[mythtv-users] Anyone tested the DVB-T2 dual tuner TBS6280?

[Simon Hobson]

Lawrence Rust wrote:

>I haven't used the 6280 but I do have a 6981 (dual dvb-S2).
>
>The driver is shipped as a mixture of source and pre-compiled object
>files for their custom frontends.  You re-build all the v4l drivers 'out
>of tree' using a v4l-media environment (see www.linuxtv.org/repo/) by
>running 'sudo make install'.
>
>The stock v4l sources supplied are old (from around 2.6.35) and don't
>contain many current fixes.  This isn't a problem per-se unless you
>intend to use the card with another v4l card.  In this case your brand
>new, bug fixed drivers are replaced by TBS's version which may or, as in
>my case, may not work.
>
>I repeatedly mailed TBS support at support at tbsdtv.com to ask how I could
>only install the 6981 driver but never got an answer.  In desperation I
>setup a git tree of 2.6.35 and merged it with the TBS drivers in order
>to separate their changes.  Finally after many hours I have a set of
>patches that I can apply to 2.6.39 that produce a working driver.
>
>Be warned that if you run a 2.6.38 or later kernel then the IR RC won't
>work because of significant changes to the RC architecture that TBS
>don't like (see http://www.tbsdtv.com/forum/viewtopic.php?f=22&t=929 and
>http://www.tbsdtv.com/forum/viewtopic.php?f=22&t=110&start=90#p2693 )
>
>Not for the faint hearted.

Since I'm also looking for a DVB-T2 tuner, and this aspect concerned 
me, I email TBS. I got a reply (on Saturday evening !) which I've had 
permission to forward :


My email read :

>As a long time user o Myth TV, I've been waiting for a usable DVB-T2 
>tuner to add to my setup as our Freeview HD uses the T2 standard. I 
>already have DVB-T.
>When someone posted about your TBS6280 I was a little excited, until 
>the thread progressed and it seems your Linux drivers & support just 
>aren't up to scratch.
>
>Personally, if there were "in tree" support of this card then I'd 
>already have bought one. Looking at the "me too"s in the thread, it 
>seems there are several others in the same position.
>
>The technical situation seems to be summed up by this posting :
>http://www.gossamer-threads.com/lists/mythtv/users/489003?do=post_view_threaded#489003
>
>Since I (along with others) don't have the skills to embark on such 
>efforts to get a working setup, it seems I'll have to wait. If 
>someone else comes up with a working card in the meantime then 
>you'll have lost out.
>
>Are you working with anyone to get support for your cards included 
>in the standard kernel tree ? If not, are you planning on doing so ?
>Having been down the road of "compiled drivers that break on every 
>kernel update" before, it's not somewhere I intend to go again.
>
>I think you only have a fairly narrow window of opportunity here. 
>Watching the threads in the mailing list there is pent up demand for 
>good DVB-T2 tuners with "support in the kernel" - whoever is second 
>with such a device will miss out.



And the reply :

>Date: Sat, 13 Aug 2011 22:08:30 +0300
>From: Turbosight Europe <euro_support at tbsdtv.com>
>CC: support <support at tbsdtv.com>
>Subject: Re: 6280 Dual DVB-T2 PCIe card & Linux
>
>
>hi Simon,
>
>i'm going to try to give you more full picture of the situation, 
>because the forum post you referred to is quite subjective and it's 
>not even giving small part of the whole picture. first of all, we've 
>never heard from the person made that post: Lawrence Rust 
><lvr at softsystem.co.uk>. so, he didn't get an answer from us simply 
>because his e-mails have never reached our support e-mail inbox - 
>most likely his e-mails were filtered by the spam filters along the 
>way, but that's not our fault - actually it's his responsibility to 
>make sure his e-mail address and/or e-mail server is not 
>black-listed by spam lists and filters. anyway, as you can see even 
>without any help from our side Lawrence Rust still managed to merge 
>the driver to his own V4L tree for 2.6.39:
>
>"Finally after many hours I have a set of patches that I can apply 
>to 2.6.39 that produce a working driver."
>
>which in fact should clearly show you it's not quite hard to figure 
>that out even by yourself (depending on your Linux experience) and 
>most importantly that our drivers are completely Linux kernel and 
>distribution independent - after all Lawrence Rust took the 
>necessary code to support his TBS board from Linux driver package 
>for 2.6.35 and merged it in 2.6.39 V4L tree successfully. so, even 
>his post may seems negative at first glance it's the contrary and it 
>actually confirms what in my opinion is the biggest advantage of TBS 
>Linux drivers - even they are not entirely open-source they are 
>still Linux kernel and distribution independent and if you wish you 
>can use your own up-to-date V4L tree to patch it with support for 
>TBS hardware.
>
>so, to make it more clear let me leave now Lawrence Rust post aside 
>and start from the beginning - we always use latest 
>chips/technologies on all our cards and thus we need to sign NDAs 
>with the companies making those chips - that's why we have some 
>restrictions and can't fully open source the Linux drivers. however, 
>we don't fully closed source them either (like for example nVidia 
>and AMD/ATi did for their video hardware) - we decided to choose 
>something in the middle - keep as much as possible open-source and 
>keep close source as little as possible, but still sufficient to not 
>result in breaching the NDAs we signed. we did that exactly to give 
>maximum freedom to our Linux customers and don't cause them the 
>inconvenience of proprietary fully closed source drivers (aka 
>'blobs') and at the same time deal with the restrictions enforced by 
>the NDAs. so, as i mentioned the small part of the driver, which is 
>closed source doesn't give any restrictions in practice, because it 
>still keeps the driver Linux kernel and distribution independent. 
>basically, that means you can patch any V4L tree and add support for 
>TBS hardware to it. unfortunately, it turned out that distributing 
>the driver as a patch to V4L tree that everyone can apply to his/her 
>own and up-to-date V4L tree is hard for more inexperienced Linux 
>users (i.e. average users) - that's why we provide Linux driver 
>package that contains V4L tree (limited to support kernels up to 
>some version) patched with the necessary changes to add support for 
>TBS cards. still, all our customers that are building embedded Linux 
>systems with TBS hardware use their own V4L tree, but usually they 
>are companies, which have experienced Linux engineers that are 
>skilled enough to get the necessary parts from our Linux driver 
>package and patch their own V4L tree with support for TBS cards. so, 
>the V4L tree in our Linux driver package is not always up-to-date, 
>but sometimes we even keep it older on purpose - for example at the 
>moment we don't update the V4L tree in our Linux driver package, 
>because there are some issues with the new 'rc-core' remote control 
>infrastructure in current V4L trees (i won't elaborate more about it 
>here, but if you're interested you can search our forum for full 
>details). so, by the time of upcoming Linux kernel 3.0 we're 
>planning to migrate the package to the latest V4L tree.
>
>so, the facts are the facts - due to NDA restrictions it's always 
>the same story and latest and high performance hardware don't have 
>fully open-source Linux drivers, but we did our best to make the 
>small closed source part of the drivers for our hardware in a way 
>that it doesn't give any restrictions in practice and behaves as 
>open-source, i.e. no Linux kernel version or distribution 
>restrictions (our Linux driver package is restricted up to 
>particular Linux kernel though, but that's because it contains V4L 
>tree which adds the limitation).
>
>as you can see there are many things to be considered and many 
>points of view and of course many opinions about it too. however, 
>our experience in the last several years shows that we had more 
>satisfied Linux customers than the opposite, for example last week i 
>helped Linux user from UK to get his TBS 6280 running and he's very 
>happy being able to enjoy DVB-T2 in his Linux setup especially 
>considering the fact that no other DVB-T2 card on the market has 
>Linux support and at the same time it's latest generation hardware 
>like TBS 6280 with high-performance Samsung DVB-T2 Network Interface 
>Module (NIM) based on the latest technologies - for example NXP 
>tuner inside the Samsung module is the only one certified for DVB-T2 
>(other companies use tuner that is for DVB-T). anyway, i know that 
>no matter how much effort we put and no matter how good support we 
>provide there always will be people who are complaining - i don't 
>mean you here, because your haven't used any TBS product in Linux in 
>practice to form your own opinion, but usually the complains i'm 
>talking about are more subjective than something objective and real 
>and often even due to limited knowledge of the customer - after all 
>we can't teach our customer how to use Linux.
>
>also, in case of TBS 6280 there is something else you should know - 
>currently we keep two separate Linux drivers: one that support DVB-T 
>only and another that supports DVB-T2 only. that's because we want 
>more people to test those two drivers before we merge them and it's 
>more efficient to locate the problem in case a customer report such. 
>so, the driver for TBS 6280 you can find on our website supports 
>DVB-T only, but it can be switched to support DVB-T2 very easy. so, 
>single driver that supports DVB-T and DVB-T2 will be released after 
>we're sure both codes are bug-free. so, far feedback we received 
>doesn't reveal any serious bugs and reported problems were fixed - 
>as i mentioned we already have customers in UK that are Linux users 
>and use the card to watch DVB-T2 signal in UK.
>
>anyway, i spent time to write you this long e-mail on Saturday 
>evening, because i'm Linux user myself and wanted to try to give you 
>the whole picture. in any case it's up to you and your choice which 
>hardware you believe fits your requirements best and purchase it.
>
>best regards,
>constantine


It follow up emails, it does seem like Constantine is still aiming a 
bit high in terms of user experience - he still believes these 
drivers are easier to install than a kernel upgrade. This suggests 
he's not really used to dealing with people for whom "apt-get update 
&& apt-get upgrade" on the command line is "advanced".

I've said I'll wait and see how things pan out. I realise there is 
stuff under NDA, and not being a dev I've no idea how hard it is to 
get part of the driver into the kernel while leaving the stuff that's 
under NDA as a binary blob that's independent of kernel/V4L.
-- 
Simon Hobson

Visit http://www.magpiesnestpublishing.co.uk/ for books by acclaimed
author Gladys Hobson. Novels - poetry - short stories - ideal as
Christmas stocking fillers. Some available as e-books.