Hello Vitaliy, On Sat, May 23, 2020 at 3:07 PM Vitaliy Makkoveev <henscheltig...@yahoo.com> wrote: > > On 23 May 2020, at 13:11, Sergey Ryazanov <ryazanov....@gmail.com> wrote: > > On Wed, May 20, 2020 at 10:13 PM Vitaliy Makkoveev > > <henscheltig...@yahoo.com> wrote: > >> On Wed, May 20, 2020 at 04:08:01AM +0300, Sergey Ryazanov wrote: > >>> On Tue, May 19, 2020 at 12:11 PM Vitaliy Makkoveev > >>> <henscheltig...@yahoo.com> wrote: > >>>> On Mon, May 04, 2020 at 10:03:40PM +0300, Sergey Ryazanov wrote: > >>>>> Split checks from frame accepting with header removing in the common > >>>>> PPP input function. This should fix packet capture on a PPP interfaces. > >>>> > >>>> Can you describe the problem you fix? As mpi@ pointed to me, reviewers > >>>> are stupid and have no telepathy skills :) > >>> > >>> When I tried to capture packets on a ppp (4) interface (with pipex > >>> activated), I noticed that all the PPP CCP frames were ok, but all the > >>> ingress PPP IP frames were mangled, and they did not contain the PPP > >>> header at all. > >> > >> This time only pppx(4) and pppac(4) have pipex(4) support. > > > > Yes, and as I wrote in the first mail, now I am working on ppp(4) & > > pipex(4) integration to speed up client side of L2TP. > > May be you can share you work? Not for commit, but for feedback. >
I send a couple of diffs in separate mails. First change is for ppp(4) to support pipex(4) acceleration in DL path. Second diff adds a new option for pppd to control pipex activation. We will need also to update xl2tpd package to teach it how to configure pppd for pipex usage. This is quite simple change, but it still require some cleanup so I do not publish it yet. > For example, each pipex session should have unique pair of `protocol’ and > `session_id’. These values are passed from userland. While the only > instance of npppd(8) uses pipex(4) this is not the problem. But you > introduce the case while pipex(4) will be used by multiple independent > userland programs. At least, I have interest how you handle this. This should not be a problem here. npppd(8) support server mode only. While my work is to implement acceleration for client side of L2TP connection. Anyway if a session collision become a case, we could additionally demux ingress traffic by the destination UDP port. -- Sergey
