Hi,
On 03/08/2018 08:44 AM, Richard Cochran wrote: > On Wed, Mar 07, 2018 at 09:47:40AM -0800, Eric Dumazet wrote: >> I would love if skb->tstamp could be either 0 or expressed in >> ktime_get() base all the time. >> >> ( Even if we would have to convert this to other bases when/if needed) > > We really do need variable clock IDs. Otherwise the HW offloading > case won't work. The desired transmit time must be expressed in terms > of the clock inside the MAC. This clock is not necessarily related to > the system time at all. > > But in addition to the performance concerns, I think putting this into > a socket option is the more natural solution. Ok, so we have it settled for clockid now. Providing it per-socket was what we'd proposed previously, so this was just an attempt to accommodate all the feedback we got on the v2 RFC. What about the tc_drop_if_late bit, though? Would it be acceptable to keep it per-packet, thus eating the 1-bit hole from skbuff if we would #if guard it (e.g. with CONFIG_NET_SCH_TBS)? Thanks, Jesus
