em(4) + ALTQ broken

[Jack Vogel]

It should never get to the drbr code, look at net/if_var.h, in the inline
definition
of drbr_dequeue() there is an #ifdef ALTQ that will effectively vector if to
use
the old IFQ_DRV_DEQUEUE() method.

I guess I can test the build on a system here, stick some syntax error in
and see if it hits.

Right now the inserted code looks solid enough to me, so somehow I think
its not being defined.

Jack


On Tue, Feb 2, 2010 at 2:38 PM, Pyun YongHyeon <pyunyh at gmail.com> wrote:

> On Tue, Feb 02, 2010 at 01:41:01PM -0800, Jack Vogel wrote:
> > LOL, and I can answer my own question, I just looked and the ONLY
> > 1Gig drivers using multiqueue are mine, so I guess not eh? :)
> >
>
> I was wrong. ALTQ is defined in opt_global.h so drbr_ interface
> should already see ALTQ. I have to look into drbr_ code.
>
> > J.
> >
> > On Tue, Feb 2, 2010 at 1:39 PM, Jack Vogel <jfvogel at gmail.com> wrote:
> >
> > > Thanks Max, yes, i've done some digging myself and now see how things
> > > work, the rubber meets the road in the defines in if_var.h.
> > >
> > > And what it does is effectively short circuit Kip Macy's multiqueue
> code
> > > in favor of the old method.
> > >
> > > Right now I can see two possibilities, either the defines are not set
> in
> > > the build, OR there is something wrong in the logic of the short
> circuit
> > > approach in Kip's code.
> > >
> > > A question might be if ANY driver that is usinig TX Multiqueue has been
> > > successfully used with ALTQ?
> > >
> > > Jack
> > >
> > >
> > >
> > > On Tue, Feb 2, 2010 at 12:37 PM, Max Laier <max at love2party.net> wrote:
> > >
> > >> On Tuesday 02 February 2010 18:48:02 Jack Vogel wrote:
> > >> > So apparently this thing needs no special knowledge in the driver,
> yet
> > >> > something in
> > >> > the new code breaks it, can someone explain tersely how the altq app
> > >> > actually
> > >> > "pokes" or "hooks up" to the driver? I am not clear about that and I
> > >> >  suspect if I was
> > >> > this would all be clearer.
> > >>
> > >> The whole story is in
> > >>
> > >>  man 9 altq
> > >>
> > >> long story short, as long as you consistently use the IFQ_* macros to
> > >> manage
> > >> the interface queue, things should just work.  if_var.h used to
> > >> conditionally
> > >> define these macros to avoid ALTQ overhead when the kernel is built
> > >> without
> > >> ALTQ.  This has changed a long time ago and should not make any
> difference
> > >> anymore.
> > >>
> > >> I can't figure out who the OP is, but could you make sure that the
> > >> includes
> > >> that are used to built the kernel are up to date?  You are building
> with
> > >> the
> > >> buildkernel target and not "the old way", right?  Also, if you build
> just
> > >> the
> > >> module, the build might pick up the includes from /usr/include instead
> of
> > >> src/sys ...
> > >>
> > >> > Jack
> > >> >
> > >> > On Tue, Feb 2, 2010 at 9:37 AM, Pyun YongHyeon <pyunyh at gmail.com>
> > >> wrote:
> > >> > > On Tue, Feb 02, 2010 at 09:30:52AM -0800, Nick Rogers wrote:
> > >> > > > > I guess the problem comes from multi-queue support. The drbr
> > >> > > > > interface is implemented with inline function so em(4)/igb(4)
> may
> > >> > > > > have to define ALTQ to the header. I have not tested the
> patch(no
> > >> > > > > time at this moment) but would you give it try?
> > >> > > > >
> > >> > > > > I tried the patch and it did not work.
> > >> > >
> > >> > > You rebuilt kernel, right? Rebuilding kernel module has no effect.
> > >> >
> > >> > _______________________________________________
> > >> > freebsd-stable at freebsd.org mailing list
> > >> > http://lists.freebsd.org/mailman/listinfo/freebsd-stable
> > >> > To unsubscribe, send any mail to "
> > >> freebsd-stable-unsubscribe at freebsd.org"
> > >> >
> > >> >
> > >> > !DSPAM:4b686584144321871135632!
> > >> >
> > >>
> > >
> > >
>