Let's add the LKML to this.
On 9/13/07, Markus Rechberger <[email protected]> wrote:
> On 9/12/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > Markus,
> >
> > Em Ter, 2007-08-14 ?s 16:31 +0200, Markus Rechberger escreveu:
> > > Following patch adds the possibility to implement tuner drivers in
> > > userspace.
> >
> > As you asked me about userspace driver, at Linux Conf Europe, let me
> > give you my feedback about it.
> >
> > On Linux, userspace-to-kernelspace APIs are meant to be forever. This
> > means that, once a newer API is created, this should remain supported
> > for all future versions. So, such APIs should be carefully analyzed and
> > accepted by the community, before going to mainstream.
> >
>
> The V4L and DVB API is stable at the moment because it's at a stage
> which is sufficient for older devices but not sufficient for newer
> devices anymore.
> To support newer device it needs a change.
>
> > I don't see any technical reason why tuner drivers should be moved to
> > userspace. Looking at xc3028 device, the driver is very simple and
> > doesn't require any special treatment that it isn't possible to be done
> > at kernel. There are already some implementations on kernelspace that
> > works fine.
> >
>
> As from my side to support the xceive driver properly it needs a
> rewrite and a proper API description. Since it's not possible to
> discuss any API changes I will work around at least for those devices
> which I can support for.
>
> > On the other hand, a TV driver without a tuner is a broken driver. With
> > parts of the driver being at userspace, this means to add undesired
> > complexity at the drivers architecture, while not bringing any benefit.
> >
> > If you look at V4L history, the first drivers started at userspace,
> > being migrated to kernelspace, where we have the proper scenario for
> > managing those devices.
> >
> > Another aspect that should be analyzed is what is desired by the
> > community:
>
> don't get me wrong but the existing community is rather small and
> kicking off people who are interested in changing things.
> I recently had a talk with someone and I've been told that I'm kicking
> off people.
> Guess why I kick off people? -> because they do not contribute in a
> productive way which also means submitting patches. Optical useless
> changes don't make any difference at the functionality in the end. And
> my requirements are ignored constantly here.
>
> > kernelspace tuners or userspace tuners. Keeping support for
> > both at long term doesn't seem reasonable. The Linux community should
> > decide what is the better way. Currently, only you are pushing for
> > userspace tuners, mainly due to non-technical reasons.
>
> read the project site and you will see the reasons.
> http://mcentral.de/wiki/index.php/Userspace_tuner#Advantages
> Another advantage is that I have cygwin based code here which I can
> easily reuse with all that work I'm not going to reinvent the wheel
> even for newer devices which I work on.
>
> > Almost all the
> > other developers are comfortable with kernelspace tuners. So, creating
> > an userspace interface just to make you happy is not the way we should
> > go.
> >
>
> I'm afraid of giving the people which are against what I submitted the
> responsibility over the project. Initially there was an RFC which
> didn't get commented either (well there was one useless comment, I
> tried to discuss it on IRC before with the same guy) after I
> implemented exactly what I proposed there I got the first non
> technical comments - also keep in mind that working on something costs
> alot of time and talking about something unknown is rather cheap.
>
> > A final aspect is that having an userspace driver for tuner will mean
> > that the kernel driver will depend on an userspace counterpart in order
> > to work. This will allow a vendor with bad intentions to release a
> > partially broken userspace driver, with limited capabilities, and a
> > closed source driver for full support. This model is likely to occur, if
> > you take a look at the past. For example: ATI and Nvidia closed source
> > drivers, several soft modem drivers, some network drivers, ...
> >
>
> Please go forward and discuss the UIO driver with Greg Kroah Hartmann
> and the fuse driver with the other people. If companies want to
> release binary drivers they can easily use the existing code put it
> into an RPM or DEB package and Ubuntu will pick it up.
>
> > With all those issues, I'm against to add an userspace interface for
> > tuners.
> >
>
> I'm against how the project works out at the moment and how it worked
> out in history. Exactly this way will kick off companies to be
> interested in future like Avermedia. A driver can easily be written
> within a few weeks and I've been struggling with it for 2 years(!!!)
> now just for nothing finally telling me that some guys want to steal
> my code and move it to kernelspace although it would raise more
> complications with upcoming and current devices which have even more
> requirements.
> I spent more time in rewriting and discussing everything than to get
> any of those requirements done.
> Look at the dvb hotplug patch which came from my side, also look at
> device node locking where the Hauppauge guy submitted a patch which
> doesn't work properly because of the spinning thread in the end. I
> would take my considerations and requirements a bit more serious.
>
> Markus
>
--
Markus Rechberger
On Thu, Sep 13, 2007, Markus Rechberger wrote:
> Let's add the LKML to this.
>
> On 9/13/07, Markus Rechberger <[email protected]> wrote:
> > On 9/12/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > > I don't see any technical reason why tuner drivers should be moved to
> > > userspace. Looking at xc3028 device, the driver is very simple and
> > > doesn't require any special treatment that it isn't possible to be done
> > > at kernel. There are already some implementations on kernelspace that
> > > works fine.
> >
> > As from my side to support the xceive driver properly it needs a
> > rewrite and a proper API description. Since it's not possible to
> > discuss any API changes
Not possible? We're doing it all the time...
However, your ideas were rejected in this discussion,
and you can't seem to get over it.
> > don't get me wrong but the existing community is rather small and
> > kicking off people who are interested in changing things.
IMHO there is a lack of openness caused by people being burned
in past flamewars. This makes it a bit difficult to see through
what happens and why, and to participate. However, I think it
is completely wrong to say that the community is "kicking off people".
> > I'm against how the project works out at the moment and how it worked
> > out in history. Exactly this way will kick off companies to be
> > interested in future like Avermedia. A driver can easily be written
> > within a few weeks and I've been struggling with it for 2 years(!!!)
> > now just for nothing finally telling me that some guys want to steal
> > my code and move it to kernelspace although it would raise more
> > complications with upcoming and current devices which have even more
> > requirements.
Oh dear, there we go again... more flame bait...
I reality, 95% of your driver code could have been merged
without problems, but you refused to take the small, objectionable
part out of the picture.
(For those interested:
http://mcentral.de/~mrec/patches/v4l-dvb/hg_v4l-dvb-experimental_01.patch
The patch changed the internal tuner API and required changes
to all tuner drivers.)
Your all-or-nothing approach didn't work out.
Out of curiosity: How does your userspace approach solve
the hybrid (analog/digital TV) tuner problem which was the
only objectionable part of your work? And why are the kernel
parts of your new interface now less objectionable? What changed?
Johannes
On 9/13/07, Johannes Stezenbach <[email protected]> wrote:
> On Thu, Sep 13, 2007, Markus Rechberger wrote:
> > Let's add the LKML to this.
> >
> > On 9/13/07, Markus Rechberger <[email protected]> wrote:
> > > On 9/12/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > > > I don't see any technical reason why tuner drivers should be moved to
> > > > userspace. Looking at xc3028 device, the driver is very simple and
> > > > doesn't require any special treatment that it isn't possible to be
> done
> > > > at kernel. There are already some implementations on kernelspace that
> > > > works fine.
> > >
> > > As from my side to support the xceive driver properly it needs a
> > > rewrite and a proper API description. Since it's not possible to
> > > discuss any API changes
>
> Not possible? We're doing it all the time...
>
Then ask Mauro about the audio standard discussion which I had with him.
> However, your ideas were rejected in this discussion,
> and you can't seem to get over it.
>
As for future projects I see other people having the same problems if
they want to join the project. If deeper requirements and/or ideas will
come up some particular people will try to run their own game.
If I'm doing something wrong technically then state it out show me
the bugs that I can understand what I did wrong or what I am doing
wrong. As for design changes if someone thinks he has to deny
something he has to state out _why_ and not only some overall
nontechnical statements which do not help anyone.
> > > don't get me wrong but the existing community is rather small and
> > > kicking off people who are interested in changing things.
>
> IMHO there is a lack of openness caused by people being burned
> in past flamewars. This makes it a bit difficult to see through
> what happens and why, and to participate. However, I think it
> is completely wrong to say that the community is "kicking off people".
>
You identified it already the right way in one mail much earlier. It's a
messup caused by many people and not only by one single person.
And that's also how I see it.
> > > I'm against how the project works out at the moment and how it worked
> > > out in history. Exactly this way will kick off companies to be
> > > interested in future like Avermedia. A driver can easily be written
> > > within a few weeks and I've been struggling with it for 2 years(!!!)
> > > now just for nothing finally telling me that some guys want to steal
> > > my code and move it to kernelspace although it would raise more
> > > complications with upcoming and current devices which have even more
> > > requirements.
>
> Oh dear, there we go again... more flame bait...
>
> I reality, 95% of your driver code could have been merged
> without problems, but you refused to take the small, objectionable
> part out of the picture.
>
the driver itself still evolves, so the main point is in getting those incomplete
APIs to support further requirements. Instead of acknowlidging and seriously
discussing the solutions of others all that's beeing done now is to redo things
hundred times and splitting development one side beeing more open (which
is the userspace work) and the other part beeing more closed to a few people
only (the inkernel work).
It's not my task to convince myself to rewrite the base of something which
I don't think that it would be valueable. The one who wants me that I
spend days in changing my code should try to convince me to change
my mind on that, unless he can provide patches which demonstrate
that his changes are definitelly an advantage over what has been done
from my side.
I will for sure acknowlidge everything that will seriously improve the work
which has been done.
> (For those interested:
> http://mcentral.de/~mrec/patches/v4l-dvb/hg_v4l-dvb-experimental_01.patch
> The patch changed the internal tuner API and required changes
> to all tuner drivers.)
>
The main problem is moreover that the RFC didn't get discussed properly, afterwards people wondered what happened, even though the sources
were also available all the time in a public repository.
> Your all-or-nothing approach didn't work out.
>
well we got far enough that I end up to step away and preparing the sources
to be able to get submitted beside linuxtv since I don't/didn't get any useful
technical feedback there.
Convince me that my work is welcome then I might start to submit smaller
patches, I already spent days for exporting patches in history and it all
end up nowhere but in unfriendly discussions - which also turned me to be
unfriendly to some people (which I've been told that I should ignore them
recently and also not continue to to argue on a nontechnical level .. althought
as mentioned I'm not the only one who made mistakes :-)
> Out of curiosity: How does your userspace approach solve
> the hybrid (analog/digital TV) tuner problem which was the
> only objectionable part of your work? And why are the kernel
> parts of your new interface now less objectionable? What changed?
>
It's only a step in development, I do not intend to keep the kernel
stub in the end, but I do intend to keep and use the userspace drivers
with i2c-dev in the long run, this requires a v4l/dvb library at the front
of everything. Till such a library shows up I plan to use the
kernelspace-userspace stub.
They are less objectionable because it adds a wrapper into the
subsystem which it didn't do before, the previous approach merged
the structs and further work would have been focussed on getting it
merged even better.
Markus
> It's only a step in development, I do not intend to keep the kernel
> stub in the end, but I do intend to keep and use the userspace drivers
> with i2c-dev in the long run, this requires a v4l/dvb library at the front
> of everything.
Well, this was what adq and myself did with libdvbapi and mti, (much
before UIO was announced at LK.) It is not tied to I2C either.
But, according to your statements (with regards to i2c-dev) you can
handle only some I2C based devices, which is in fact a subset of a
subset. Also not to forget that hardly a few I2C devices are in fact
"truly" I2C compatible. In fact many variables to be considered.
Also there is to consider a non technical aspect, whether vendors will
misuse this interface for binary only, undermining the efforts put in
for OSS drivers.
Manu
On 9/13/07, Manu Abraham <[email protected]> wrote:
>
> > It's only a step in development, I do not intend to keep the kernel
> > stub in the end, but I do intend to keep and use the userspace drivers
> > with i2c-dev in the long run, this requires a v4l/dvb library at the front
> > of everything.
>
> Well, this was what adq and myself did with libdvbapi and mti, (much
> before UIO was announced at LK.) It is not tied to I2C either.
>
I2C is the main communication path for it, although there are callback
mechanisms available which add the possibility for different configuration
paths.
> But, according to your statements (with regards to i2c-dev) you can
> handle only some I2C based devices, which is in fact a subset of a
> subset. Also not to forget that hardly a few I2C devices are in fact
> "truly" I2C compatible. In fact many variables to be considered.
>
The analogue tuner core is also only for i2c only devices which most
devices rely on.
> Also there is to consider a non technical aspect, whether vendors will
> misuse this interface for binary only, undermining the efforts put in
> for OSS drivers.
>
What holds companies for using the current available code putting it
into an rpm or deb package and releasing such code now?
The Avermedia example I pointed out to is a good example already.
As from my side I won't release binary drivers.
Although on the other side:
* are drivers from vendors which work through several kernel versions
that bad?
* Why did someone duallicense videodev2 with BSD/GPL?
I would appreciate if someone else on the list could also comment
the reason that drivers should all be included in the linuxkernel just
because forcing the companies to release binary drivers because
of that. My opinion about that is if a company wants to go opensource
they will do so, if not they will either not release a driver or release
nothing.
Markus
Markus Rechberger wrote:
> On 9/13/07, Manu Abraham <[email protected]> wrote:
>>> It's only a step in development, I do not intend to keep the kernel
>>> stub in the end, but I do intend to keep and use the userspace drivers
>>> with i2c-dev in the long run, this requires a v4l/dvb library at the front
>>> of everything.
>> Well, this was what adq and myself did with libdvbapi and mti, (much
>> before UIO was announced at LK.) It is not tied to I2C either.
>>
>
> I2C is the main communication path for it, although there are callback
> mechanisms available which add the possibility for different configuration
> paths.
Sorry, i must say that what you said is wrong.
The example implementation in libdvbapi/mti is I2C only with a STV0299
on the TTPCI, if that was what you meant.
But if you need examples for every possible interface, then probably you
are out of luck.
Manu
On 9/13/07, Manu Abraham <[email protected]> wrote:
> Markus Rechberger wrote:
> > On 9/13/07, Manu Abraham <[email protected]> wrote:
> >>> It's only a step in development, I do not intend to keep the kernel
> >>> stub in the end, but I do intend to keep and use the userspace drivers
> >>> with i2c-dev in the long run, this requires a v4l/dvb library at the
> front
> >>> of everything.
> >> Well, this was what adq and myself did with libdvbapi and mti, (much
> >> before UIO was announced at LK.) It is not tied to I2C either.
> >>
> >
> > I2C is the main communication path for it, although there are callback
> > mechanisms available which add the possibility for different configuration
> > paths.
>
> Sorry, i must say that what you said is wrong.
>
> The example implementation in libdvbapi/mti is I2C only with a STV0299
> on the TTPCI, if that was what you meant.
> But if you need examples for every possible interface, then probably you
> are out of luck.
>
I didn't comment the libdvbapi/mti driver.
I'm quite confident that non I2C protocols can be handled by a callback
function. In the end it's either a usb control command or pci mmio writes
in the current usual cases and such calls could be handled behind a
callback function which is implemented in the driver.
Markus
Markus Rechberger wrote:
> On 9/13/07, Manu Abraham <[email protected]> wrote:
>> Markus Rechberger wrote:
>>> On 9/13/07, Manu Abraham <[email protected]> wrote:
>>>>> It's only a step in development, I do not intend to keep the kernel
>>>>> stub in the end, but I do intend to keep and use the userspace drivers
>>>>> with i2c-dev in the long run, this requires a v4l/dvb library at the
>> front
>>>>> of everything.
>>>> Well, this was what adq and myself did with libdvbapi and mti, (much
>>>> before UIO was announced at LK.) It is not tied to I2C either.
>>>>
>>> I2C is the main communication path for it, although there are callback
>>> mechanisms available which add the possibility for different configuration
>>> paths.
>> Sorry, i must say that what you said is wrong.
>>
>> The example implementation in libdvbapi/mti is I2C only with a STV0299
>> on the TTPCI, if that was what you meant.
>> But if you need examples for every possible interface, then probably you
>> are out of luck.
>>
>
> I didn't comment the libdvbapi/mti driver.
> I'm quite confident that non I2C protocols can be handled by a callback
> function. In the end it's either a usb control command or pci mmio writes
There's also DTL in some cases. It's not USB msgs and or PCI MMIO alone.
The actual DTL spec is unfortunately not open.
Manu
>
>> Also there is to consider a non technical aspect, whether vendors will
>> misuse this interface for binary only, undermining the efforts put in
>> for OSS drivers.
>>
>>
>
> What holds companies for using the current available code putting it
> into an rpm or deb package and releasing such code now?
>
> The Avermedia example I pointed out to is a good example already.
> As from my side I won't release binary drivers.
> Although on the other side:
> * are drivers from vendors which work through several kernel versions
> that bad?
> * Why did someone duallicense videodev2 with BSD/GPL?
>
> I would appreciate if someone else on the list could also comment
> the reason that drivers should all be included in the linuxkernel just
> because forcing the companies to release binary drivers because
> of that. My opinion about that is if a company wants to go opensource
> they will do so, if not they will either not release a driver or release
> nothing.
>
>
I know for certain that adding a userland API tuner/demod interface to
the kernel, allowing non-caring opportunistic silicon or board vendors
to developer closed source proprietary drivers, will have a negative
effect on the community and we'd set back linuxtv 3-5 years.
I know for certain that it would happen. Trust me.
I've told you this countless times and you're not hearing me.
Hauppauge have some leverage with Conexant and NXP to release public
datasheets. If they just have to release a demod.so (or similar)
loadable, they'll defer to the board vendors and we'll see the certain
board vendors 'locking other board vendors' out of their drivers. We'll
see embedded firmware, not shared between drivers.
Except, it won't stop at demod.so. It will extend into unfixable bugs
for VendorB's board, because VendorA doesn't want to release a new
demod.so, and VendorB has no linux resources. What happens next? For
financial reasons - demod.so will begin to include checks to see if
specific PCI or USB devices are present in the system, and will fail to
work properly (if at all) when they're not being used with the preferred
products.
Read my lips: For commercial reasons, this enables driver components
that only work if specific boards are present.
- Steve
On 9/13/07, Steven Toth <[email protected]> wrote:
>
> >
> >> Also there is to consider a non technical aspect, whether vendors will
> >> misuse this interface for binary only, undermining the efforts put in
> >> for OSS drivers.
> >>
> >>
> >
> > What holds companies for using the current available code putting it
> > into an rpm or deb package and releasing such code now?
> >
> > The Avermedia example I pointed out to is a good example already.
> > As from my side I won't release binary drivers.
> > Although on the other side:
> > * are drivers from vendors which work through several kernel versions
> > that bad?
> > * Why did someone duallicense videodev2 with BSD/GPL?
> >
> > I would appreciate if someone else on the list could also comment
> > the reason that drivers should all be included in the linuxkernel just
> > because forcing the companies to release binary drivers because
> > of that. My opinion about that is if a company wants to go opensource
> > they will do so, if not they will either not release a driver or release
> > nothing.
> >
> >
>
>
> I know for certain that adding a userland API tuner/demod interface to
> the kernel, allowing non-caring opportunistic silicon or board vendors
> to developer closed source proprietary drivers, will have a negative
> effect on the community and we'd set back linuxtv 3-5 years.
>
> I know for certain that it would happen. Trust me.
>
> I've told you this countless times and you're not hearing me.
>
> Hauppauge have some leverage with Conexant and NXP to release public
> datasheets. If they just have to release a demod.so (or similar)
> loadable, they'll defer to the board vendors and we'll see the certain
> board vendors 'locking other board vendors' out of their drivers. We'll
> see embedded firmware, not shared between drivers.
>
> Except, it won't stop at demod.so. It will extend into unfixable bugs
> for VendorB's board, because VendorA doesn't want to release a new
> demod.so, and VendorB has no linux resources. What happens next? For
> financial reasons - demod.so will begin to include checks to see if
> specific PCI or USB devices are present in the system, and will fail to
> work properly (if at all) when they're not being used with the preferred
> products.
>
Steven,
what stops vendors of using the current existing code to achieve that
goal. They could provide binary drivers with the existing API.
What stops companies to intercept the ioctl calls and overriding some
I2C commands?
Since you know about windows drivers (at least I think that you know
about it) you might know about the limitations of the v4l/dvb API
in general the reason just put as much code as possible into the
kernel just for forcing companies to release code under GPL doesn't
seem to be valid.
How about proprietary video formats, would you also place the decoding
algorithms in kernel just to force companies to release their code
for it?
What do you think about the existing usbfs implementation, which
allows to implement usb drivers completly in userspace?
What do you think about IOMMU?
please answer those questions.
thanks,
Markus
Am Donnerstag, den 13.09.2007, 16:36 -0400 schrieb Steven Toth:
> >
> >> Also there is to consider a non technical aspect, whether vendors will
> >> misuse this interface for binary only, undermining the efforts put in
> >> for OSS drivers.
> >>
> >>
> >
> > What holds companies for using the current available code putting it
> > into an rpm or deb package and releasing such code now?
> >
> > The Avermedia example I pointed out to is a good example already.
> > As from my side I won't release binary drivers.
> > Although on the other side:
> > * are drivers from vendors which work through several kernel versions
> > that bad?
> > * Why did someone duallicense videodev2 with BSD/GPL?
> >
> > I would appreciate if someone else on the list could also comment
> > the reason that drivers should all be included in the linuxkernel just
> > because forcing the companies to release binary drivers because
> > of that. My opinion about that is if a company wants to go opensource
> > they will do so, if not they will either not release a driver or release
> > nothing.
> >
> >
>
>
> I know for certain that adding a userland API tuner/demod interface to
> the kernel, allowing non-caring opportunistic silicon or board vendors
> to developer closed source proprietary drivers, will have a negative
> effect on the community and we'd set back linuxtv 3-5 years.
>
> I know for certain that it would happen. Trust me.
>
> I've told you this countless times and you're not hearing me.
>
> Hauppauge have some leverage with Conexant and NXP to release public
> datasheets. If they just have to release a demod.so (or similar)
> loadable, they'll defer to the board vendors and we'll see the certain
> board vendors 'locking other board vendors' out of their drivers. We'll
> see embedded firmware, not shared between drivers.
>
> Except, it won't stop at demod.so. It will extend into unfixable bugs
> for VendorB's board, because VendorA doesn't want to release a new
> demod.so, and VendorB has no linux resources. What happens next? For
> financial reasons - demod.so will begin to include checks to see if
> specific PCI or USB devices are present in the system, and will fail to
> work properly (if at all) when they're not being used with the preferred
> products.
>
> Read my lips: For commercial reasons, this enables driver components
> that only work if specific boards are present.
>
> - Steve
>
I do confirn that I have all this, Steve mentions, really have seen
already!
Markus, sorry, they did abuse it and will do it again.
Hermann
Markus Rechberger wrote:
> On 9/13/07, Steven Toth <[email protected]> wrote:
>
>>>> Also there is to consider a non technical aspect, whether vendors will
>>>> misuse this interface for binary only, undermining the efforts put in
>>>> for OSS drivers.
>>>>
>>>>
>>>>
>>> What holds companies for using the current available code putting it
>>> into an rpm or deb package and releasing such code now?
>>>
>>> The Avermedia example I pointed out to is a good example already.
>>> As from my side I won't release binary drivers.
>>> Although on the other side:
>>> * are drivers from vendors which work through several kernel versions
>>> that bad?
>>> * Why did someone duallicense videodev2 with BSD/GPL?
>>>
>>> I would appreciate if someone else on the list could also comment
>>> the reason that drivers should all be included in the linuxkernel just
>>> because forcing the companies to release binary drivers because
>>> of that. My opinion about that is if a company wants to go opensource
>>> they will do so, if not they will either not release a driver or release
>>> nothing.
>>>
>>>
>>>
>> I know for certain that adding a userland API tuner/demod interface to
>> the kernel, allowing non-caring opportunistic silicon or board vendors
>> to developer closed source proprietary drivers, will have a negative
>> effect on the community and we'd set back linuxtv 3-5 years.
>>
>> I know for certain that it would happen. Trust me.
>>
>> I've told you this countless times and you're not hearing me.
>>
>> Hauppauge have some leverage with Conexant and NXP to release public
>> datasheets. If they just have to release a demod.so (or similar)
>> loadable, they'll defer to the board vendors and we'll see the certain
>> board vendors 'locking other board vendors' out of their drivers. We'll
>> see embedded firmware, not shared between drivers.
>>
>> Except, it won't stop at demod.so. It will extend into unfixable bugs
>> for VendorB's board, because VendorA doesn't want to release a new
>> demod.so, and VendorB has no linux resources. What happens next? For
>> financial reasons - demod.so will begin to include checks to see if
>> specific PCI or USB devices are present in the system, and will fail to
>> work properly (if at all) when they're not being used with the preferred
>> products.
>>
>>
> Steven,
>
> what stops vendors of using the current existing code to achieve that
> goal. They could provide binary drivers with the existing API.
>
>
Because the good people in this mailing list are keeping them honest.
Give any board or silicon company the ability to protect their IP, even
in the smallest way and they'll do it, and for no good technical reason.
It's a cut throat market and it's not clear that everyone understands
just how thin sales margins really are.
That means Hauppauge potentially releasing a binary driver, because it's
much easier than seeking silicon vendor permission for a public diver.
The net result of that would mean I'd have to leave the company and find
another company that practices the one thing I truly care about ....
open source and open development groups.
I'm keeping Hauppauge honest with their Linux involvement and I'm not
alone. Other devs in other linux subsystems in other companies are doing
the same thing.
Binary drivers (or binary components) leads Linux back in time.
I can't believe your so passionate about protecting secrets.
> What stops companies to intercept the ioctl calls and overriding some
> I2C commands?
>
>
Why would a company want to do that? Companies don't do that, hackers do
that.
> Since you know about windows drivers (at least I think that you know
> about it) you might know about the limitations of the v4l/dvb API
> in general the reason just put as much code as possible into the
> kernel just for forcing companies to release code under GPL doesn't
> seem to be valid.
>
It seems perfectly valid to me.
> How about proprietary video formats, would you also place the decoding
> algorithms in kernel just to force companies to release their code
> for it?
>
The kernel has no good API for those, each new type of video device and
suggested API is judged on it's own merits and discussed on the mailing
lists.
> What do you think about the existing usbfs implementation, which
> allows to implement usb drivers completly in userspace?
>
Those are not my problem and I don't use them, you should raise that
with the relevant usb-dev mailing lists. I'm here because I care about
linuxtv. Please stay on topic.
> What do you think about IOMMU?
>
>
Just because AMD or INTEL want to invent some whizzy new technology it
doesn't say anything about the TV card development and retail business.
Intel and AMD have teams of Linux engineers helping operating system
developers bring their ideas and technologies to new platforms. That's a
million miles away from any of the TV board vendors I know of, who have
little or NO fulltime linux developers and consider the < 5% market
fringe at best.
Markus, senior devs in the LinuxTV group are telling you, based on their
commercial experience, that userspace access is technically great, but
long term it will be used against the community and will ultimately hurt
linuxtv development.
If you want to reply and have the last word, go ahead, but repeating
what I've said on this list in the past - I'll never support the
userland tuning/demod idea.
I wanted to work directly with you on the em28xx tree, helping you
remove the 5% of code that Johannes referred to, but you said no. I
wanted to help you make the tree conform to the linuxtv standards for
frameworks, you said no.
If you care about LinuxTV you'll work with the core subsystem developers
to bring your em28xx tree inline. If you don't care then why are you here?
- Steve
On 9/14/07, Steven Toth <[email protected]> wrote:
> Markus Rechberger wrote:
> > On 9/13/07, Steven Toth <[email protected]> wrote:
> >
> >>>> Also there is to consider a non technical aspect, whether vendors will
> >>>> misuse this interface for binary only, undermining the efforts put in
> >>>> for OSS drivers.
> >>>>
> >>>>
> >>>>
> >>> What holds companies for using the current available code putting it
> >>> into an rpm or deb package and releasing such code now?
> >>>
> >>> The Avermedia example I pointed out to is a good example already.
> >>> As from my side I won't release binary drivers.
> >>> Although on the other side:
> >>> * are drivers from vendors which work through several kernel versions
> >>> that bad?
> >>> * Why did someone duallicense videodev2 with BSD/GPL?
> >>>
> >>> I would appreciate if someone else on the list could also comment
> >>> the reason that drivers should all be included in the linuxkernel just
> >>> because forcing the companies to release binary drivers because
> >>> of that. My opinion about that is if a company wants to go opensource
> >>> they will do so, if not they will either not release a driver or release
> >>> nothing.
> >>>
> >>>
> >>>
> >> I know for certain that adding a userland API tuner/demod interface to
> >> the kernel, allowing non-caring opportunistic silicon or board vendors
> >> to developer closed source proprietary drivers, will have a negative
> >> effect on the community and we'd set back linuxtv 3-5 years.
> >>
> >> I know for certain that it would happen. Trust me.
> >>
> >> I've told you this countless times and you're not hearing me.
> >>
> >> Hauppauge have some leverage with Conexant and NXP to release public
> >> datasheets. If they just have to release a demod.so (or similar)
> >> loadable, they'll defer to the board vendors and we'll see the certain
> >> board vendors 'locking other board vendors' out of their drivers. We'll
> >> see embedded firmware, not shared between drivers.
> >>
> >> Except, it won't stop at demod.so. It will extend into unfixable bugs
> >> for VendorB's board, because VendorA doesn't want to release a new
> >> demod.so, and VendorB has no linux resources. What happens next? For
> >> financial reasons - demod.so will begin to include checks to see if
> >> specific PCI or USB devices are present in the system, and will fail to
> >> work properly (if at all) when they're not being used with the preferred
> >> products.
> >>
> >>
> > Steven,
> >
> > what stops vendors of using the current existing code to achieve that
> > goal. They could provide binary drivers with the existing API.
> >
> >
>
> Because the good people in this mailing list are keeping them honest.
> Give any board or silicon company the ability to protect their IP, even
> in the smallest way and they'll do it, and for no good technical reason.
> It's a cut throat market and it's not clear that everyone understands
> just how thin sales margins really are.
>
> That means Hauppauge potentially releasing a binary driver, because it's
> much easier than seeking silicon vendor permission for a public diver.
> The net result of that would mean I'd have to leave the company and find
> another company that practices the one thing I truly care about ....
> open source and open development groups.
>
well you could already release binary drivers if you would be
concerned about it, so again all that seems to be no reason for me.
What stops Hauppauge in implementing drivers another way?
For example the USB drivers completly in userspace.
> I'm keeping Hauppauge honest with their Linux involvement and I'm not
> alone. Other devs in other linux subsystems in other companies are doing
> the same thing.
>
For example the UIO thing, accessing MMIO through userspace, also
accessing usb control messages from userspace. Because of that reason
since it is already possible to provide binary drivers your reason is
again not valid.
The code which the userspace tuners are connected to is GPL so what.
Beside that Hauppauge is not the only company out there although I also
have contacts there at Hauppauge Europe.
> Binary drivers (or binary components) leads Linux back in time.
>
> I can't believe your so passionate about protecting secrets.
>
Please point me to the part where I am passionated about protecting
any of my opensource code which is currently available.
I have other arguments why to put that part into userspace.
> > What stops companies to intercept the ioctl calls and overriding some
> > I2C commands?
> >
> >
>
> Why would a company want to do that? Companies don't do that, hackers do
> that.
>
just to achieve what you're trying to argue with that companies could
use some odd constructs which could be used to publish their drivers as
binary only. I don't see the problem there if they want to do so.
> > Since you know about windows drivers (at least I think that you know
> > about it) you might know about the limitations of the v4l/dvb API
> > in general the reason just put as much code as possible into the
> > kernel just for forcing companies to release code under GPL doesn't
> > seem to be valid.
> >
>
> It seems perfectly valid to me.
>
> > How about proprietary video formats, would you also place the decoding
> > algorithms in kernel just to force companies to release their code
> > for it?
> >
>
> The kernel has no good API for those, each new type of video device and
> suggested API is judged on it's own merits and discussed on the mailing
> lists.
>
my experience is that there's no way of discussing things properly
with some guys there. Hey if someone wants to get his device work
somehow he's invited to join the whole project, what happened with
that project during the last 2 years (and probably before already)
does not seem to be an open community to anything.
> > What do you think about the existing usbfs implementation, which
> > allows to implement usb drivers completly in userspace?
> >
>
> Those are not my problem and I don't use them, you should raise that
> with the relevant usb-dev mailing lists. I'm here because I care about
> linuxtv. Please stay on topic.
>
I'm not sure if you are aware of that userspace usb video driver which got
published on the video4linux mailinglist. It hijacked ioctl calls and used usbfs
for having the whole driver in userspace.
I would like you to have a look at:
http://www.harmwal.nl/pccam880/
"This is a user space video4linux driver for the PC-CAM 880 and other
Zoran/Coach based USB webcams"
> > What do you think about IOMMU?
> >
> >
> Just because AMD or INTEL want to invent some whizzy new technology it
> doesn't say anything about the TV card development and retail business.
> Intel and AMD have teams of Linux engineers helping operating system
> developers bring their ideas and technologies to new platforms. That's a
> million miles away from any of the TV board vendors I know of, who have
> little or NO fulltime linux developers and consider the < 5% market
> fringe at best.
>
it helps to virtualize devices and introduces newer features for that.
Some interesting projects could be derrived out of that, there are
quite a few interesting papers floating around how drivers could be
handled in future.
> Markus, senior devs in the LinuxTV group are telling you, based on their
> commercial experience, that userspace access is technically great, but
> long term it will be used against the community and will ultimately hurt
> linuxtv development.
>
I've seriously seen where senior devs lead other people during the last year,
repeatly pointing to wrong solutions and finally redoing things a completly
other way. Protecting their own interests and not seriously looking to get
more people on board for fixing up those API issues.
> If you want to reply and have the last word, go ahead, but repeating
> what I've said on this list in the past - I'll never support the
> userland tuning/demod idea.
>
> I wanted to work directly with you on the em28xx tree, helping you
> remove the 5% of code that Johannes referred to, but you said no. I
> wanted to help you make the tree conform to the linuxtv standards for
> frameworks, you said no.
>
> If you care about LinuxTV you'll work with the core subsystem developers
> to bring your em28xx tree inline. If you don't care then why are you here?
It doesn't really work out to work with those 3-5 "core" people who are active
there.
It starts at the point where RFCs are submitted and not answered,
discussions are avoided and finally some people start to complain.
In case of pointing people to wrong or bad solutions, one "Maintainer"
pointed out it would be like manipulating others work ... this exactly
happened with the em28xx project. It's not only about the current
implementation, as from my side I also take upcoming products into
account and how long it would take to get something which isn't
supported by the API fixed.. it's just something that is too hard and
I don't want to debate about things where I know that the outcome
is that I have to wait till someone of those 3-5 "core" people come up
with an own idea.
This is my impression of the overall situation you might
try to change it during the next few months but in case of further
implementations I will for sure try to get around that situation because
in history I only received nontechnical arguments why my stuff wasn't
accepted (again this starts at the RFCs).
If you want to comment this, please take all parts of that response into
account, especially the userspace v4l driver which I pointed to.
Markus
>>> What do you think about IOMMU?
>>>
>>>
>> Just because AMD or INTEL want to invent some whizzy new technology it
>> doesn't say anything about the TV card development and retail business.
>> Intel and AMD have teams of Linux engineers helping operating system
>> developers bring their ideas and technologies to new platforms. That's a
>> million miles away from any of the TV board vendors I know of, who have
>> little or NO fulltime linux developers and consider the < 5% market
>> fringe at best.
>>
>
> it helps to virtualize devices and introduces newer features for that.
> Some interesting projects could be derrived out of that, there are
> quite a few interesting papers floating around how drivers could be
> handled in future.
>
IOMMU can be considered similar to the AGP GART, which is similar,
remapping the Addresses, as far as i understand.
Though you get a physical to virtual translation, what about interrupts,
yet to be seen with how to do it with multipath scenarios.
Something that i happened to read
https://www.gelato.unsw.edu.au/archives/comp-arch/2007-March/007370.html
Even Documentation/Intel-IOMMU.txt, doesn't seem to paint a very rosy
picture
Manu
On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
> >>> What do you think about IOMMU?
> >>>
> >>>
> >> Just because AMD or INTEL want to invent some whizzy new technology it
> >> doesn't say anything about the TV card development and retail business.
> >> Intel and AMD have teams of Linux engineers helping operating system
> >> developers bring their ideas and technologies to new platforms. That's a
> >> million miles away from any of the TV board vendors I know of, who have
> >> little or NO fulltime linux developers and consider the < 5% market
> >> fringe at best.
> >>
> >
> > it helps to virtualize devices and introduces newer features for that.
> > Some interesting projects could be derrived out of that, there are
> > quite a few interesting papers floating around how drivers could be
> > handled in future.
> >
>
> IOMMU can be considered similar to the AGP GART, which is similar,
> remapping the Addresses, as far as i understand.
Common new IOMMUs have only very few in common with the AGP GART. In
fact, with current modern IOMMU hardware it will be possible to
implement secure userspace device drivers that are even able to do DMA.
This is not possible with the GART.
> Though you get a physical to virtual translation, what about interrupts,
Modern IOMMUs are able to remap interrupts. This will solve the problem
with PCI interrupt sharing.
Joerg
On Fri, Sep 14, 2007, Markus Rechberger wrote:
> On 9/14/07, Steven Toth <[email protected]> wrote:
> >
> > If you care about LinuxTV you'll work with the core subsystem developers
> > to bring your em28xx tree inline. If you don't care then why are you here?
>
> It doesn't really work out to work with those 3-5 "core" people who are active
> there.
> It starts at the point where RFCs are submitted and not answered,
> discussions are avoided and finally some people start to complain.
> In case of pointing people to wrong or bad solutions, one "Maintainer"
> pointed out it would be like manipulating others work ... this exactly
> happened with the em28xx project. It's not only about the current
> implementation, as from my side I also take upcoming products into
> account and how long it would take to get something which isn't
> supported by the API fixed.. it's just something that is too hard and
> I don't want to debate about things where I know that the outcome
> is that I have to wait till someone of those 3-5 "core" people come up
> with an own idea.
Maybe you still don't realize how tiresome it is to talk to you.
What you present as "linuxtv people block my contributions" is
IMHO "linuxtv people got fed up talking to you". Because when
people disagree with you, you keep rambling on and on instead
of just accepting it. See, working with an Open source community
requires that you don't piss everyone off, but instead find
ways to _motivate_ them to help you fix the problems which
prevent your code from being merged. When people are doing
software development _for fun_, then it should be a _pleasure_
for them to work with you, and not a pain in the ass.
You code and your arguments in this discussion may be
different than before, but the thread follows a familiar
pattern and will likely go nowhere, just like the many
discussions before.
Johannes
Joerg Roedel wrote:
> On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
>>>>> What do you think about IOMMU?
>>>>>
>>>>>
>>>> Just because AMD or INTEL want to invent some whizzy new technology it
>>>> doesn't say anything about the TV card development and retail business.
>>>> Intel and AMD have teams of Linux engineers helping operating system
>>>> developers bring their ideas and technologies to new platforms. That's a
>>>> million miles away from any of the TV board vendors I know of, who have
>>>> little or NO fulltime linux developers and consider the < 5% market
>>>> fringe at best.
>>>>
>>> it helps to virtualize devices and introduces newer features for that.
>>> Some interesting projects could be derrived out of that, there are
>>> quite a few interesting papers floating around how drivers could be
>>> handled in future.
>>>
>> IOMMU can be considered similar to the AGP GART, which is similar,
>> remapping the Addresses, as far as i understand.
>
> Common new IOMMUs have only very few in common with the AGP GART. In
> fact, with current modern IOMMU hardware it will be possible to
> implement secure userspace device drivers that are even able to do DMA.
> This is not possible with the GART.
>
>> Though you get a physical to virtual translation, what about interrupts,
>
> Modern IOMMUs are able to remap interrupts. This will solve the problem
> with PCI interrupt sharing.
What CPU's are we talking about ?
Thanks for the explanation.
Manu
> > * Why did someone duallicense videodev2 with BSD/GPL?
Originally the BSD people wanted to share the interface for user space
compatibility.
The kernel however is GPL so the BSD licence on some bits of the code
isn't really relevant as the combination of bits you depend upon is GPL,
will remain GPL and the only way to do a binary module would be to show
it is not in any way a derivative work in law.
> I know for certain that adding a userland API tuner/demod interface to
> the kernel, allowing non-caring opportunistic silicon or board vendors
> to developer closed source proprietary drivers, will have a negative
> effect on the community and we'd set back linuxtv 3-5 years.
Agreed entirely
On Fri, Sep 14, 2007 at 01:17:05AM +0200, Markus Rechberger wrote:
> what stops vendors of using the current existing code to achieve that
> goal. They could provide binary drivers with the existing API.
If you feel lucky about the GPL
> What stops companies to intercept the ioctl calls and overriding some
> I2C commands?
The GPL - derivative work is the boundary not code linkage. Possibly a userspace
tuner hack would probably fit this too. Especially if a specific vendor is
producing both bits together and trying to claim they are independant.
> How about proprietary video formats, would you also place the decoding
> algorithms in kernel just to force companies to release their code
> for it?
No, I would assume they'd provide a proprietary conversion library that
no nobody would use (just like their hw). We keep format conversion firmly
seperated from hardware I/O processing.
Alan
On 9/14/07, Manu Abraham <[email protected]> wrote:
> Joerg Roedel wrote:
> > On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
> >>>>> What do you think about IOMMU?
> >>>>>
> >>>>>
> >>>> Just because AMD or INTEL want to invent some whizzy new technology it
> >>>> doesn't say anything about the TV card development and retail business.
> >>>> Intel and AMD have teams of Linux engineers helping operating system
> >>>> developers bring their ideas and technologies to new platforms. That's
> a
> >>>> million miles away from any of the TV board vendors I know of, who have
> >>>> little or NO fulltime linux developers and consider the < 5% market
> >>>> fringe at best.
> >>>>
> >>> it helps to virtualize devices and introduces newer features for that.
> >>> Some interesting projects could be derrived out of that, there are
> >>> quite a few interesting papers floating around how drivers could be
> >>> handled in future.
> >>>
> >> IOMMU can be considered similar to the AGP GART, which is similar,
> >> remapping the Addresses, as far as i understand.
> >
> > Common new IOMMUs have only very few in common with the AGP GART. In
> > fact, with current modern IOMMU hardware it will be possible to
> > implement secure userspace device drivers that are even able to do DMA.
> > This is not possible with the GART.
> >
> >> Though you get a physical to virtual translation, what about interrupts,
> >
> > Modern IOMMUs are able to remap interrupts. This will solve the problem
> > with PCI interrupt sharing.
>
> What CPU's are we talking about ?
>
upcoming x86-64 processors from the AMD side. I'm not into what Intel
is doing in that area at the moment.
Markus
On Fri, Sep 14, 2007 at 04:00:30PM +0400, Manu Abraham wrote:
> Joerg Roedel wrote:
> > On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
> >>>>> What do you think about IOMMU?
> >>>>>
> >>>>>
> >>>> Just because AMD or INTEL want to invent some whizzy new technology it
> >>>> doesn't say anything about the TV card development and retail business.
> >>>> Intel and AMD have teams of Linux engineers helping operating system
> >>>> developers bring their ideas and technologies to new platforms. That's a
> >>>> million miles away from any of the TV board vendors I know of, who have
> >>>> little or NO fulltime linux developers and consider the < 5% market
> >>>> fringe at best.
> >>>>
> >>> it helps to virtualize devices and introduces newer features for that.
> >>> Some interesting projects could be derrived out of that, there are
> >>> quite a few interesting papers floating around how drivers could be
> >>> handled in future.
> >>>
> >> IOMMU can be considered similar to the AGP GART, which is similar,
> >> remapping the Addresses, as far as i understand.
> >
> > Common new IOMMUs have only very few in common with the AGP GART. In
> > fact, with current modern IOMMU hardware it will be possible to
> > implement secure userspace device drivers that are even able to do DMA.
> > This is not possible with the GART.
> >
> >> Though you get a physical to virtual translation, what about interrupts,
> >
> > Modern IOMMUs are able to remap interrupts. This will solve the problem
> > with PCI interrupt sharing.
>
> What CPU's are we talking about ?
IOMMUs are not necessarily a CPU feature. These IOMMUS will be found in
the South/North Bridge or even on PCI devices itself (uncommon). The
Calgary IOMMU is such an example of an IOMMU not implemented on the CPU
itself.
Joerg
On 9/14/07, Alan Cox <[email protected]> wrote:
> On Fri, Sep 14, 2007 at 01:17:05AM +0200, Markus Rechberger wrote:
> > what stops vendors of using the current existing code to achieve that
> > goal. They could provide binary drivers with the existing API.
>
> If you feel lucky about the GPL
>
> > What stops companies to intercept the ioctl calls and overriding some
> > I2C commands?
>
> The GPL - derivative work is the boundary not code linkage. Possibly a
> userspace
> tuner hack would probably fit this too. Especially if a specific vendor is
> producing both bits together and trying to claim they are independant.
>
> > How about proprietary video formats, would you also place the decoding
> > algorithms in kernel just to force companies to release their code
> > for it?
>
> No, I would assume they'd provide a proprietary conversion library that
> no nobody would use (just like their hw). We keep format conversion firmly
> seperated from hardware I/O processing.
>
In general there are known formats available, the drawback is that every TV
application deals with it in a non-unified way at the moment, meaning alot
code duplication in userspace since there's no library available at the moment
which does the videoconversion from one to another format. Such a library is
beeing developed at the moment which gets plugged infront of accessing
the devicenodes.
Although in the long run I'm looking forward to reuse the userspace tuners with
such a library infront of everything by using i2c-dev.
This would also make it easy to reuse the sample code of several companies,
without having to cut out several features of the drivers and to rewrite them
to an inkernel format.
Markus
On 9/14/07, Johannes Stezenbach <[email protected]> wrote:
> On Fri, Sep 14, 2007, Markus Rechberger wrote:
> > On 9/14/07, Steven Toth <[email protected]> wrote:
> > >
> > > If you care about LinuxTV you'll work with the core subsystem developers
> > > to bring your em28xx tree inline. If you don't care then why are you
> here?
> >
> > It doesn't really work out to work with those 3-5 "core" people who are
> active
> > there.
> > It starts at the point where RFCs are submitted and not answered,
> > discussions are avoided and finally some people start to complain.
> > In case of pointing people to wrong or bad solutions, one "Maintainer"
> > pointed out it would be like manipulating others work ... this exactly
> > happened with the em28xx project. It's not only about the current
> > implementation, as from my side I also take upcoming products into
> > account and how long it would take to get something which isn't
> > supported by the API fixed.. it's just something that is too hard and
> > I don't want to debate about things where I know that the outcome
> > is that I have to wait till someone of those 3-5 "core" people come up
> > with an own idea.
>
> Maybe you still don't realize how tiresome it is to talk to you.
> What you present as "linuxtv people block my contributions" is
> IMHO "linuxtv people got fed up talking to you". Because when
> people disagree with you, you keep rambling on and on instead
> of just accepting it. See, working with an Open source community
> requires that you don't piss everyone off, but instead find
> ways to _motivate_ them to help you fix the problems which
> prevent your code from being merged. When people are doing
> software development _for fun_, then it should be a _pleasure_
> for them to work with you, and not a pain in the ass.
>
Johannes,
people do contribute to the em28xx project.
If noone keeps finding solutions for requirements I will of course
go on to find my own way.
Although upcoming and even the current requirements are not met
by the existing API.
It's worth nothing to merge what's available now since I'm not even
ok with how several issues are solved with the driver itself at the
moment.
I'm going forward step by step with it now.
there's also an active and even problem solving oriented ML available:
http://mcentral.de/pipermail/em28xx/
Also if you look at the mercurial code you'll see several people
contributing to that project.
Markus
Joerg Roedel wrote:
>>> Common new IOMMUs have only very few in common with the AGP GART. In
>>> fact, with current modern IOMMU hardware it will be possible to
>>> implement secure userspace device drivers that are even able to do DMA.
>>> This is not possible with the GART.
>>>
>>>> Though you get a physical to virtual translation, what about interrupts,
>>> Modern IOMMUs are able to remap interrupts. This will solve the problem
>>> with PCI interrupt sharing.
>> What CPU's are we talking about ?
>
> IOMMUs are not necessarily a CPU feature. These IOMMUS will be found in
> the South/North Bridge or even on PCI devices itself (uncommon). The
> Calgary IOMMU is such an example of an IOMMU not implemented on the CPU
> itself.
>
I do understand that (an earlier reply from Grant Grundler on the same
[1], while working on something else), but that wasn't exactly what i
was getting at. The bridges are in fact tied up with a certain CPU class.
Though your argument holds true: "secure userpsace device drivers can be
implemented with modern hardware" There is a large flaw in it. (From an
academic POV, you are correct)
Now, if all the drivers were to depend on that certain feature, what
happens to all other CPU class users ? Looking at a pile of CPU's being
used, also not to forget that devices such as STB's use even very small
embedded CPU's such as the PPC405 Vulcan based [2] to mobile devices
such as mobile phones using ARM, Xtensa [3], OMAP CPU's/platforms, which
do not in any way use the bridges nor the CPU class which you however
mention.
So .. we are looking at a small segment, ie. a subset of the PC users
even, even if the larger segments like STB's can be ignored. This would
mean that only a small subset of users using a certain CPU class can use
those drivers (eventhough the devices themselves don't have that
limitation, the limitation being the implementation of the driver
alone), which is absurd.
Manu
[1] http://lkml.org/lkml/2007/5/26/217
[2] http://abraham.manu.googlepages.com/p3160033.jpg
[3] http://tensilica.com/
Markus,
> >
> > Maybe you still don't realize how tiresome it is to talk to you.
> > What you present as "linuxtv people block my contributions" is
> > IMHO "linuxtv people got fed up talking to you". Because when
> > people disagree with you, you keep rambling on and on instead
> > of just accepting it. See, working with an Open source community
> > requires that you don't piss everyone off, but instead find
> > ways to _motivate_ them to help you fix the problems which
> > prevent your code from being merged. When people are doing
> > software development _for fun_, then it should be a _pleasure_
> > for them to work with you, and not a pain in the ass.
> >
>
> Johannes,
>
> people do contribute to the em28xx project.
> If noone keeps finding solutions for requirements I will of course
> go on to find my own way.
> Although upcoming and even the current requirements are not met
> by the existing API.
> It's worth nothing to merge what's available now since I'm not even
> ok with how several issues are solved with the driver itself at the
> moment.
> I'm going forward step by step with it now.
>
> there's also an active and even problem solving oriented ML available:
> http://mcentral.de/pipermail/em28xx/
>
> Also if you look at the mercurial code you'll see several people
> contributing to that project.
Solutions for your requirements can be reachable via a kernelspace
solution:
- The hybrid tuner support, that where your requirement, when all those
discussions started, were already added to the subsystem. So now, an
hybrid tuner can be accessed by both DVB and V4L devices;
- Audio standard selection is already possible via S_STD (it is already
working for a few standards, like NTSC/M JP and NTSC/M KR). Maybe more
standards should be needed, but hey, we still have 34 bits available at
std mask.
The point is: there's no technical need for userspace. This will just
add userless complexity.
However, you insist with your selfish idea that every other solution,
except the one you're proposing are bogus. This is not the way Open
Source development works. You should listen to people.
Cheers,
Mauro
On Fri, Sep 14, 2007 at 08:56:40PM +0400, Manu Abraham wrote:
>
> I do understand that (an earlier reply from Grant Grundler on the same
> [1], while working on something else), but that wasn't exactly what i
> was getting at. The bridges are in fact tied up with a certain CPU class.
>
> Though your argument holds true: "secure userpsace device drivers can be
> implemented with modern hardware" There is a large flaw in it. (From an
> academic POV, you are correct)
>
> Now, if all the drivers were to depend on that certain feature, what
> happens to all other CPU class users ? Looking at a pile of CPU's being
> used, also not to forget that devices such as STB's use even very small
> embedded CPU's such as the PPC405 Vulcan based [2] to mobile devices
> such as mobile phones using ARM, Xtensa [3], OMAP CPU's/platforms, which
> do not in any way use the bridges nor the CPU class which you however
> mention.
This is true. These platforms do not (and afaik will not) have an
IOMMU and thus its impossible to implement a secure userspace driver
interface that supports DMA. In general, IOMMUs are only expected on
platforms which implement virtualization, have a processor address space
larger than the io address space or both.
> So .. we are looking at a small segment, ie. a subset of the PC users
> even, even if the larger segments like STB's can be ignored. This would
> mean that only a small subset of users using a certain CPU class can use
> those drivers (eventhough the devices themselves don't have that
> limitation, the limitation being the implementation of the driver
> alone), which is absurd.
This is also true. But looking at the current development of
virtualization in hardware on the PC market (and also the increasing
amount of main memory) makes me sure that in a few years every new sold
_PC_ will haven an IOMMU inside. But the problems with the other
platforms or older hardware will certainly remain.
Joerg
On 9/14/07, Mauro Carvalho Chehab <[email protected]> wrote:
> Markus,
>
> > >
> > > Maybe you still don't realize how tiresome it is to talk to you.
> > > What you present as "linuxtv people block my contributions" is
> > > IMHO "linuxtv people got fed up talking to you". Because when
> > > people disagree with you, you keep rambling on and on instead
> > > of just accepting it. See, working with an Open source community
> > > requires that you don't piss everyone off, but instead find
> > > ways to _motivate_ them to help you fix the problems which
> > > prevent your code from being merged. When people are doing
> > > software development _for fun_, then it should be a _pleasure_
> > > for them to work with you, and not a pain in the ass.
> > >
> >
> > Johannes,
> >
> > people do contribute to the em28xx project.
> > If noone keeps finding solutions for requirements I will of course
> > go on to find my own way.
> > Although upcoming and even the current requirements are not met
> > by the existing API.
> > It's worth nothing to merge what's available now since I'm not even
> > ok with how several issues are solved with the driver itself at the
> > moment.
> > I'm going forward step by step with it now.
> >
> > there's also an active and even problem solving oriented ML available:
> > http://mcentral.de/pipermail/em28xx/
> >
> > Also if you look at the mercurial code you'll see several people
> > contributing to that project.
>
> Solutions for your requirements can be reachable via a kernelspace
> solution:
>
> - The hybrid tuner support, that where your requirement, when all those
> discussions started, were already added to the subsystem. So now, an
> hybrid tuner can be accessed by both DVB and V4L devices;
>
It's far more complex as the thing which is implemented there.
The only thing that has been implemented is that one moduleformat
can be loaded by the v4l and by the dvb framework nothing else at the
moment. I told you at the kernel summit about that and I think you
knew about that before.
Just to quote some text:
"Right now, a separate instance of the driver is used for analog /
digital tuning. In order to use a single instance, we will have to
store a pointer to the dvb_frontend structure on the bridge level, but
there are various other prerequisites that must be dealt with before we
get to that point.
We _will_ get there though, eventually."
> - Audio standard selection is already possible via S_STD (it is already
> working for a few standards, like NTSC/M JP and NTSC/M KR). Maybe more
> standards should be needed, but hey, we still have 34 bits available at
> std mask.
>
Let me quote some text where you've been in CC and which didn't get
far enough to get a solution implemented.
(Michael Schimek)
"> xc3028_BG_PAL_A2_A.i2c FW_78 > B/G PAL A2
> Group delay: See RecITU-R BT.470-6 p.21 Response "A"
> xc3028_BG_PAL_A2_A_MTS.i2c FW_78_MTS B/G PAL A2 ZWEITON
> xc3028_BG_PAL_A2_B.i2c FW_78 B/G PAL A2
> Group delay: See RecITU-R BT.470-6 p.21 Response "B"
> xc3028_BG_PAL_A2_B_MTS.i2c FW_78_MTS B/G PAL A2 ZWEITON
> xc3028_BG_PAL_NICAM_A.i2c FW_78 B/G PAL NICAM
> Group delay: See RecITU-R BT.470-6 p.21 Response "A"
> xc3028_BG_PAL_NICAM_A_MTS.i2c FW_78_MTS B/G PAL FM
> xc3028_BG_PAL_NICAM_B.i2c FW_78 B/G PAL NICAM
> Group delay: See RecITU-R BT.470-6 p.21 Response "B"
> xc3028_BG_PAL_NICAM_B_MTS.i2c FW_78_MTS B/G PAL FM
We cannot add new standards for each of these files because only six
bits are unassigned in the lower half of v4l2_std_id. It seems
unecessary too, please correct me if I'm wrong.
(Well the driver could define its own video standards for each of the
firmwares and put them into the upper 32 bits of v4l2_std_id, which were
set aside for this purpose. But adding standards to the API also has its
advantages. Maybe it's time to reserve bits 40-55 for future expansion.)
I suppose you choose firmwares with IF or baseband sound output
depending on the design of the card?"
> The point is: there's no technical need for userspace. This will just
> add userless complexity.
>
> However, you insist with your selfish idea that every other solution,
> except the one you're proposing are bogus. This is not the way Open
> Source development works. You should listen to people.
>
I pointed out a few requirements which didn't get commented at all, and
I explained why things where done in a particular way.
Markus
> > - The hybrid tuner support, that where your requirement, when all those
> > discussions started, were already added to the subsystem. So now, an
> > hybrid tuner can be accessed by both DVB and V4L devices;
> >
>
> It's far more complex as the thing which is implemented there.
> The only thing that has been implemented is that one moduleformat
> can be loaded by the v4l and by the dvb framework nothing else at the
> moment. I told you at the kernel summit about that and I think you
> knew about that before.
>
> Just to quote some text:
> "Right now, a separate instance of the driver is used for analog /
> digital tuning. In order to use a single instance, we will have to
> store a pointer to the dvb_frontend structure on the bridge level, but
> there are various other prerequisites that must be dealt with before we
> get to that point.
>
> We _will_ get there though, eventually."
Maybe it is still not perfect. Feel free to improve it. Several people
from the community, including me, already offered you help to add your
driver, reworking on the 5% of the stuff that aren't compatible with the
V4L/DVB core API design.
> > - Audio standard selection is already possible via S_STD (it is already
> > working for a few standards, like NTSC/M JP and NTSC/M KR). Maybe more
> > standards should be needed, but hey, we still have 34 bits available at
> > std mask.
> >
>
> Let me quote some text where you've been in CC and which didn't get
> far enough to get a solution implemented.
>
> (Michael Schimek)
>
> "> xc3028_BG_PAL_A2_A.i2c FW_78 > B/G PAL A2
> > Group delay: See RecITU-R BT.470-6 p.21 Response "A"
> > xc3028_BG_PAL_A2_A_MTS.i2c FW_78_MTS B/G PAL A2 ZWEITON
> > xc3028_BG_PAL_A2_B.i2c FW_78 B/G PAL A2
> > Group delay: See RecITU-R BT.470-6 p.21 Response "B"
> > xc3028_BG_PAL_A2_B_MTS.i2c FW_78_MTS B/G PAL A2 ZWEITON
>
> > xc3028_BG_PAL_NICAM_A.i2c FW_78 B/G PAL NICAM
> > Group delay: See RecITU-R BT.470-6 p.21 Response "A"
> > xc3028_BG_PAL_NICAM_A_MTS.i2c FW_78_MTS B/G PAL FM
> > xc3028_BG_PAL_NICAM_B.i2c FW_78 B/G PAL NICAM
> > Group delay: See RecITU-R BT.470-6 p.21 Response "B"
> > xc3028_BG_PAL_NICAM_B_MTS.i2c FW_78_MTS B/G PAL FM
>
> We cannot add new standards for each of these files because only six
> bits are unassigned in the lower half of v4l2_std_id. It seems
> unecessary too, please correct me if I'm wrong.
We can use the full 64 bits of v4l2_std_id. We just need to take some
care.
Due to the lack of __ucmpdi2 on ppc32 architecture, with some gcc
versions, a hack were added at v4l2-common.c, that truncates it to 32
bits, at the function v4l2_norm_to_name():
http://git.kernel.org/?p=linux/kernel/git/mchehab/v4l-dvb.git;a=commit;h=412297d31d439ba56cd4faeb3a49a6f569f40702
So, using more than 32 bits is possible, providing that we change the
implementation of v4l2_norm_to_name() or find a way for it to work with
ppc32.
Instead of just adding a standard for each possible combination, we may
just add bits for the supported audio formats. For example, we can use
the bitmask as:
#define V4L2_STD_AUDIO_NICAM_A ((v4l2_std_id)0x04000000)
#define V4L2_STD_AUDIO_A2_A ((v4l2_std_id)0x08000000)
Since for all other chipsets but xc3028, all audio standards are
supported, maybe we can, instead, use a negate bitmap logic for the
non-supported audio standards. Something like:
#define V4L2_STD_AUDIO_NOT_NICAM_A ((v4l2_std_id)0x04000000)
#define V4L2_STD_AUDIO_NOT_A2_A ((v4l2_std_id)0x08000000)
And define some macros for the specific standards you need. For example:
#define V4L2_STD_AUDIO_NOT_ALL V4L2_STD_AUDIO_NOT_A2_A | V4L2_STD_AUDIO_NOT_NICAM_A
#define V4L2_STD_PAL_BG_A2 V4L2_STD_PAL_BG | (V4L2_STD_AUDIO_NOT_ALL & !V4L2_STD_AUDIO_NOT_A2_A)
This way, V4L2_STD_PAL_BG will mean that this PAL/BG accepts all
possible audio standards (being binary compatible), while
V4L2_STD_PAL_BG_A2_A will mean that only A2 response 'A' audio format is
supported for PAL/BG.
--
Cheers,
Mauro
On 9/14/07, Markus Rechberger <[email protected]> wrote:
> On 9/14/07, Manu Abraham <[email protected]> wrote:
> > Joerg Roedel wrote:
> > > On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
> > >>>>> What do you think about IOMMU?
> > >>>>>
> > >>>>>
> > >>>> Just because AMD or INTEL want to invent some whizzy new technology it
> > >>>> doesn't say anything about the TV card development and retail business.
> > >>>> Intel and AMD have teams of Linux engineers helping operating system
> > >>>> developers bring their ideas and technologies to new platforms. That's
> > a
> > >>>> million miles away from any of the TV board vendors I know of, who have
> > >>>> little or NO fulltime linux developers and consider the < 5% market
> > >>>> fringe at best.
> > >>>>
> > >>> it helps to virtualize devices and introduces newer features for that.
> > >>> Some interesting projects could be derrived out of that, there are
> > >>> quite a few interesting papers floating around how drivers could be
> > >>> handled in future.
> > >>>
> > >> IOMMU can be considered similar to the AGP GART, which is similar,
> > >> remapping the Addresses, as far as i understand.
> > >
> > > Common new IOMMUs have only very few in common with the AGP GART. In
> > > fact, with current modern IOMMU hardware it will be possible to
> > > implement secure userspace device drivers that are even able to do DMA.
> > > This is not possible with the GART.
> > >
> > >> Though you get a physical to virtual translation, what about interrupts,
> > >
> > > Modern IOMMUs are able to remap interrupts. This will solve the problem
> > > with PCI interrupt sharing.
> >
> > What CPU's are we talking about ?
> >
>
> upcoming x86-64 processors from the AMD side. I'm not into what Intel
> is doing in that area at the moment.
All AMD64 chips have an IOMMU. Only Intel's most recent chips do.
Alex
On 9/14/07, Alex Deucher <[email protected]> wrote:
> On 9/14/07, Markus Rechberger <[email protected]> wrote:
> > On 9/14/07, Manu Abraham <[email protected]> wrote:
> > > Joerg Roedel wrote:
> > > > On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
> > > >>>>> What do you think about IOMMU?
> > > >>>>>
> > > >>>>>
> > > >>>> Just because AMD or INTEL want to invent some whizzy new technology
> it
> > > >>>> doesn't say anything about the TV card development and retail
> business.
> > > >>>> Intel and AMD have teams of Linux engineers helping operating
> system
> > > >>>> developers bring their ideas and technologies to new platforms.
> That's
> > > a
> > > >>>> million miles away from any of the TV board vendors I know of, who
> have
> > > >>>> little or NO fulltime linux developers and consider the < 5% market
> > > >>>> fringe at best.
> > > >>>>
> > > >>> it helps to virtualize devices and introduces newer features for
> that.
> > > >>> Some interesting projects could be derrived out of that, there are
> > > >>> quite a few interesting papers floating around how drivers could be
> > > >>> handled in future.
> > > >>>
> > > >> IOMMU can be considered similar to the AGP GART, which is similar,
> > > >> remapping the Addresses, as far as i understand.
> > > >
> > > > Common new IOMMUs have only very few in common with the AGP GART. In
> > > > fact, with current modern IOMMU hardware it will be possible to
> > > > implement secure userspace device drivers that are even able to do
> DMA.
> > > > This is not possible with the GART.
> > > >
> > > >> Though you get a physical to virtual translation, what about
> interrupts,
> > > >
> > > > Modern IOMMUs are able to remap interrupts. This will solve the
> problem
> > > > with PCI interrupt sharing.
> > >
> > > What CPU's are we talking about ?
> > >
> >
> > upcoming x86-64 processors from the AMD side. I'm not into what Intel
> > is doing in that area at the moment.
>
> All AMD64 chips have an IOMMU. Only Intel's most recent chips do.
>
It's not available yet, you can read more about it in the following article:
http://developer.amd.com/articlex.jsp?id=101
If you're interested in it I can put together some more information about
it.
Markus
On 9/14/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > > - The hybrid tuner support, that where your requirement, when all those
> > > discussions started, were already added to the subsystem. So now, an
> > > hybrid tuner can be accessed by both DVB and V4L devices;
> > >
> >
> > It's far more complex as the thing which is implemented there.
> > The only thing that has been implemented is that one moduleformat
> > can be loaded by the v4l and by the dvb framework nothing else at the
> > moment. I told you at the kernel summit about that and I think you
> > knew about that before.
> >
> > Just to quote some text:
> > "Right now, a separate instance of the driver is used for analog /
> > digital tuning. In order to use a single instance, we will have to
> > store a pointer to the dvb_frontend structure on the bridge level, but
> > there are various other prerequisites that must be dealt with before we
> > get to that point.
> >
> > We _will_ get there though, eventually."
>
> Maybe it is still not perfect. Feel free to improve it. Several people
> from the community, including me, already offered you help to add your
> driver, reworking on the 5% of the stuff that aren't compatible with the
> V4L/DVB core API design.
For clarification, Markus is quoting me, above.
The idea is to eventually store a pointer to the dvb_frontend
structure on the bridge level to be used as a single entry point
between the analog and digital subsystems. However, we are not yet
ready for this, as the refactoring process has a lot more to be done
beforehand.
Phase 1 of the refactoring was to implement the core changes required
for a single module to be used by both v4l and dvb, and to convert the
existing tuner modules to the new internal API.
Phase 2, a work in progress, involves the removal of duplicated code.
For example, the current code in the master branch still has tda8275 /
tda8275a analog code inside of tda8290.c, where the digital tuning
code is in tda827x.c ... This was resolved in this changeset:
Move all tda8275/8275a tuning code from tda8290 module into tda827x module
http://linuxtv.org/hg/~mkrufky/philipsNXP/rev/09c2e16a8cdd
This code is working fine, and I have it pushed to linuxtv.org for the
sake of testing. I have not requested merge to master because I still
have some cleanups to do, and I do not want this to go to 2.6.24.
(side note: basic support for TDA8295 + TDA18271 has been added to my
philipsNXP tree, as well)
Tuner-simple and dvb-pll will have to undergo a similar treatment, and
it's not going to happen overnight. But I *am* working on it.
I've outlined a basic roadmap to the refactoring plans in my original RFC:
http://linuxtv.org/pipermail/linux-dvb/2007-August/019950.html
What I didn't mention in that RFC, however, is the method in which I
plan to remove the multiple instantiations of the tuner code for a
single piece of hardware, by moving the dvb_frontend pointer to the
bridge level. Since this change depends on the other refactoring to
be completed first, I found it unnecessary to explain this in detail
at this point.
When the time comes, a new RFC will be sent out to deal with that matter.
There is no reason why the Xceive driver cannot be merged into the
current development tree using the hybrid tuner framework as it stands
today.
Regards,
Mike
On 9/14/07, Markus Rechberger <[email protected]> wrote:
> On 9/14/07, Alex Deucher <[email protected]> wrote:
> > On 9/14/07, Markus Rechberger <[email protected]> wrote:
> > > On 9/14/07, Manu Abraham <[email protected]> wrote:
> > > > Joerg Roedel wrote:
> > > > > On Fri, Sep 14, 2007 at 11:57:59AM +0400, Manu Abraham wrote:
> > > > >>>>> What do you think about IOMMU?
> > > > >>>>>
> > > > >>>>>
> > > > >>>> Just because AMD or INTEL want to invent some whizzy new technology
> > it
> > > > >>>> doesn't say anything about the TV card development and retail
> > business.
> > > > >>>> Intel and AMD have teams of Linux engineers helping operating
> > system
> > > > >>>> developers bring their ideas and technologies to new platforms.
> > That's
> > > > a
> > > > >>>> million miles away from any of the TV board vendors I know of, who
> > have
> > > > >>>> little or NO fulltime linux developers and consider the < 5% market
> > > > >>>> fringe at best.
> > > > >>>>
> > > > >>> it helps to virtualize devices and introduces newer features for
> > that.
> > > > >>> Some interesting projects could be derrived out of that, there are
> > > > >>> quite a few interesting papers floating around how drivers could be
> > > > >>> handled in future.
> > > > >>>
> > > > >> IOMMU can be considered similar to the AGP GART, which is similar,
> > > > >> remapping the Addresses, as far as i understand.
> > > > >
> > > > > Common new IOMMUs have only very few in common with the AGP GART. In
> > > > > fact, with current modern IOMMU hardware it will be possible to
> > > > > implement secure userspace device drivers that are even able to do
> > DMA.
> > > > > This is not possible with the GART.
> > > > >
> > > > >> Though you get a physical to virtual translation, what about
> > interrupts,
> > > > >
> > > > > Modern IOMMUs are able to remap interrupts. This will solve the
> > problem
> > > > > with PCI interrupt sharing.
> > > >
> > > > What CPU's are we talking about ?
> > > >
> > >
> > > upcoming x86-64 processors from the AMD side. I'm not into what Intel
> > > is doing in that area at the moment.
> >
> > All AMD64 chips have an IOMMU. Only Intel's most recent chips do.
> >
>
> It's not available yet, you can read more about it in the following article:
>
> http://developer.amd.com/articlex.jsp?id=101
>
> If you're interested in it I can put together some more information about
> it.
It (the IOMMU) is available on all AMD64 chips. The newer
virtualization stuff is only on the newer chips.
Alex
On Fri, Sep 14, 2007, Markus Rechberger wrote:
>
> people do contribute to the em28xx project.
...
> there's also an active and even problem solving oriented ML available:
> http://mcentral.de/pipermail/em28xx/
>
> Also if you look at the mercurial code you'll see several people
> contributing to that project.
Of course, people who own such a device and want to use
it with Linux have no choice but to work with you.
And you do a good job for your users, you support them
well and in return they contribute info and patches
to support new devices.
But the thing is that at mcentral.de you're the man at the top,
and your users will hardly disagree with you on core
technical questions. (Well, admittedly I'm speculating
here as I don't read your em28xx list.)
On for drivers/media/ OTOH you are just one developer among others,
and some of them choose to disagree with you. Even worse,
IIRC there wasn't even _a single_ other developer willing
to ACK your offending patch.
Now, doesn't _that_ get you thinking?
Johannes
Hi,
On 9/14/07, Michael Krufky <[email protected]> wrote:
> On 9/14/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > > > - The hybrid tuner support, that where your requirement, when all those
> > > > discussions started, were already added to the subsystem. So now, an
> > > > hybrid tuner can be accessed by both DVB and V4L devices;
> > > >
> > >
> > > It's far more complex as the thing which is implemented there.
> > > The only thing that has been implemented is that one moduleformat
> > > can be loaded by the v4l and by the dvb framework nothing else at the
> > > moment. I told you at the kernel summit about that and I think you
> > > knew about that before.
> > >
> > > Just to quote some text:
> > > "Right now, a separate instance of the driver is used for analog /
> > > digital tuning. In order to use a single instance, we will have to
> > > store a pointer to the dvb_frontend structure on the bridge level, but
> > > there are various other prerequisites that must be dealt with before we
> > > get to that point.
> > >
> > > We _will_ get there though, eventually."
> >
> > Maybe it is still not perfect. Feel free to improve it. Several people
> > from the community, including me, already offered you help to add your
> > driver, reworking on the 5% of the stuff that aren't compatible with the
> > V4L/DVB core API design.
>
> For clarification, Markus is quoting me, above.
>
> The idea is to eventually store a pointer to the dvb_frontend
> structure on the bridge level to be used as a single entry point
> between the analog and digital subsystems. However, we are not yet
> ready for this, as the refactoring process has a lot more to be done
> beforehand.
I think this will require a rethink of either how the em2880-dvb
driver works or how frontend drivers work. The current API expects
users to initialise their frontend and then bind a tuner to it.
em2880-dvb is a sort of subdriver that attaches to the main driver,
and doesn't have any control over when or how it initialises its
tuner, so it can't delay tuner initialisation until the frontend has
been initialised. (I don't think it's the only hybrid driver that
works this way either). Of course, I could be missing something.
> There is no reason why the Xceive driver cannot be merged into the
> current development tree using the hybrid tuner framework as it stands
> today.
I'm not convinced this is entirely true. In order to avoid unnecessary
reinitialisation of the device, the driver needs to know whether the
device is in analog or digital mode, and I can't see a way of doing it
with the current API. (I think existing drivers, such as the xc2028
driver in one branch, use the older analog API and make the digital
driver a wrapper around it.) Again, I may be missing something.
Aidan
> > There is no reason why the Xceive driver cannot be merged into the
> > current development tree using the hybrid tuner framework as it stands
> > today.
>
> I'm not convinced this is entirely true. In order to avoid unnecessary
> reinitialisation of the device, the driver needs to know whether the
> device is in analog or digital mode, and I can't see a way of doing it
> with the current API. (I think existing drivers, such as the xc2028
> driver in one branch, use the older analog API and make the digital
> driver a wrapper around it.) Again, I may be missing something.
For sure there are some ways. One dirty way would be to add an static
lock at xc3028 driver. You can rise the lock when firmware is being
uploaded, removing it at the end. This would prevent the troubles you've
mentioned.
A cleaner way would be to create a tasklet for firmware upload. This
will also prevent the driver for a long load time, due to firmware
initialization (a similar change were recently added at ivtv driver).
For sure there are other ways of doing this.
Cheers,
Mauro
Aidan Thornton wrote:
>
> I think this will require a rethink of either how the em2880-dvb
> driver works or how frontend drivers work. The current API expects
> users to initialise their frontend and then bind a tuner to it.
> em2880-dvb is a sort of subdriver that attaches to the main driver,
> and doesn't have any control over when or how it initialises its
> tuner, so it can't delay tuner initialisation until the frontend has
> been initialised. (I don't think it's the only hybrid driver that
> works this way either). Of course, I could be missing something.
The em28xx/xc3028 is in fact not too complex. Just for sake of
demonstration, some time back i had posted a dummy driver how it can be
done in a nice and clean way as an example.
The patch assumes some additional standards, you can ignore them. But
you get the general idea from in there.
http://marc.info/?l=linux-video&m=117613833119350&w=2
>
>> There is no reason why the Xceive driver cannot be merged into the
>> current development tree using the hybrid tuner framework as it stands
>> today.
>
> I'm not convinced this is entirely true. In order to avoid unnecessary
> reinitialisation of the device, the driver needs to know whether the
> device is in analog or digital mode, and I can't see a way of doing it
> with the current API. (I think existing drivers, such as the xc2028
> driver in one branch, use the older analog API and make the digital
> driver a wrapper around it.) Again, I may be missing something.
You can read this post also for some additional information.
http://marc.info/?l=linux-video&m=117922735929375&w=2
Use the ideas as you deem fit. ;-)
Manu
On 9/14/07, Johannes Stezenbach <[email protected]> wrote:
> On Fri, Sep 14, 2007, Markus Rechberger wrote:
> >
> > people do contribute to the em28xx project.
> ...
> > there's also an active and even problem solving oriented ML available:
> > http://mcentral.de/pipermail/em28xx/
> >
> > Also if you look at the mercurial code you'll see several people
> > contributing to that project.
>
> Of course, people who own such a device and want to use
> it with Linux have no choice but to work with you.
> And you do a good job for your users, you support them
> well and in return they contribute info and patches
> to support new devices.
>
> But the thing is that at mcentral.de you're the man at the top,
> and your users will hardly disagree with you on core
> technical questions. (Well, admittedly I'm speculating
> here as I don't read your em28xx list.)
>
they can. Put technical issues infront of everything also see the
whole picture that userspace tv applications have to support
all the codecs which are available.
> On for drivers/media/ OTOH you are just one developer among others,
> and some of them choose to disagree with you. Even worse,
> IIRC there wasn't even _a single_ other developer willing
> to ACK your offending patch.
>
there was not a single developer of the old crufted developers who
didn't bring the project forward during the last 2 years.
yes you're right.
> Now, doesn't _that_ get you thinking?
>
it gets me thinking. Some core developers who I met during
the last few weeks (kernel summit, suse conference in czech)
told me to go on with it actually because the final plan isn't that
bad.. I don't expect anyone of the old crufted v4l/dvb (well many
of them already left the party) will join the game...
(I'll leave that open here) spend some time read the whole history
of this thread and it will show up what this all is targeting at.
I'll answer your questions with technical reasons why I'm doing
all that stuff that way if you'll just ask.
Someone told me during the last 2 weeks
"but v4l2 was about to solve all those problems" it didn't. (point)
and I think I explained good enough why all that crap still goes
on as it is.
Beside that I'm just curious how much did you contribute
during the last 2 years to the lkml/linux kernel, and how much
do you want to contribute in future? (also from my side
talk is cheap (even for me) but getting something done costs
quite some time and feedback from other people)
Markus
> Beside that I'm just curious how much did you contribute
> during the last 2 years to the lkml/linux kernel, and how much
> do you want to contribute in future? (also from my side
> talk is cheap (even for me) but getting something done costs
> quite some time and feedback from other people)
Contributing doesn't necessarily means submitting patches. There are
several good guys at the community offering very valuable contributions,
like patchset reviews, good comments, userspace experiences, etc.
Johannes does a very impressive work of maintaining, almost alone,
LinuxTV website, upgrading the system, monitoring disk spaces, taking
backups, etc. Also, he is always available to discuss the most important
changes at APIs and to defend the Open Source community, providing his
very clear point of view.
Thank you, Johannes for all your good work!
--
Cheers,
Mauro
On 9/14/07, Johannes Stezenbach <[email protected]> wrote:
> On Fri, Sep 14, 2007, Markus Rechberger wrote:
> >
> > people do contribute to the em28xx project.
> ...
> > there's also an active and even problem solving oriented ML available:
> > http://mcentral.de/pipermail/em28xx/
> >
> > Also if you look at the mercurial code you'll see several people
> > contributing to that project.
>
> Of course, people who own such a device and want to use
> it with Linux have no choice but to work with you.
> And you do a good job for your users, you support them
> well and in return they contribute info and patches
> to support new devices.
>
> But the thing is that at mcentral.de you're the man at the top,
why is it mcentral.de at the moment? I started several discussions about
the whole topic. The thing which is in v4l-dvb on linuxtv does not satisfy
any requirements, also further additions will be hard to achieve with it.
> and your users will hardly disagree with you on core
> technical questions. (Well, admittedly I'm speculating
> here as I don't read your em28xx list.)
>
they can disagree, I even pointed out to design flaws lately when
some users posted bugreports.
> On for drivers/media/ OTOH you are just one developer among others,
> and some of them choose to disagree with you. Even worse,
> IIRC there wasn't even _a single_ other developer willing
> to ACK your offending patch.
>
the current patch isn't offending anymore, see it as a chance I'm
giving you the chance to acknowlidge what has been done now from
my side.
The main discussion in this thread was about drivers in userspace
are bad because the API will allow binary drivers. The guy
who works for Hauppauge (again I also have good contacts
at Hauppauge Europe) writes it's bad - for no technical reason.
If someone points out that it is bad (after reading the whole thread)
why don't we put X.org, bash, well everything into the kernel?
GPL is the saviour seems to be the saviour for some people in this
world, but in the end it's still if people want to go that way.
Much work has been done by other people before, my work
is also just an additional contribution and I (again) don't intend to
release binary drivers. (also keep in mind that ubuntu takes
everything which makes things work - this matters to the enduser)
Hey I can also write I can help you to get things right with some other
people, and I can financially support people by giving away
hardware and even specs for free in some cases. Who is able
to do that from the old crufted v4l/dvb guys?
Manu throws his drivers over the wall to the OSS community, although
I don't mind.
Mauro is copying the logic of my code and writes I told him I'm ok with
taking my code without just adding a single line of how his driver
got developed. (I still wonder why he skipped some significant parts
of the driver .. because he used the existing one as logic template)
http://linuxtv.org/hg/~mchehab/tm6000-new/log/14352ab89146/linux/drivers/media/video/tuner-xc2028.c
(not looking at the specific changeset but he copied the firmware
loading instructions without taking care about the copyright?)
> Now, doesn't _that_ get you thinking?
>
oh yes.
Markus
> The main discussion in this thread was about drivers in userspace
> are bad because the API will allow binary drivers.
No. The focus is that userspace API is not needed at all, and the
community believe that this is a regression from all efforts that are
being done by the community to have good quality OSS software.
> The guy
> who works for Hauppauge (again I also have good contacts
> at Hauppauge Europe) writes it's bad - for no technical reason.
Please stop with personal attacks.
> I (again) don't intend to
> release binary drivers. (also keep in mind that ubuntu takes
> everything which makes things work - this matters to the enduser)
Markus, you are thinking that all the community are fool? You used to
state on your website that your intention were to release binary-only
xc5000 drivers. So, please stop with childish and assume what you've
said.
> Mauro is copying the logic of my code and writes I told him I'm ok with
> taking my code without just adding a single line of how his driver
> got developed. (I still wonder why he skipped some significant parts
> of the driver .. because he used the existing one as logic template)
>
> http://linuxtv.org/hg/~mchehab/tm6000-new/log/14352ab89146/linux/drivers/media/video/tuner-xc2028.c
The basis for tuner-xc2028 driver is tea5767. The xc2028/xc3028 drivers
is very simple:
a) load the proper firmware;
b) send one 32 bits command to select frequency + the frequency divider.
All the rest is the common logic of a tuner driver.
If you take a look at my driver, you will see that the implementation is
different, providing also those functionalities:
- provides a sync during frequency setting, needed by tm6000;
- has the logic to retrieve signal status;
- part of the firmware need to be reload every time you change a freq
(tm6000 driver needs it);
- supports just the firmwares I've identified as being used by tm6000
driver;
The only thing I used is your usbreplay.pl, as properly stated at
README.first (properly pointing to your site).
Again, please stop with personal attacks. This leads to nowhere.
---
>From my side, I've nacked your userspace tuner.
However, I keep open to accept your kernelspace em28xx/xc3028 drivers,
providing that they fits at the current V4L/DVB core.
Changes at common APIs, and especially at Kernel to userspace API should
be discussed with the community. If you accept this fact, you may also
propose improvements at the APIs.
If, after all that were discussed, you're willing to do a serious work,
please send us the patches for em28xx/xc3028 kernelspace drivers.
Otherwise, I'll kindly ask you to take your own way and stop with those
flamewars.
Cheers,
Mauro
Hello Markus,
Markus Rechberger wrote:
> The main discussion in this thread was about drivers in userspace
> are bad because the API will allow binary drivers. The guy
> who works for Hauppauge (again I also have good contacts
> at Hauppauge Europe) writes it's bad - for no technical reason.
AFAICS, Steven raised the same thoughts what i had.
> If someone points out that it is bad (after reading the whole thread)
> why don't we put X.org, bash, well everything into the kernel?
I am not saying that userspace is bad. In fact i am all for userspace,
_if_ there is much of a complication. For example we have had largely
complex devices. You might like to read this thread a while back.
This was the reason why we started up libdvbapi/mti (For those who don't
know what it is, libdvbapi/mti is a userspace approach for having device
support in userspace with complicated tuning algorithms with a lot of
calculations)
http://search.gmane.org/?query=Re%3A+%5BRFC%5D+Userspace+extensions%2C+was+Re%3A+%5Blinux-dvb%5D+%5Bpatch%5D+Add+support+for+different+tuning+algorithms&author=&group=linux.drivers.dvb&sort=relevance&DEFAULTOP=and&query=
For that demodulator and a successor for the same, i had finally moved
the same to in kernel with a lot of trouble. Maybe it is not as precise
as it should have been.
But what i mean is that we should use such approaches if there needs to
be a heavy valid reason and if there are many more devices going that
way, we should definitely move to userspace. Maybe the userspace idea is
a bit still immature.
That said, i don't see any such complexities with the XC3028
Manu
On 9/15/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > The main discussion in this thread was about drivers in userspace
> > are bad because the API will allow binary drivers.
>
> No. The focus is that userspace API is not needed at all, and the
> community believe that this is a regression from all efforts that are
> being done by the community to have good quality OSS software.
>
please have a look at:
http://mcentral.de/wiki/index.php/Bugtracker
> > The guy
> > who works for Hauppauge (again I also have good contacts
> > at Hauppauge Europe) writes it's bad - for no technical reason.
>
> Please stop with personal attacks.
>
This is no personal attack.
> > I (again) don't intend to
> > release binary drivers. (also keep in mind that ubuntu takes
> > everything which makes things work - this matters to the enduser)
>
> Markus, you are thinking that all the community are fool?
You're using the word "community" in a quite abstract way here.
Please document how the linuxtv community behaves behind the
lines, and I would even like that those people who discussed several
things would start to write about their other personal issues with
that project which I'm not part of.
> You used to
> state on your website that your intention were to release binary-only
> xc5000 drivers.
people talk alot, in the end only the result counts. So if you've seen
any binary driver from my side point out to it. To be honest
I think if I would have gone the same way as Avermedia from the
beginning on to release binary drivers it would have saved months
of development and the main distros could already have _full_
support for all the features.
The driver as it is now is not perfect either it requires quite some
more work to get all features work flawlessly around the globe for
all the users.
> So, please stop with childish and assume what you've said.
>
> > Mauro is copying the logic of my code and writes I told him I'm ok with
> > taking my code without just adding a single line of how his driver
> > got developed. (I still wonder why he skipped some significant parts
> > of the driver .. because he used the existing one as logic template)
> >
> >
> http://linuxtv.org/hg/~mchehab/tm6000-new/log/14352ab89146/linux/drivers/media/video/tuner-xc2028.c
>
> The basis for tuner-xc2028 driver is tea5767. The xc2028/xc3028 drivers
> is very simple:
>
> a) load the proper firmware;
> b) send one 32 bits command to select frequency + the frequency divider.
>
well people see if they compare the code what you took from the existing
driver.
> All the rest is the common logic of a tuner driver.
>
> If you take a look at my driver, you will see that the implementation is
> different, providing also those functionalities:
>
> - provides a sync during frequency setting, needed by tm6000;
> - has the logic to retrieve signal status;
> - part of the firmware need to be reload every time you change a freq
> (tm6000 driver needs it);
> - supports just the firmwares I've identified as being used by tm6000
> driver;
>
> The only thing I used is your usbreplay.pl, as properly stated at
> README.first (properly pointing to your site).
>
> Again, please stop with personal attacks. This leads to nowhere.
>
If you look at my current implementation and even the implementation
which I had a year ago many problems are solved there.
I don't mind if you don't want to use the userspace tuner API, it's not
a replacement for the inkernel version, it makes it just easy to
implement it.
The current and upcoming em28xx devices will use it, just to avoid
that I have to redo all the work hundret times for no real reason.
>
> >From my side, I've nacked your userspace tuner.
>
> However, I keep open to accept your kernelspace em28xx/xc3028 drivers,
> providing that they fits at the current V4L/DVB core.
>
I don't even want to use that existing driver in future (as I wrote earlier
already). The newer module will be a dropin userspace replacement for
what's available. I got around 70 devices work with it at the moment,
although I don't even want to reinvent the wheel so I'll take what I've
received from some companies. That way all I have to do is to use
the provided API of those silicon sample drivers.
> Changes at common APIs, and especially at Kernel to userspace API should
> be discussed with the community. If you accept this fact, you may also
> propose improvements at the APIs.
>
That for I wrote the RFCs which didn't get discussed properly and where
2 people of that "community" told me to use something which won't work
out at all.
And hey, why didn't those guys do what they told me to do in the end?
Because they wanted to do it by themself only. Although it's the base
of the driver without a proper base it just screams for more work in the
end.
> If, after all that were discussed, you're willing to do a serious work,
> please send us the patches for em28xx/xc3028 kernelspace drivers.
> Otherwise, I'll kindly ask you to take your own way and stop with those
> flamewars.
>
You don't have to use it if you don't like it, and I get around those
restrictions and I'm able to fix up and extend the features for the
upcoming devices without interfering your work.
Markus
On Sat, Sep 15, 2007, Markus Rechberger wrote:
>
> it gets me thinking. Some core developers who I met during
> the last few weeks (kernel summit, suse conference in czech)
> told me to go on with it actually because the final plan isn't that
> bad..
I was referring to your code you posted for merging on linux-dvb,
and which got rejected. Anyway, it's easy to agree with you if one
has just heard _your_ version -- the purpose of this thread is to
give people an alternate version of the story to complement
yours, which would allow the people you talked to to think it over
and change their mind. See if you can get of those people you
talked to to post here to support you.
> Beside that I'm just curious how much did you contribute
> during the last 2 years to the lkml/linux kernel, and how much
> do you want to contribute in future? (also from my side
> talk is cheap (even for me) but getting something done costs
> quite some time and feedback from other people)
IOW you say if I don't write code I should shut up?
Is that also what you tell users on your em28xx list
when they dare to disagree with you?
Nice...
Johannes
On Sat, Sep 15, 2007, Markus Rechberger wrote:
> The main discussion in this thread was about drivers in userspace
> are bad because the API will allow binary drivers. The guy
> who works for Hauppauge (again I also have good contacts
> at Hauppauge Europe) writes it's bad - for no technical reason.
It sends a shiver down my spine that seem to imply that you
use your Hauppauge Europe contact to lobby against the
efforts of Hauppauge USA employees to promote support for
Open Source kernel drivers within their company.
> If someone points out that it is bad (after reading the whole thread)
> why don't we put X.org, bash, well everything into the kernel?
> GPL is the saviour seems to be the saviour for some people in this
> world, but in the end it's still if people want to go that way.
> Much work has been done by other people before, my work
> is also just an additional contribution and I (again) don't intend to
> release binary drivers. (also keep in mind that ubuntu takes
> everything which makes things work - this matters to the enduser)
>
> Hey I can also write I can help you to get things right with some other
> people, and I can financially support people by giving away
> hardware and even specs for free in some cases. Who is able
> to do that from the old crufted v4l/dvb guys?
>
> Manu throws his drivers over the wall to the OSS community, although
> I don't mind.
> Mauro is copying the logic of my code and writes I told him I'm ok with
> taking my code without just adding a single line of how his driver
> got developed. (I still wonder why he skipped some significant parts
> of the driver .. because he used the existing one as logic template)
>
> http://linuxtv.org/hg/~mchehab/tm6000-new/log/14352ab89146/linux/drivers/media/video/tuner-xc2028.c
> (not looking at the specific changeset but he copied the firmware
> loading instructions without taking care about the copyright?)
After you dismissed all arguments except your own as invalid, you've
now moved on to the "post bullshit and personal accusations" stage.
I hope this means we're now done with this "discussion".
> > Now, doesn't _that_ get you thinking?
>
> oh yes.
Yeah, right.
Johannes
On 9/15/07, Johannes Stezenbach <[email protected]> wrote:
> On Sat, Sep 15, 2007, Markus Rechberger wrote:
> >
> > it gets me thinking. Some core developers who I met during
> > the last few weeks (kernel summit, suse conference in czech)
> > told me to go on with it actually because the final plan isn't that
> > bad..
>
> I was referring to your code you posted for merging on linux-dvb,
> and which got rejected. Anyway, it's easy to agree with you if one
> has just heard _your_ version -- the purpose of this thread is to
> give people an alternate version of the story to complement
> yours, which would allow the people you talked to to think it over
> and change their mind. See if you can get of those people you
> talked to to post here to support you.
>
Everyone knows that there are some issues even some internal
ones which I'm not part of.
Not sure if you met some other kernel developers recently, all you
can hear is "those crazy media guys" (which just includes everyone
who works with media related drivers)
> > Beside that I'm just curious how much did you contribute
> > during the last 2 years to the lkml/linux kernel, and how much
> > do you want to contribute in future? (also from my side
> > talk is cheap (even for me) but getting something done costs
> > quite some time and feedback from other people)
>
> IOW you say if I don't write code I should shut up?
> Is that also what you tell users on your em28xx list
> when they dare to disagree with you?
>
> Nice...
>
This is a free world you can write whatever you want,
but if you want me to change my code you need to
convince me that I should acknowlidge your ideas.
And by not acknowlidging my requirements don't expect
that I will go back to the start and try to reinvent the
wheel. After almost 2 years I'm quite into those things
and I know what I want for my project and what I'll try
to achieve with what I'm doing.
Markus
> Everyone knows that there are some issues even some internal
> ones which I'm not part of.
With respect to your kernel-userspace API for xc3028, you made something
that seemed to be a dream: there's a consensus: not a single developer
believed that this is the better way; nobody seems that this is the
better approach.
So, or you are the only media developer with good sense in the face of
the Earth, or you're incapable of understand the obvious: you're wrong
with this approach. IMO, the answer is quite obvious.
--
Cheers,
Mauro
On 9/15/07, Mauro Carvalho Chehab <[email protected]> wrote:
> > Everyone knows that there are some issues even some internal
> > ones which I'm not part of.
>
> With respect to your kernel-userspace API for xc3028, you made something
> that seemed to be a dream: there's a consensus: not a single developer
> believed that this is the better way; nobody seems that this is the
> better approach.
>
Not a single developer out of 3-5 people? Seems fine with me, because
people didn't agree with anything else before either and tried to point
me to wrong solutions.
http://threebit.net/mail-archive/video4linux/msg07548.html
> So, or you are the only media developer with good sense in the face of
> the Earth, or you're incapable of understand the obvious: you're wrong
> with this approach. IMO, the answer is quite obvious.
>
that for you don't have to use it.
http://mcentral.de/wiki/index.php/Userspace_tuner#Advantages
I'm off for the weekend now so have a nice one :-)
Markus
Em Sáb, 2007-09-15 às 16:33 +0200, Markus Rechberger escreveu:
> I'm off for the weekend now so have a nice one :-)
Enjoy your weekend. I really hope that you have some time to reflect and
review your positions during the weekend.
--
Cheers,
Mauro
Johannes Stezenbach wrote:
> On Sat, Sep 15, 2007, Markus Rechberger wrote:
>
>> The main discussion in this thread was about drivers in userspace
>> are bad because the API will allow binary drivers. The guy
>> who works for Hauppauge (again I also have good contacts
>> at Hauppauge Europe) writes it's bad - for no technical reason.
>>
>
>
lol, oh man, how to make friends and influence people.
> It sends a shiver down my spine that seem to imply that you
> use your Hauppauge Europe contact to lobby against the
> efforts of Hauppauge USA employees to promote support for
> Open Source kernel drivers within their company.
>
>
Ack.
On Sat, Sep 15, 2007 at 11:04:42AM -0300, Mauro Carvalho Chehab wrote:
> With respect to your kernel-userspace API for xc3028, you made something
> that seemed to be a dream: there's a consensus: not a single developer
> believed that this is the better way; nobody seems that this is the
> better approach.
>
> So, or you are the only media developer with good sense in the face of
> the Earth, or you're incapable of understand the obvious: you're wrong
> with this approach. IMO, the answer is quite obvious.
Yes, as a newbie observer on the v4l list, the answer is obvious to me,
at last, and the reason is not entirely technical. I can't read so many
bogus arguments on so few lines without reacting.
Rephrasing Mauro:
"Not a single developer out of a few SEEMS to believe that it is the BETTER
approach, so since the FEW represent ALL media developers in the world, and
since there is only ONE RIGHT way to do things, and since the GROUP is
always RIGHT and always knows better than the individual, then YOU're WRONG
and I'm right. Conform to the group and do as the group says, whatever the
consequences!"
Geeks are decidedly as prone as others to blindly accept travelling on the
slippery road of herd mentality and "obvious" conclusions based on
appearances. Is this OPEN source development or a dictatorship or what?
So in the end Mauro will be right. And Markus will continue to develop and
defend his stuff as HE sees fit. He knows his own work better than anyone
else. It will be HIS way or nothing with his own stuff, it's his inalienable
right.
And don't be naive, if there's no solution more viable than Markus' one,
then the latter will eventually be widely adopted somehow, sometime,
whatever the amount of grumbling from the establishment. No
dictatorship/forced consensus can decide future's direction, nor improve
its already low own viability.
Cheers,
Bernard.
On Saturday 15 September 2007 18:58:34 Bernard Jungen wrote:
> On Sat, Sep 15, 2007 at 11:04:42AM -0300, Mauro Carvalho Chehab wrote:
> > With respect to your kernel-userspace API for xc3028, you made
> > something that seemed to be a dream: there's a consensus: not a
> > single developer believed that this is the better way; nobody seems
> > that this is the better approach.
> >
> > So, or you are the only media developer with good sense in the face
> > of the Earth, or you're incapable of understand the obvious: you're
> > wrong with this approach. IMO, the answer is quite obvious.
>
> Yes, as a newbie observer on the v4l list, the answer is obvious to
> me, at last, and the reason is not entirely technical. I can't read
> so many bogus arguments on so few lines without reacting.
>
> Rephrasing Mauro:
>
> "Not a single developer out of a few SEEMS to believe that it is the
> BETTER approach, so since the FEW represent ALL media developers in
> the world, and since there is only ONE RIGHT way to do things, and
> since the GROUP is always RIGHT and always knows better than the
> individual, then YOU're WRONG and I'm right. Conform to the group and
> do as the group says, whatever the consequences!"
>
> Geeks are decidedly as prone as others to blindly accept travelling
> on the slippery road of herd mentality and "obvious" conclusions
> based on appearances. Is this OPEN source development or a
> dictatorship or what?
It's called peer-review. It's why the linux kernel is as good as it is
today. Yes, the tuner belongs in kernel-space. I'll use the user-space
version from Markus in my ivtv driver as long as there is no
alternative, but as soon as the same tuner is merged in the kernel I'll
switch to that one. Why? Because the end-user shouldn't have to install
a userspace daemon just to support a stupid little tuner.
I've said this before, and I'll say it again: the sole reason for this
mess is personality clashes. Technically it should have gone in two
years ago. I worked two years on getting the ivtv driver (and
supporting i2c drivers) merged into the kernel, in the process of which
many V4L2 (and a few DVB) API additions and refinements were made, all
by working with the core developers. The end result was much better
than if I would have done it all by myself.
It can be a difficult process (it's always hard to accept that the other
person is right), and sometimes it means you have to sleep on it for a
few nights before you realize that you are wrong and the other person
is right. It can also go the other way, but even then it helps you
because it forces you to express the technical reasons why you are
right. And that gives you a better understanding of the issues at
stake.
> So in the end Mauro will be right. And Markus will continue to
> develop and defend his stuff as HE sees fit. He knows his own work
> better than anyone else. It will be HIS way or nothing with his own
> stuff, it's his inalienable right.
You're never alone. You work within the kernel framework and within the
v4l-dvb framework. You want to get something done, then you'll have to
work together. The linux project is no different there then working for
a company. And no, Mauro isn't always right. But just saying 'I'm
right, you're wrong' never helps. Never, ever. Arguing your case based
on technical arguments is the best way to go. Always remain respectful
with whomever you're arguing, always remain polite.
The time for rational arguments in this situation has long since passed,
unfortunately.
> And don't be naive, if there's no solution more viable than Markus'
> one, then the latter will eventually be widely adopted somehow,
> sometime, whatever the amount of grumbling from the establishment. No
> dictatorship/forced consensus can decide future's direction, nor
> improve its already low own viability.
Sigh.
Hans
On 9/15/07, Bernard Jungen <[email protected]> wrote:
> On Sat, Sep 15, 2007 at 11:04:42AM -0300, Mauro Carvalho Chehab wrote:
> > With respect to your kernel-userspace API for xc3028, you made something
> > that seemed to be a dream: there's a consensus: not a single developer
> > believed that this is the better way; nobody seems that this is the
> > better approach.
> >
> > So, or you are the only media developer with good sense in the face of
> > the Earth, or you're incapable of understand the obvious: you're wrong
> > with this approach. IMO, the answer is quite obvious.
>
> Yes, as a newbie observer on the v4l list, the answer is obvious to me,
> at last, and the reason is not entirely technical. I can't read so many
> bogus arguments on so few lines without reacting.
>
> Rephrasing Mauro:
>
> "Not a single developer out of a few SEEMS to believe that it is the BETTER
> approach, so since the FEW represent ALL media developers in the world, and
> since there is only ONE RIGHT way to do things, and since the GROUP is
> always RIGHT and always knows better than the individual, then YOU're WRONG
> and I'm right. Conform to the group and do as the group says, whatever the
> consequences!"
>
> Geeks are decidedly as prone as others to blindly accept travelling on the
> slippery road of herd mentality and "obvious" conclusions based on
> appearances. Is this OPEN source development or a dictatorship or what?
>
> So in the end Mauro will be right. And Markus will continue to develop and
> defend his stuff as HE sees fit. He knows his own work better than anyone
> else. It will be HIS way or nothing with his own stuff, it's his inalienable
> right.
>
> And don't be naive, if there's no solution more viable than Markus' one,
> then the latter will eventually be widely adopted somehow, sometime,
> whatever the amount of grumbling from the establishment. No
> dictatorship/forced consensus can decide future's direction, nor improve
> its already low own viability.
>
I see it exactly the same way.
Well I will continue to provide an alternative stack with alternative drivers.
The peer review shows that it doesn't work too well without people participating
fixing problems, the initial drivers were inkernel based and due some updates in
modules which are used by the em28xx some devices which previously worked
don't work anymore in the kernel.
As for the em28xx the driver will use userspace i2c based tuners, decoders and
demodulators.
The userspace modules will use unaltered sample drivers which are provided by
several companies, which also saves alot time in future for that project.
Those drivers will also officially be provided and recommended by the
manufacturing company of those devices, including the proper firmware.
Personally I won't spend any time on reinventing the wheel and fixing drivers
which get broken by not taking care of it.
Beside that people are welcome to contribute without having to fear a political
overhead of discussing requirements and other issues when changing the
corresponding alternative core code. The project and how it will/should go on
is documented at mcentral.de.
I also see that as a good way for the community, that way the linuxtv guys
have to prove that they can be open to other people without adding too
much noise and overhead otherwise people will contribute to the alternative
project as some already do.
Markus
Markus Rechberger wrote:
> On 9/14/07, Alan Cox <[email protected]> wrote:
>> On Fri, Sep 14, 2007 at 01:17:05AM +0200, Markus Rechberger wrote:
>>> what stops vendors of using the current existing code to achieve that
>>> goal. They could provide binary drivers with the existing API.
>> If you feel lucky about the GPL
>>
>>> What stops companies to intercept the ioctl calls and overriding some
>>> I2C commands?
>> The GPL - derivative work is the boundary not code linkage. Possibly a
>> userspace
>> tuner hack would probably fit this too. Especially if a specific vendor is
>> producing both bits together and trying to claim they are independant.
>>
>>> How about proprietary video formats, would you also place the decoding
>>> algorithms in kernel just to force companies to release their code
>>> for it?
>> No, I would assume they'd provide a proprietary conversion library that
>> no nobody would use (just like their hw). We keep format conversion firmly
>> seperated from hardware I/O processing.
>>
>
> In general there are known formats available, the drawback is that every TV
> application deals with it in a non-unified way at the moment, meaning alot
> code duplication in userspace since there's no library available at the moment
> which does the videoconversion from one to another format. Such a library is
> beeing developed at the moment which gets plugged infront of accessing
> the devicenodes.
IMHO...
The reason why there is no single 'format conversion library' that
everybody uses is because of the large differences between requirements
for such a thing. The line between 'format conversion' and things such
as a video codec, or image processing is very vague. For example: Some
apps just want compressed video format output. Would video compression
be part of such a 'format conversion library'? If so, then which
compressed video formats? Proprietary ones too? If not all formats that
exist are supported, it would not be complete for some or many apps, and
maybe not even useful at all (because integrating any necessary
pixel-format conversion into the compression routines may be more
efficient than making multiple passes over the data with separate
libraries). Will the library include resizing? If so, which resampling
algorithms? Then how about rotation? Then maybe geometrical mapping
(games could want that)? Will the library be able to adjust
brightness/contrast/saturation in software? If so, then what about noise
filtering (which algorithms?), etc... perhaps the library should go all
the way, bind to port 80 and serve-up a live video stream
'youtube-style'? ok, now that would definitely go too far...
The question is: Where exactly to put the boundary?
My point is that format conversion is not a video capture driver issue.
Sure, it is convenient for apps to be able to use standard libraries
that perform certain functions with optimized code, etc, but for the
purpose of capturing video (media) it's not always necessary to convert
the output into something different before the data is useful for an app.
Certain format conversions can be done very efficiently inside video
cards, for example, and an app may prefer to use that. If a video card
has such functionality, access to it should be part of its driver.
Applications needing format conversions would benefit tremendously from
a good, powerful, flexible, efficient, etc, library that removes the
necessity for each application to choose between using a video-card
accelerator, or MMX code, or SSE code, or another method to do the work.
Maybe it should allow applications to request a 'media stream' in a
given format and container, with conversion automagically happening when
needed, as efficiently as possible (or at the requested quality level).
But that doesn't mean that that library should be an integrated part of
the video capture interface. The boundary of the 'media capture
interface' driver should be on the data, as delivered by the video
capture interface.
The reason is that the 'best' way to do format conversions depend on the
application requirements (what formats can the app use directly?), on
the available hardware (is there an accelerator that can help?), and on
the quality expectation or system issues (is the CPU the bottleneck or
is the USB bus saturated? Do we request JPEG from the webcam and
decompress into YUV, or do we request the RAW bayer from the
high-quality image sensor and convert that into YUV?). Choices, choises...
That means that IMHO, to prevent having runaway complexity plus chasing
a moving target, format conversion should definitely not be embedded in
the API to access video capture hardware, unless the hardware itself
offers format conversion functions (for example, as it does now, by
allowing choice of the output format, but only for formats that the
hardware itself supports directly).
> Although in the long run I'm looking forward to reuse the userspace tuners with
> such a library infront of everything by using i2c-dev.
> This would also make it easy to reuse the sample code of several companies,
> without having to cut out several features of the drivers and to rewrite them
> to an inkernel format.
It may make things 'easy' if it makes it easy to use existing (?) code
from other sources, but that's not a reason why such an interface is
technically superior. For example, 'ndiswrapper' makes it very easy to
use certain wireless card drivers, but that doesn't mean (by far) that
ndiswrapper's method results in the best wireless card drivers for Linux.
It sounds to me that your approach is more similar to ndiswrapper, for
practical reasons 'this is how I can make it work with less effort' than
it is similar to a full-blown Linux driver 'this is the best way to make
a reliable and efficient driver for this kind of hardware'. Linux
_users_ may need to use ndiswrapper right now, but Linux itself needs
real drivers, not a compromise.
Just my 2cts...
Jelle.
>
> Markus
>
> --
> video4linux-list mailing list
> Unsubscribe mailto:[email protected]?subject=unsubscribe
> https://www.redhat.com/mailman/listinfo/video4linux-list
>
On 9/18/07, Jelle Foks <[email protected]> wrote:
> Markus Rechberger wrote:
> > On 9/14/07, Alan Cox <[email protected]> wrote:
> >> On Fri, Sep 14, 2007 at 01:17:05AM +0200, Markus Rechberger wrote:
> >>> what stops vendors of using the current existing code to achieve that
> >>> goal. They could provide binary drivers with the existing API.
> >> If you feel lucky about the GPL
> >>
> >>> What stops companies to intercept the ioctl calls and overriding some
> >>> I2C commands?
> >> The GPL - derivative work is the boundary not code linkage. Possibly a
> >> userspace
> >> tuner hack would probably fit this too. Especially if a specific vendor
> is
> >> producing both bits together and trying to claim they are independant.
> >>
> >>> How about proprietary video formats, would you also place the decoding
> >>> algorithms in kernel just to force companies to release their code
> >>> for it?
> >> No, I would assume they'd provide a proprietary conversion library that
> >> no nobody would use (just like their hw). We keep format conversion
> firmly
> >> seperated from hardware I/O processing.
> >>
> >
> > In general there are known formats available, the drawback is that every
> TV
> > application deals with it in a non-unified way at the moment, meaning alot
> > code duplication in userspace since there's no library available at the
> moment
> > which does the videoconversion from one to another format. Such a library
> is
> > beeing developed at the moment which gets plugged infront of accessing
> > the devicenodes.
>
> IMHO...
>
> The reason why there is no single 'format conversion library' that
> everybody uses is because of the large differences between requirements
> for such a thing. The line between 'format conversion' and things such
> as a video codec, or image processing is very vague. For example: Some
> apps just want compressed video format output. Would video compression
> be part of such a 'format conversion library'? If so, then which
> compressed video formats? Proprietary ones too? If not all formats that
> exist are supported, it would not be complete for some or many apps, and
> maybe not even useful at all (because integrating any necessary
> pixel-format conversion into the compression routines may be more
> efficient than making multiple passes over the data with separate
> libraries). Will the library include resizing? If so, which resampling
> algorithms? Then how about rotation? Then maybe geometrical mapping
> (games could want that)? Will the library be able to adjust
> brightness/contrast/saturation in software? If so, then what about noise
> filtering (which algorithms?), etc... perhaps the library should go all
> the way, bind to port 80 and serve-up a live video stream
> 'youtube-style'? ok, now that would definitely go too far...
>
> The question is: Where exactly to put the boundary?
>
good statement, the absolute goal should be that enduser applications can
handle videostreams as easy as possible. This is currently not given.
For example I have devices which work with ekiga, and some devices
which don't work with ekiga.
Why so? Ekiga already uses a rather big library for handling videodevices.
Another point is why doesn't tvtime support digital audio?
People have to run sox in the background to get any sound piped from
the corresponding dsp node to the soundcard, and also this is no general
way. So first people have to look up what the correct audio node of their
TV device is... (there are small scripts available which lets the users select)
But I also think this should happen automatically.
There are many devices out there which don't even include all ioctl calls
which are documented in the specs.
And since sometimes the API needs to be updated a library could handle this
in a better way.
I see the whole v4l-dvb project as half way done at the moment because of
the unflexibility of the whole system.
> My point is that format conversion is not a video capture driver issue.
> Sure, it is convenient for apps to be able to use standard libraries
> that perform certain functions with optimized code, etc, but for the
> purpose of capturing video (media) it's not always necessary to convert
> the output into something different before the data is useful for an app.
>
this is just a thought, I think a pluggable mechanism would be nice for that.
The library could query the device about it's capabilities and if it returns a
non standard videoformat just add a mechanism for loading the corresponding
videocodec if available.
> Certain format conversions can be done very efficiently inside video
> cards, for example, and an app may prefer to use that. If a video card
> has such functionality, access to it should be part of its driver.
>
yes, this would always be the optimal way
> Applications needing format conversions would benefit tremendously from
> a good, powerful, flexible, efficient, etc, library that removes the
> necessity for each application to choose between using a video-card
> accelerator, or MMX code, or SSE code, or another method to do the work.
> Maybe it should allow applications to request a 'media stream' in a
> given format and container, with conversion automagically happening when
> needed, as efficiently as possible (or at the requested quality level).
>
yes I also agree with that.
> But that doesn't mean that that library should be an integrated part of
> the video capture interface. The boundary of the 'media capture
> interface' driver should be on the data, as delivered by the video
> capture interface.
>
I would prefer to abstract the ioctl calls from the media applications
if possible.
Sometimes API changes are unavoidable and even make sense instead of
carrying around old crufted mechanisms, and since there's currently not a
single TV application out there which supports all available devices it would
make more sense to work on a better infrastructure for all that.
> The reason is that the 'best' way to do format conversions depend on the
> application requirements (what formats can the app use directly?), on
> the available hardware (is there an accelerator that can help?), and on
> the quality expectation or system issues (is the CPU the bottleneck or
> is the USB bus saturated? Do we request JPEG from the webcam and
> decompress into YUV, or do we request the RAW bayer from the
> high-quality image sensor and convert that into YUV?). Choices, choises...
>
yes there are many ways. I like the idea of such a library because it could
even allow to integrate complete userspace drivers in certain cases.
I posted a link within that thread that there's a driver which intercepts all
ioctl calls and which is completly done in userspace. By using a library
the LD_PRELOAD hack wouldn't even be needed and that device could be
integrated flawlessly into a video infrastructure.
I know that such a userspace driver would be slower than the inkernel driver
although I think it's up to the developer what he wants to do.
> That means that IMHO, to prevent having runaway complexity plus chasing
> a moving target, format conversion should definitely not be embedded in
> the API to access video capture hardware, unless the hardware itself
> offers format conversion functions (for example, as it does now, by
> allowing choice of the output format, but only for formats that the
> hardware itself supports directly).
>
> > Although in the long run I'm looking forward to reuse the userspace tuners
> with
> > such a library infront of everything by using i2c-dev.
> > This would also make it easy to reuse the sample code of several
> companies,
> > without having to cut out several features of the drivers and to rewrite
> them
> > to an inkernel format.
>
> It may make things 'easy' if it makes it easy to use existing (?) code
> from other sources, but that's not a reason why such an interface is
> technically superior. For example, 'ndiswrapper' makes it very easy to
> use certain wireless card drivers, but that doesn't mean (by far) that
> ndiswrapper's method results in the best wireless card drivers for Linux.
>
> It sounds to me that your approach is more similar to ndiswrapper, for
> practical reasons 'this is how I can make it work with less effort' than
> it is similar to a full-blown Linux driver 'this is the best way to make
> a reliable and efficient driver for this kind of hardware'. Linux
> _users_ may need to use ndiswrapper right now, but Linux itself needs
> real drivers, not a compromise.
>
ndiswrapper is quite different, I still nativly compile the code on Linux and
even the sources are available.
I just tried to integrate a newer kind of module, as for me I don't want to
rewrite the algorithms of drivers which I'm able to reuse in userspace,
the example which I have here right now even uses C++ in userspace
and all I do is to use the API of these sample sources and package everything
into such a userspace library which can directly be used by the driver.
It's not my job to reinvent the work of companies (and I'm not even
interested in that) if they agree with delivering code which can be easily
integrated, as for me it's reaching the goal by the shortest path first
approach.
Although there are 2 projects going on at the moment, my one is capable
of pushing some drivers to userspace, the other one is still in progress and
I don't know when it will be finished (since I don't even work on that)
During that time the driver will internally depend on the kernel to userspace
bridge; As soon as the userspace drivers can be plugged to the media library
infront of the devicenodes I intend to remove the kernel-userspace stub..
but this might take a while if not several months.
Markus
> The reason why there is no single 'format conversion library' that
> everybody uses is because of the large differences between requirements
> for such a thing. The line between 'format conversion' and things such
> as a video codec, or image processing is very vague.
Agreed. What I think it should happen is that the userspace library
should focus at the "weird" codecs. E. g. those which uses some sort of
proprietary format. This way, an userspace app may use the userspace
library as a "fallback method" for unknown FOURCC formats. The result
will be probably far away from an optimal result on some cases (since it
probably mean double buffering), but this will at least allow userspace
apps to work. As performance become an issue, the userspace app
developer may use the GPL code at userspace API as a reference to write
a proper optimized format driver for its apps.
Just my 2 cents.
Cheers,
Mauro
On Tue, Sep 18, 2007 at 07:56:05PM -0300, Mauro Carvalho Chehab wrote:
> proprietary format. This way, an userspace app may use the userspace
> library as a "fallback method" for unknown FOURCC formats. The result
> will be probably far away from an optimal result on some cases (since it
> probably mean double buffering), but this will at least allow userspace
> apps to work. As performance become an issue, the userspace app
> developer may use the GPL code at userspace API as a reference to write
> a proper optimized format driver for its apps.
You can dynamically load libraries based on constructed path names which
means you can write a simple library for media conversions which in turn
will try and open libv4l-format-ABCD.so for any format it doesn't know - and
thus is extensible
