[ofa-general] [PATCH 1/3] mcast loopback block

[Olga Shern (Voltaire)]

On 6/2/08, Ralph Campbell <ralph.campbell at qlogic.com> wrote:
>
> On Mon, 2008-06-02 at 21:06 +0300, Olga Shern (Voltaire) wrote:
> >
> >
> > On 6/2/08, Ralph Campbell <ralph.campbell at qlogic.com> wrote:
> >         On Mon, 2008-06-02 at 20:52 +0300, Olga Shern (Voltaire)
> >         wrote:
> >         >
> >         >
> >         > On 6/2/08, Ralph Campbell <ralph.campbell at qlogic.com> wrote:
> >         >         What do you mean by "block loopback"?
> >         >         Does this mean a posted send WR for a multicast
> >         packet
> >         >         that matches the local LID is not sent at all or
> >         >         the packet is sent on the wire but not sent to
> >         >         the local QP? Or is it something else?
> >         >
> >         >
> >         > It controls the ConnectX HCA multicast packets block
> >         loopback
> >         > (blck_lb) for self QP. The patch is designed to enable or
> >         >
> >         > disable blocking of all multicast packets on self QP created
> >         on the
> >         > ConnectX HCA
> >
> >         I understand that the patch enables/disables the bit.
> >         You didn't really answer my question which is what does
> >         the bit really do. I'm pretty sure it means that multicast
> >         packets aren't forwarded to any locally attached QPs
> >         but I wanted to be sure that is what it does.
> >
> >
> > No,
> > It isn't  disabled  for every QP (by default loopback is enabled),
> > when you attach QP to multicast group you can decide to disable  it
> > (the decision is done per QP)
>
> So, in other words, if you create a QP with this option and then
> attach it to a multicast group and a multicast packet is posted
> on the send work request queue of the "disabled" QP, it won't
> receive a copy of the packet it sent. If some other QP on the
> same HCA is also attached to the same multicast group, it will
> receive a copy of the multicast packet.
>
> Yes, you are rigth
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/general/attachments/20080603/ee7fe96b/attachment.html>