Hi Ido, Andrew!

I'm happy to receive these emails and to see that my old patch had
some interest. Yes, at that time there was no official support for
QSFP-DD in the upstream kernel. I was able to test my code using a
custom driver we developed in my company. If everything works out, I'd
be happy to re-submit it.

> It is a while ago, but i thought there was something odd about the
> order of the pages, or the number of the pages? And there was no clear
> indication from the kernel about QSPF page format vs QSPF-DD page
> format?
>
> So Adrian's patch is probably a good starting point, but i think it
> needs further work.

It is true, my patch was basically defining a KAPI. The decision to
have the pages 0x10 and 0x11 right after 0x00, 0x01 and 0x02 was made
based on the need to provide the same stats for QSFP-DD as for QSFP.
At that point, page 0x03 was not needed. Anyway, due to the way I
defined the offset values in qsfp-dd.h, if we add page 0x03 it would
be extremely easy to just change

#define PAG11H_OFFSET (0x04 * 0x80)
to
#define PAG11H_OFFSET (0x05 * 0x80)

and probably everything else would stay the same.
We can continue our discussion about the needed pages in the "[PATCH
net-next 1/2]" e-mails chain.


On Fri, 26 Jun 2020 at 16:06, Andrew Lunn <[email protected]> wrote:
>
> On Fri, Jun 26, 2020 at 05:53:42PM +0300, Ido Schimmel wrote:
> > On Fri, Jun 26, 2020 at 05:47:22PM +0300, Ido Schimmel wrote:
> > > From: Ido Schimmel <[email protected]>
> > >
> > > This patch set from Vadim adds support for Quad Small Form Factor
> > > Pluggable Double Density (QSFP-DD) modules in mlxsw.
> >
> > Adrian,
> >
> > In November you sent a patch that adds QSFP-DD support in ethtool user
> > space utility:
> > https://patchwork.ozlabs.org/project/netdev/patch/[email protected]/
> >
> > Back then Andrew rightfully noted that no driver in the upstream kernel
> > supports QSFP-DD and the patch was deferred.
>
> Hi Ido
>
> It is a while ago, but i thought there was something odd about the
> order of the pages, or the number of the pages? And there was no clear
> indication from the kernel about QSPF page format vs QSPF-DD page
> format?
>
> So Adrian's patch is probably a good starting point, but i think it
> needs further work.
>
>       Andrew

Reply via email to