auto detection of Flytv duo/hybrid and pci/cardbus confusion

Hi,

I've finally gotten around to reading the code and trying to get my PCI MSI TV@nywhere A/D card auto detected.

First clarification, duo versus hybrid.
Are "duo" cards equipped with two independent tuners that can both be used at the same time?
Are "hybrid" cards necessarily equipped with digital and analogue tuners? Can a two tuner card be both a duo and a hybrid, if one tuner is digital the other is analogue and they can both be used at the same time?

Second clarification, PCI versus cardbus.
They don't look anything like each other, but can they be logically interchangeable? If the code for a cardbus tuner happens to work for a PCI tuner is there anything wrong with referring to the PCI tuner as a cardbus device?

Looking at <http://www.linuxtv.org/wiki/index.php/DVB-T_PCMCIA_Cards>
there does not appear to be any such thing as a SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS, despite the entry (number 94) in saa7134.h. Looking at
<http://www.linuxtv.org/wiki/index.php/DVB-T_PCI_Cards#LifeView>
there is a PCI version - but there is no PCI version in saa7134.h.

Should "SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS" be changed to "SAA7134_BOARD_FLYDVBT_HYBRID"?

It appears that both PCI and cardbus versions of the Flytv duo exist and are listed in saa7134.h - despite slightly inconsistent punctuation; SAA7134_BOARD_FLYDVBTDUO versus SAA7134_BOARD_FLYDVBT_DUO_CARDBUS.

Should
"SAA7134_BOARD_FLYDVBTDUO" be changed to
"SAA7134_BOARD_FLYDVBT_DUO"?

I have an MSI TV@nywhere A/D PCI card that works with the option card=94

There appears to not be an entry in struct pci_device_id saa7134_pci_tbl[] in saa7134-cards.c for my card. There is a reference to a "TV@nywhere DUO" which I guess is a valid entry for a different card.

Is the entry;

          {
                .vendor       = PCI_VENDOR_ID_PHILIPS,
                .device       = PCI_DEVICE_ID_PHILIPS_SAA7133,
                .subvendor    = 0x4e42,
                .subdevice    = 0x3502,
                .driver_data  = SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS
        },

supposed to be;

           {
                .vendor       = PCI_VENDOR_ID_PHILIPS,
                .device       = PCI_DEVICE_ID_PHILIPS_SAA7133,
                .subvendor    = 0x4E42,         /* MSI */
                .subdevice    = 0x3306,         /* TV@nywhere Hybrid A/D */
                driver_data  = SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS,
        },

with the subdevice changed, or possibly;

           {
                .vendor       = PCI_VENDOR_ID_PHILIPS,
                .device       = PCI_DEVICE_ID_PHILIPS_SAA7133,
                .subvendor    = 0x4E42,         /* MSI */
                .subdevice    = 0x3306,         /* TV@nywhere Hybrid A/D */
                driver_data  = SAA7134_BOARD_FLYDVBT_HYBRID,
        },

with the subdevice and driver_data changed, or should there be an extra entry in the list?

Quoted 6 lines

First clarification, duo versus hybrid.
Are "duo" cards equipped with two independent tuners that can both be used at the same time?
Are "hybrid" cards necessarily equipped with digital and analogue tuners? Can a two tuner card be both a duo and a hybrid, if one tuner is digital the other is analogue and they can both be used at the same time?

Greetings all!

Let me clear things up a bit.

Quoted 7 lines

Quoted 6 lines

First clarification, duo versus hybrid.
Are "duo" cards equipped with two independent tuners that can both be used at the same time?
Are "hybrid" cards necessarily equipped with digital and analogue tuners? Can a two tuner card be both a duo and a hybrid, if one tuner is digital the other is analogue and they can both be used at the same time?

LifeView are using two vendor IDs - 4E42h for all (!) their OEMs, and their own one for LifeView branded cards. Hence we need two PCI ID entries for everything, each pair pointing back to the same card data.

Then, card types.

The analog-only and "hybrid" have one single tuner, for DVB-T or analog. The "Duo" cards have two tuner frontends, one for DVB-T and the other for analog.
"Trio" cards add a DVB-S frontend, which cannot be used at the same time as the DVB-T frontend. Like the Duo, these can run one digital and one analog stream in parallel.

Finally, card shapes.

Each card type comes in CardBus, PCI, and MiniPCI shape. The flavors are compatible, so that again, the PCI ID data point back to the same card entry for e.g. the PCI and CardBus Duo.

The card type/shape combinations are distinctly identified by their subsystem ID. No need to guesstimate anything.

That's the plan at least.

regards,
Peter

Am Sonntag, den 17.02.2008, 21:53 +0100 schrieb Peter Missel:

Quoted 31 lines

Greetings all!

Let me clear things up a bit.

Quoted 7 lines

Quoted 7 lines

First clarification, duo versus hybrid.
Are "duo" cards equipped with two independent tuners that can both be used at the same time?
Are "hybrid" cards necessarily equipped with digital and analogue tuners? Can a two tuner card be both a duo and a hybrid, if one tuner is digital the other is analogue and they can both be used at the same time?

LifeView are using two vendor IDs - 4E42h for all (!) their OEMs, and their own one for LifeView branded cards. Hence we need two PCI ID entries for everything, each pair pointing back to the same card data.

Then, card types.

The analog-only and "hybrid" have one single tuner, for DVB-T or analog. The "Duo" cards have two tuner frontends, one for DVB-T and the other for analog.
"Trio" cards add a DVB-S frontend, which cannot be used at the same time as the DVB-T frontend. Like the Duo, these can run one digital and one analog stream in parallel.

Finally, card shapes.

Each card type comes in CardBus, PCI, and MiniPCI shape. The flavors are compatible, so that again, the PCI ID data point back to the same card entry for e.g. the PCI and CardBus Duo.

Quoted 7 lines

The card type/shape combinations are distinctly identified by their subsystem ID. No need to guesstimate anything.

That's the plan at least.

regards,
Peter

Hi Peter!

Your plan is fine so far.

We might add some more comments to group devices obviously together, since those looking first time at it are a bit lost.

For such i2c IR limits, we have your and Eddi's comments.

Since we can't help it easily, Peter D. should suggest the older version of the MSI A/D for auto detection. It won't make anything more worse on that not fully clear Vivanco stuff, except Hartmut might have ideas.

Cheers,
Hermann

Quoted 53 lines

Am Sonntag, den 17.02.2008, 21:53 +0100 schrieb Peter Missel:

Quoted 31 lines

Greetings all!

Let me clear things up a bit.

Quoted 7 lines

Quoted 7 lines

First clarification, duo versus hybrid.
Are "duo" cards equipped with two independent tuners that can both be used at the same time?
Are "hybrid" cards necessarily equipped with digital and analogue tuners? Can a two tuner card be both a duo and a hybrid, if one tuner is digital the other is analogue and they can both be used at the same time?

LifeView are using two vendor IDs - 4E42h for all (!) their OEMs, and their own one for LifeView branded cards. Hence we need two PCI ID entries for everything, each pair pointing back to the same card data.

Then, card types.

The analog-only and "hybrid" have one single tuner, for DVB-T or analog. The "Duo" cards have two tuner frontends, one for DVB-T and the other for analog.
"Trio" cards add a DVB-S frontend, which cannot be used at the same time as the DVB-T frontend. Like the Duo, these can run one digital and one analog stream in parallel.

Finally, card shapes.

Each card type comes in CardBus, PCI, and MiniPCI shape. The flavors are compatible, so that again, the PCI ID data point back to the same card entry for e.g. the PCI and CardBus Duo.

Quoted 7 lines

The card type/shape combinations are distinctly identified by their subsystem ID. No need to guesstimate anything.

That's the plan at least.

regards,
Peter

Hi Peter!

Your plan is fine so far.

We might add some more comments to group devices obviously together, since those looking first time at it are a bit lost.

For such i2c IR limits, we have your and Eddi's comments.

Since we can't help it easily, Peter D. should suggest the older version of the MSI A/D for auto detection. It won't make anything more worse on that not fully clear Vivanco stuff, except Hartmut might have ideas.

Quoted 2 lines

Cheers,
Hermann

I think that you just suggested something. I'm going to stand at the side and nod my head. ;-)

What should I do to make it an official suggestion?

Hi Peter,

Am Sonntag, den 17.02.2008, 14:28 +1100 schrieb Peter D.:

Quoted 5 lines

Hi,

I've finally gotten around to reading the code and trying to get my PCI MSI TV@nywhere A/D card auto detected.

Quoted 7 lines

First clarification, duo versus hybrid.
Are "duo" cards equipped with two independent tuners that can both be used at the same time?
Are "hybrid" cards necessarily equipped with digital and analogue tuners? Can a two tuner card be both a duo and a hybrid, if one tuner is digital the other is analogue and they can both be used at the same time?

for that all I give some examples further below.

Quoted 5 lines

Second clarification, PCI versus cardbus.
They don't look anything like each other, but can they be logically interchangeable? If the code for a cardbus tuner happens to work for a PCI tuner is there anything wrong with referring to the PCI tuner as a cardbus device?

No, we do such and vice versa, but then we add a comment PCI or cardbus version usually and another important compatible category is Mini PCI. You should also add LR306 as the board type.

Quoted 6 lines

Looking at <http://www.linuxtv.org/wiki/index.php/DVB-T_PCMCIA_Cards>
there does not appear to be any such thing as a SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS, despite the entry (number 94) in saa7134.h. Looking at
<http://www.linuxtv.org/wiki/index.php/DVB-T_PCI_Cards#LifeView>
there is a PCI version - but there is no PCI version in saa7134.h.

They do exist, see bttv-gallery.de,

also Mauro has already a MSI OEM
version. They are special for DVB-T, since they switch the mode by a gpio pin and have the silent i2c connection to the tuner open and don't use the i2c gate of the tda8290 analog demod inside the saa7131e for DVB-T tuning.

Quoted 3 lines

Should "SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS" be changed to "SAA7134_BOARD_FLYDVBT_HYBRID"?

No. Just also add "PCI" and "LR306" to your comments.
The gpio 21 is used for the TV/radio switch. The gpio 22 = 0 in the mask is used to switch the tuner AGC to the analog demod.
On some cardbus version, which have the gpio 27 high, switching it to 0 turns the fan on, on yours it does nothing, since it is 0 anyway.

Quoted 8 lines

It appears that both PCI and cardbus versions of the Flytv duo exist and are listed in saa7134.h - despite slightly inconsistent punctuation; SAA7134_BOARD_FLYDVBTDUO versus SAA7134_BOARD_FLYDVBT_DUO_CARDBUS.

Should
"SAA7134_BOARD_FLYDVBTDUO" be changed to
"SAA7134_BOARD_FLYDVBT_DUO"?

Not related to your device. Subdevice is 0x0306 and you have 0x3306, but yes, these are in a different family of devices too.

Quoted 7 lines

I have an MSI TV@nywhere A/D PCI card that works with the option card=94

There appears to not be an entry in struct pci_device_id saa7134_pci_tbl[] in saa7134-cards.c for my card. There is a reference to a
"TV@nywhere DUO" which I guess is a valid entry for a different card.

Yes it is. Peter Missel bought some and added the entry.

Duo in LifeView terminology means two separate tuners. One for analog TV/FM and one for DVB-T.

On the older Duo versions those two tuners were _not hybrid_ and different for analog TV and DVB-T.
Only the newer Duo variants with tda8275A use two identical such tuners, but not in hybrid mode. One is always in analog and the other always in digital mode. The older types are out of production and using two still allows to have DVB-T and analog TV from the tuners at once.

This is not possible with one single hybrid tuner. DVB-T and analog video from an external input, VCR tuner or something, is possible at once, but needs also to use packed video formats for analog during that due to limitations of the dma engines with planar formats.

Then the Trio has even one more tuner for DVB-S, but there is only _one_ saa713x PCI bridge on all such and only one channel decoder/demodulator for each mode. Since the single saa713x chip has only one TS/mpeg/host interface, either usable in parallel or serial mode, DVB-S and DVB-T can not be used at the same time. (also not an mpeg encoder)

This is only possible with two saa713x bridges, like the md7134 in the CTX925 version has them and when we enable DVB-S support on the second bridge hopefully soon. Else functionality is like on the Trio. (BTW, the md7134 represents lots of different cards through eeprom detection)

Next category is the Medion md8800 Quad(ro) with two saa7131e, two fully usable tda8275ac1 hybrid (2 analog + 2 DVB-T demods, no FM) and two tda8263 DVB-S tuners plus two tda10086 channel decoders plus an isl6405 dual LNB controller, where we just have support of the first SAT connector now, diseqc for multiswitches is not fully tested yet. There is RF loopthrough, in and out, respectively to the other.

So it should be already possible to view and record from both DVB-S devices at once, within that tunable spectrum the active one limits for the passive by controlling tone and LNB voltage.

That loopthrough can be switched off, for example when we have active control for the second SAT connector too.

On the two hybrid tuners the RF feed can also be switched, either to loopthrough mode from TV1 antenna connector to both tuners or have input of choice from both analog/DVB-T antenna connectors separately.

Quoted 29 lines

Is the entry;

          {
                .vendor       = PCI_VENDOR_ID_PHILIPS,
                .device       = PCI_DEVICE_ID_PHILIPS_SAA7133,
                .subvendor    = 0x4e42,
                .subdevice    = 0x3502,
                .driver_data  = SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS
        },

supposed to be;

           {
                .vendor       = PCI_VENDOR_ID_PHILIPS,
                .device       = PCI_DEVICE_ID_PHILIPS_SAA7133,
                .subvendor    = 0x4E42,         /* MSI */
                .subdevice    = 0x3306,         /* TV@nywhere Hybrid

A/D */ driver_data = SAA7134_BOARD_FLYDVBT_HYBRID_CARDBUS, },

with the subdevice changed, or possibly;

           {
                .vendor       = PCI_VENDOR_ID_PHILIPS,
                .device       = PCI_DEVICE_ID_PHILIPS_SAA7133,
                .subvendor    = 0x4E42,         /* MSI */
                .subdevice    = 0x3306,         /* TV@nywhere Hybrid
A/D */ driver_data  = SAA7134_BOARD_FLYDVBT_HYBRID,
        },

Quoted 2 lines

with the subdevice and driver_data changed, or should there be an extra entry in the list?

We have no SAA7134_BOARD_FLYDVBT_HYBRID, this would require already the new entry.

The 4e42:3502 is Mauro's MSI TV @nywhere A/D NB cardbus and likely other OEMs too. On the subdevice 0x3306 are the related PCI cards and 0x3307 are Mini PCI.

Thought you did not pick up my patch suggestion that time,
http://www.linuxtv.org/pipermail/linux-dvb/2007-June/018678.html
because we had no means to make a difference to the Vivanco 21056 and we did not get the analog TV mode right on it.

For the i2c remote with mdt10p55 (pic 16c505) Eddi had a patch. We talked about it and remaining implementation questions, but should be usable for you in private if you replace the appearance of SAA7134_BOARD_FLYDVB_TRIO with your device or add it accordingly.

Since the cardbus and mini pci devices seem not to have a remote so far, this could eventually rise the question of a separate entry later on IMHO.

Cheers,
Hermann