On Fri, Feb 10, 2017 at 7:42 PM, Jakub Kicinski <kubak...@wp.pl> wrote: > On Fri, 10 Feb 2017 18:33:13 +0200, Or Gerlitz wrote: >> On Fri, Feb 10, 2017 at 3:22 AM, Jakub Kicinski <kubak...@wp.pl> wrote:
>> > In cls_bpf we do store information about whether program is offloaded or >> > not already (see the @offloaded member). Could we simplify the code >> > thanks to this? >> yeah, I felt like I don't fully understand the role of the offloaded >> member. As I wrote, this patch is compile tested only, I will be happy >> if you can test it post here a better version, I don't think we need >> to add/change the flags semantics, see next > The @offloaded member just tells us whether the program is offloaded. > We need it because unlike u32 and flower (I think) we have explicit > ADD and REPLACE. Other filters just always do REPLACE. I assume the > driver keeps track if there is already an associated rule in that case? In flower you always add new filter through the replace (TC_CLSFLOWER_REPLACE) call
