[mythtv-users] Ceton drivers for kernel?

[Gary Buhrmaster]

On Thu, Jul 22, 2021 at 10:42 PM Steve Greene <sgreene820 at gmail.com> wrote:

> I miss the 4 cablecard tuners available from my Ceton Infinitv 4 PCI-E card. The manufacturer insisted over the years they were still in business, but it's pretty clear they have crashed and burned.

Ceton made a pivot that bankrupted themselves.  I believe
the company still exists in legal limbo for certain parties to
continue to try to pick at whatever assets remain on the
carcass.

> I've tried the github drivers, but they haven't kept pace
> with the latest 5.x kernels. What is stopping "the kernel"
> from supporting the tuners.   I realized I'm probably
> drastically oversimplifying both technical and legal
> challenges. Can someone enlighten me?

Some Ceton staff themselves were Linux users (they
even initially thought the product might have a useful
niche in LInux), but the company mgmt realized that
that a protected content path requirement for the core
target audience basically made it a Windows (mostly)
product, and as such, never really formally supported
Linux, and never productized the driver.

To get something into the kernel, the developer(s) need
to write to the GPL license, agree to certain T&Cs,
and also to agree to long term support (since, as you
experience, the kernel itself changes, and the
maintainer is responsible for at least some of the
changes some of the time).

And, last I knew, ceton had not formally placed the
drivers under the GPL license (and the one line that
some companies use to bypass kernel checks for out of
tree drivers is not a full source license).  As the company
is no more, trying to get signoff on any license change
might be complex to achieve (although a few of the
principals have moved on to better pastures, the
code itself is presumably still owned by the corporate
zombie).

Since the license status is unclear, if you are a
capable kernel developer, you probably would need
to write a completely new driver and get it upstreamed.

Note since the existing driver(s) license is unclear,
you may need to build a chinese wall between those
in your team performing the reverse engineering of
the hardware and your writing of a new driver.  Those
that have looked at the existing driver may be
considered tainted depending on your jurisdiction's
legal frameworks.

Oh, and last I knew, the latest beta firmware that
dropped just before the implosion (the beta firmware
was required for certain MSOs due to an "interesting"
interpretation of a CableLabs reg) has a memory
leak under some conditions that may require a
tuner reboot from time to time otherwise the tuner
starts to randomly fail to operate.  It would have
presumably been resolved before the final release
of the firmware, but it never got that far.