[linux-dvb] Multiproto API/Driver Update

Steven Toth stoth at linuxtv.org
Fri Sep 5 15:45:51 CEST 2008

Markus Rechberger wrote:
> Hi,
> On Thu, Sep 4, 2008 at 10:47 PM, Johannes Stezenbach <js at linuxtv.org> wrote:
>> On Thu, Sep 04, 2008, Manu Abraham wrote:
>>> Does it support ISDB-T, ATSC-MH, CMMB, DBM-T/H?
>>> Intentionally, no!  Experience with the old api development has proven
>>> that making blind assumptions about delivery systems is a bad idea.
>>> It's better to add in support for these when the hardware actually arrives
>>> and can be properly tested.

We have ISDB-T running under linux in the lab, we have a pretty good 
idea about what we need to generalize for an API, but you can't make a 
standard out of one demod.

All API's developers should be cautious here.

> I have Empia ISDB-T and DMB-T/H hardware and the corresponding signal
> generator for it here,
> it's right on my roadmap and work can be started within a few days.

A big difference between can and will, the em28xx fiasco tells us this.

I hope this next question isn't seen as a flame, because I ask this as a 
genuine question. Are you committing to adding ISDB-T and DMB-T/H 
support to master at linuxtv.org? Or would this be something you'd plan 
to keep at you own site, like your other code?

>> Full ACK on this one. Once an API is merged into the mailine
>> kernel we're stuck with it, no matter how ugly and broken it might be.
>> -> NEVER merge untested APIs


> should be the rule but there's always an exception for it too . o (
> thinking about KVM )
>>> If you would like to use any of these drivers now, you may pull the
>>> tree from http://jusst.de/hg/multiproto.  Drivers may be configured
>>> with 'make menuconfig' the same as you've done with v4l.
>>> Feedback, bug reports, etc. are welcomed and encouraged!
>> I only want to add a bit of historical perspective so people
>> are aware of the reasons why Steve came up with his alternative
>> API proposal, and why a number of developers seem to support it.

/me nods

>> First let's look at the timestamps:
>> http://jusst.de/hg/multiproto/log/2a911b8f9910/linux/include/linux/dvb/frontend.h
>> http://jusst.de/hg/multiproto_api_merge/log/4c62efb08ea6/linux/include/linux/dvb/frontend.h
>> Then at some discussion from nearly one year ago:
>> http://article.gmane.org/gmane.linux.drivers.dvb/36643

This is worth reading.

> by experience I'm sure most people won't read up the history here...

Also agreed, a lot of people have lost sight of the history - which is 
the entire reason that the S2API alternative exists.


The S2API tree is available on linuxtv.org with HVR4000 support, 
spanning 3 or 4 patches.

I have a TT-3200, do you have a complete tree with all of your pull 
req'd patches available on linuxtv.org for testing, including your 
demodulation and tuning drivers? Do you have a complete solution that I 
can evaluate for pro's and cons?



More information about the linux-dvb mailing list