List: linux-video
Subject: Question about MPEG playback on HW
From: "Alex Bennee" <trent_took () hotmail ! com>
Date: 2000-02-21 8:32:16
Hi,
This is a modified version of a posting I made to the linux kernel list that
got no response (probably because I was way off-topic). Anyway..
Given the lack of response from the kernel list I am hoping that the
Video4Linux is the correct forum. I have been unable to find a FAQ for
Video4Linux and a brief look at the API seems to show the system is more
interested in video capture than playback. Is this the case?
If I am in the right area please read on…
I've just recently acquired a Videologic mpeg decoder card in an effort to
get decent playback of DVD's. It is based on the Sigma Designs EM8300
decoder chip which according to their web site
(http://www.sigmadesigns.com/product_em8300.htm) handles the MPEG and Audio
decoding as well as the CSS algorithm. The board certainly looks simple
enough as it only has the one major piece of silicon.
Has anyone approached Sigma with any success about obtaining the specs?
My background is that of an embedded software engineer of around 6 year
experience so I am familiar with the whole process of drivers. I would like
to help in writing a driver for this card, however as a comparative newbie
to the whole x86/Linux thing it would probably be better to work with
someone more experienced.
Of course if anyone is already writing such drivers then please point me in
the right direction and I shall offer my services directly!
Regards,
Alex
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: sticke_m <sticke_m () informatik ! fh-hamburg ! de>
Date: 2000-02-21 19:17:56
Hi,
generaly NO, but ......
On Mon, 21 Feb 2000, Alex Bennee wrote:
> I've just recently acquired a Videologic mpeg decoder card in an effort to
> get decent playback of DVD's. It is based on the Sigma Designs EM8300
> decoder chip which according to their web site
> (http://www.sigmadesigns.com/product_em8300.htm) handles the MPEG and Audio
> decoding as well as the CSS algorithm. The board certainly looks simple
> enough as it only has the one major piece of silicon.
>
> Has anyone approached Sigma with any success about obtaining the specs?
Sigma is not willing (now) to give specs about the EM8300 away.
They say it is because the CSS is done in software by the Windows driver
and that they can not make it open source bacause of the DVD restrictions.
There are some discusions on the Sigma Designs news server:
news.realmagic.com or news.sigmadesigns.com (I am not shure)
They have set up a new newsgroup fealing with linux support for the new
MPEG/DVD playback card that will use the EM8400 chip that implements the
CSS algorithm in hardware and that they are writing a driver that will
be open-sourced at the end of the month. they say that the drivers are
beta an there are no nice guy tools blablabla....
It seems that they don't know us and that we don't need such GUI tools
and that we are able to write it on our own and that we only need the
drivers.
Many people want Sigma to open the specs for the other cards (also that
based on the EM8300) and that they don't need to have DVD support and
only need MPEG playback, but Sigma is not willing to do that.
But I think that someone,sometime will get the specs and will write
drivers because the comunity will not buy the new card if Sigma does
not open the specs for the others.
> My background is that of an embedded software engineer of around 6 year
> experience so I am familiar with the whole process of drivers. I would like
> to help in writing a driver for this card, however as a comparative newbie
> to the whole x86/Linux thing it would probably be better to work with
> someone more experienced.
>
> Of course if anyone is already writing such drivers then please point me in
> the right direction and I shall offer my services directly!
BTW: does anyone has a driver for the Aztech Multimedia MPEG Playback Card
It is an ISA based card and I have such a thing. It works fine under
Windows but I wan't Linux support for it. I have written the guys at
Aztech but got no answer.
Michael
---------------------------------------------------------------
Michael Stickel
michael@cubic.org
sticke_m@informatik.fh-hamburg.de
http://www.informatik.fh-hamburg.de/~sticke_m/
---------------------------------------------------------------
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: "Alex Bennee" <trent_took () hotmail ! com>
Date: 2000-02-22 5:15:14
>From: sticke_m <sticke_m@informatik.fh-hamburg.de>
>Hi,
>
>generaly NO, but ......
In that case any idea where I am best focusing my equiries?
>On Mon, 21 Feb 2000, Alex Bennee wrote:
> > I've just recently acquired a Videologic mpeg decoder card in an effort
>to
> > get decent playback of DVD's. It is based on the Sigma Designs EM8300
> > decoder chip
> > <snip>
>Sigma is not willing (now) to give specs about the EM8300 away.
>They say it is because the CSS is done in software by the Windows driver
>and that they can not make it open source bacause of the DVD restrictions.
That would imply that is was imposible to use the chip to play normal MPEG
streams (which AFAIK is all you get after to decrypt with the CSS
algorithm). Have you been involved with discussions with them or is there
someone else I should contact?
>There are some discusions on the Sigma Designs news server:
> news.realmagic.com or news.sigmadesigns.com (I am not shure)
Cheers, I shall try and access the server. Anyone know of a good WWW<->NNTP
gateway service?
>They have set up a new newsgroup fealing with linux support for the new
>MPEG/DVD playback card that will use the EM8400 chip that implements the
>CSS algorithm in hardware and that they are writing a driver that will
>be open-sourced at the end of the month. they say that the drivers are
>beta an there are no nice guy tools blablabla....
>It seems that they don't know us and that we don't need such GUI tools
>and that we are able to write it on our own and that we only need the
>drivers.
Maybe it will be possible to reverse engineer the EM83000 from the 8400
source code. Was this an anouncement they made on the news groups?
>Many people want Sigma to open the specs for the other cards (also that
>based on the EM8300) and that they don't need to have DVD support and
>only need MPEG playback, but Sigma is not willing to do that.
Seems illogicical...
>But I think that someone,sometime will get the specs and will write
>drivers because the comunity will not buy the new card if Sigma does
>not open the specs for the others.
One can hope. I think I should try an send a mail to them...
Alex.
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: pvolcko () concentric ! net
Date: 2000-02-22 14:31:26
> >Sigma is not willing (now) to give specs about the EM8300 away.
> >They say it is because the CSS is done in software by the Windows driver
> >and that they can not make it open source bacause of the DVD restrictions.
>
> That would imply that is was imposible to use the chip to play normal MPEG
> streams (which AFAIK is all you get after to decrypt with the CSS
> algorithm). Have you been involved with discussions with them or is there
> someone else I should contact?
Several people (myself included) have been in touch with Sigma Designs on a
regular basis over the past several months. In every instance the answer to
*any* driver or developer support to the EM8300/HW+ cards has been no. Many
variations on full DVD support have been asked about, including just MPEG-2
decoding and/or ac-3/spdif output access. Again no. It seems that they are
taking the philosophy, and it is a valid one, that they want to get their new
card supported. Kind of use that as their first shot at linux
compatibility. It is a much better suited card for such a task and it is what
their engineering staff have been working on for some time now. It would
appear that our best shot at getting EM8300 support would come from a positive
response from the market and development community in regards to the new
EM8400 chipset and subsquent card(s) (NetStream2000). If they get that, then
they might be swayed. Even then, however, they will likely not give full
playback support for the card, it would be a few scattered bits (however
useful in their own right). So basically, go ahead and ask them if you
want. Be civil and explain your reasons. They will come back with a "no,
sorry" but you'r voice would have been heard with those of many others,
perhaps increasing management's positive view of the opporunities they are
passing up by leaving their older hardware to "rot" away.
> Maybe it will be possible to reverse engineer the EM83000 from the 8400
> source code. Was this an anouncement they made on the news groups?
The EM8400 drivers will be, at least initially, very "thin." This is to say
that they will actually be anorexic by traditional linux driver
standards. There are actually two pieces that Sigma will be releasing. The
user space DVD Navigation Module (closed source, shared lib with published
api) and the kernel module (one IOCTL for transferring data and commands with
the card and user space, as well as some DMA and memory management
stuff). From what I gathered talking to the author of the linux port at LWE,
there is practically nothing in the kernel driver. All logic for controlling
the card is actually in the user space dvd navigation lib. This utilizes a
very simple/thin kernel space interface to the card. This is at least what we
will initially be seeing from them. I suspect that two things will happen in
short order after the release of the beta drivers:
1) Someone will modify the kernel module to trap the commands being sent to
the card. That information will be used to re-write the kernel module to fit
a V4L/V4L2 interface and match a more traditional device kernel module,
providing read, write, open, close, etc. The IOCTL and logic from the first
driver will of course need to remain since that is what the nav api will
depend on to function.
2) Sigma will re-think their approach and either use the hacked driver as
their new model for continued development. This is especially likely to
happen since the reps from Sigma talked to Alan Cox and indicated that they
would like to eventually see the kernel driver make it into the base kernel
source tree. In order for that to happen, the more open and more robust
version of the driver will be necessary.
> >Many people want Sigma to open the specs for the other cards (also that
> >based on the EM8300) and that they don't need to have DVD support and
> >only need MPEG playback, but Sigma is not willing to do that.
>
> Seems illogicical...
They have contracts and licensing restriction they must follow. Trying to
sneak around these, although legal perhaps, would jepordize their future
business with the licensors and could end up backfiring on them with lawsuits
and whatnot. It is not an illogical stance to be taking once you consider
that aspect. The other part to this is that they have a limited support
structure and it is mostly geared toward corporate clientel. They could not
effectively support the developers on both EM8300 and EM8400. We shoudl be
thankful that they are putting the resources toward the EM8400 that will be
necessary.
Yes it's illogical from a purely technical standpoint. Consider the business
and legal implications and it dosen't seem to be that illogical at all.
> >But I think that someone,sometime will get the specs and will write
> >drivers because the comunity will not buy the new card if Sigma does
> >not open the specs for the others.
>
> One can hope. I think I should try an send a mail to them...
I encourage you to. Present a strong argument from not only a technical and
developer standpoint, but a business practice standpoint and it will recieve
attention.
Paul Volcko
LSDVD
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Ralph Metzler <rjkm () netcologne ! de>
Date: 2000-02-22 15:09:32
Alex Bennee writes:
> >Sigma is not willing (now) to give specs about the EM8300 away.
> >They say it is because the CSS is done in software by the Windows dri=
ver
> >and that they can not make it open source bacause of the DVD restrict=
ions.
> =
> That would imply that is was imposible to use the chip to play normal =
MPEG =
> streams (which AFAIK is all you get after to decrypt with the CSS =
> algorithm). Have you been involved with discussions with them or is th=
ere =
> someone else I should contact?
> =
> >There are some discusions on the Sigma Designs news server:
> > news.realmagic.com or news.sigmadesigns.com (I am not shure)
> =
> Cheers, I shall try and access the server. Anyone know of a good WWW<-=
>NNTP =
> gateway service?
> =
> >They have set up a new newsgroup fealing with linux support for the n=
ew
> >MPEG/DVD playback card that will use the EM8400 chip that implements =
the
> >CSS algorithm in hardware and that they are writing a driver that wil=
l
> >be open-sourced at the end of the month. they say that the drivers ar=
e
> >beta an there are no nice guy tools blablabla....
> >It seems that they don't know us and that we don't need such GUI tool=
s
> >and that we are able to write it on our own and that we only need the=
> >drivers.
> =
> Maybe it will be possible to reverse engineer the EM83000 from the 840=
0 =
> source code. Was this an anouncement they made on the news groups?
> =
> >Many people want Sigma to open the specs for the other cards (also th=
at
> >based on the EM8300) and that they don't need to have DVD support and=
> >only need MPEG playback, but Sigma is not willing to do that.
> =
> Seems illogicical...
In one of the postings on news.sigmadesigns.com they (I think it was
Mr. Goldberg) say that in the current situation where CSS is available =
in source form it would already be a violation of their contract, NDA, =
whatever, to provide normal MPEG2 playback.
I suppose their NDA does not cover this specific case but that =
SigmaDesigns wants to be on the safe side.
I don=B4t like this but can understand their situation. When
the big companies and the movie industry control all the important
new laws concerning information technology in a country
(do they still call it the land of the free?) I would be scared too.
> >But I think that someone,sometime will get the specs and will write
> >drivers because the comunity will not buy the new card if Sigma does
> >not open the specs for the others.
> =
> One can hope. I think I should try an send a mail to them...
I have a Hollywood+ and would be willing to help any reverse
engineering effort as my time permits. =
I suppose reverse engineering for hardware compatibility is still
legal, even in the USA with DMCA? =
Lets face it, there still is not much support for Linux and open
source development of drivers for MPEG2/DVD/DVB cards. =
I only know of the drivers for the Siemens DVB card, the Stradis MPEG2
decoder card, the Matrox add-on cards (how open is the support for that?)=
and the
driver for the DVD card with LSI chip (which might still see the day
of light as a PCMCIA driver) which became open source drivers with suppor=
t by the
manufacturer with source code or at least by providing data sheets.
Other drivers like the DXR2 driver and the driver for the Skyvision DVB/D=
VD card
were only possible through reverse enigineering. The EMU8400 driver
will probably not be open sourced, only a thin layer.
Ralph
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Emmanuel Michon <emmanuel_michon () sdesigns ! com>
Date: 2000-02-22 16:35:11
pvolcko@concentric.net writes:
> regular basis over the past several months. In every instance the answer to
> *any* driver or developer support to the EM8300/HW+ cards has been no. Many
You're right.
> The EM8400 drivers will be, at least initially, very "thin." This is to say
> that they will actually be anorexic by traditional linux driver
> standards. There are actually two pieces that Sigma will be releasing. The
> user space DVD Navigation Module (closed source, shared lib with published
> api) and the kernel module (one IOCTL for transferring data and commands with
> the card and user space, as well as some DMA and memory management
> stuff). From what I gathered talking to the author of the linux port at LWE,
> there is practically nothing in the kernel driver. All logic for controlling
> the card is actually in the user space dvd navigation lib. This utilizes a
> very simple/thin kernel space interface to the card. This is at least what we
> will initially be seeing from them. I suspect that two things will happen in
> short order after the release of the beta drivers:
>
> 1) Someone will modify the kernel module to trap the commands being sent to
> the card. That information will be used to re-write the kernel module to fit
> a V4L/V4L2 interface and match a more traditional device kernel module,
> providing read, write, open, close, etc. The IOCTL and logic from the first
> driver will of course need to remain since that is what the nav api will
> depend on to function.
You're talking about some kind of block device for the EM8400 decoder.
Yes, it's fun to make a
# cat toto.mpg >/dev/realmagic0
and see video, but we provide (and need to implement) much more features than
open, write and close that do not really fit a block device driver.
Well, there is a whole family of user-space drivers people often forget, the
X servers. Could you write an X server as a block or char device kernel module...
I do not think our decoder fits your approach.
I did not want to make a big hack like mmap /dev/mem as X servers do
# cat /proc/277/maps
..
400e4000-400f4000 rw-s e1800000 03:02 129537 /dev/mem
400f4000-40104000 rw-s 000a0000 03:02 129537 /dev/mem
40104000-41104000 rw-s e3000000 03:02 129537 /dev/mem
..
because I do not want a setuid root binary. I need some interrupt handling
also, which requires a kernel module for the request_irq() call.
Moreover, people working on software decoding do not work in kernel space,
do they?
I imagine people will try, for sure, a lot of bizarre things with that
driver. My opinion about Eng the 8300 from ReEng of the 8400 is just: good luck.
There might be some guy skilled enough to do that, and I would be really
impressed by this exploit.
> 2) Sigma will re-think their approach and either use the hacked driver as
> their new model for continued development. This is especially likely to
> happen since the reps from Sigma talked to Alan Cox and indicated that they
> would like to eventually see the kernel driver make it into the base kernel
> source tree. In order for that to happen, the more open and more robust
> version of the driver will be necessary.
As a counterpart, you seem to be convinced that we cannot provide
something suitable. Paul, we are not evil.
> They have contracts and licensing restriction they must follow. Trying to
> sneak around these, although legal perhaps, would jepordize their future
> business with the licensors and could end up backfiring on them with lawsuits
> and whatnot. It is not an illogical stance to be taking once you consider
> that aspect.
Great! right and really well explained in a few lines, thank you.
Providing a file player only for H+ *is* sneaking around, as we know
that for sure the first thing people will do is piping css-cat into that.
As a *company* we must abide very strict licenses.
This situation is really different from the individual programmer's.
--
Emmanuel Michon
Ingénieur en développement logiciel
REALmagic France
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: pvolcko () concentric ! net
Date: 2000-02-22 22:40:46
> > 2) Sigma will re-think their approach and either use the hacked driver as
> > their new model for continued development. This is especially likely to
> > happen since the reps from Sigma talked to Alan Cox and indicated that they
> > would like to eventually see the kernel driver make it into the base kernel
> > source tree. In order for that to happen, the more open and more robust
> > version of the driver will be necessary.
>
> As a counterpart, you seem to be convinced that we cannot provide
> something suitable. Paul, we are not evil.
hehe... I'm not trying to imply tht you are. I'm just trying to explain
thatin order for the linux driver to become part of the standard driver set in
the kernel, there would need to be those read, write, etc interfaces. While
there is merit to the user space driver solution, as you explained, there is
also merit to having those "standard" interfaces included in the driver.
It boils down to the fact that the card is not just a dvd decoder. In order
to be a dvd decoder it must also be an mpeg-2 decoder and an ac-3 (and
pcm) decoder. The goal of a kernel module for the card should then be to
first provide necessary support for the dvd navigation implementation (user
space driver). The next step is to provide easy to use low level support for
the playback of files via redirection and standard file operators. Chances
are that this is already done in some form from within the user space driver
set, so there is overlap in functionality. So either the new functionality is
utilized as best as possible within the user space driver, or you leave the
overlap as is.
> > They have contracts and licensing restriction they must follow. Trying to
> > sneak around these, although legal perhaps, would jepordize their future
> > business with the licensors and could end up backfiring on them with lawsuits
> > and whatnot. It is not an illogical stance to be taking once you consider
> > that aspect.
>
> Great! right and really well explained in a few lines, thank you.
> Providing a file player only for H+ *is* sneaking around, as we know
> that for sure the first thing people will do is piping css-cat into that.
>
> As a *company* we must abide very strict licenses.
> This situation is really different from the individual programmer's.
This is what I'm not clear on. How is it your (Sigma's) problem if someone
uses css-cat to the board? As far as you, as a company, would be concerned
there is an unencrypted vob file being played. Those exist and they are
supported otherwise within the user space driver... How does moving that
capability to a kernel module driver change the legality for Sigma?
I reiterate I do not think Sigma is evil. :) I actually like Sigma and
applaud their work in getting linux support in place.
Paul Volcko
LSDVD
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Michael Stickel <sticke_m () informatik ! fh-hamburg ! de>
Date: 2000-02-22 23:39:14
pvolcko@concentric.net wrote:
> > > 2) Sigma will re-think their approach and either use the hacked driver as
> > > their new model for continued development. This is especially likely to
> > > happen since the reps from Sigma talked to Alan Cox and indicated that they
> > > would like to eventually see the kernel driver make it into the base kernel
> > > source tree. In order for that to happen, the more open and more robust
> > > version of the driver will be necessary.
> >
> > As a counterpart, you seem to be convinced that we cannot provide
> > something suitable. Paul, we are not evil.
>
> hehe... I'm not trying to imply tht you are. I'm just trying to explain
> thatin order for the linux driver to become part of the standard driver set in
> the kernel, there would need to be those read, write, etc interfaces. While
> there is merit to the user space driver solution, as you explained, there is
> also merit to having those "standard" interfaces included in the driver.
>
Yes, if Sigma thinks that they will get into the Linux kernel with that solution,
then I must say: never.
Linux has one major difference to Win$. Money doesn't matter as much as
quality of code. If a co programmer of mine would create such a construction
I would let em eat its source page per page. The more code.... the more to eat.
Michael
michael@cubic.org
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: "Alex Bennee" < trent_took () hotmail ! com>
Date: 2000-02-23 0:28:42
>From: Emmanuel Michon < emmanuel_michon@sdesigns.com>
>I imagine people will try, for sure, a lot of bizarre things with that
>driver. My opinion about Eng the 8300 from ReEng of the 8400 is just: good
>luck.
>There might be some guy skilled enough to do that, and I would be really
>impressed by this exploit.
You obviously know more than i do :-) see my other post about my experience
with other evolving chipsets. Of course if the EM8400 is a complete
re-design rather than just putting CSS on an existing chipset then it would
be hard. I'll happily bow to informed opinion.
>Great! right and really well explained in a few lines, thank you.
>Providing a file player only for H+ *is* sneaking around, as we know
>that for sure the first thing people will do is piping css-cat into that.
Sure, but you could make that sort of argument about a Car that could go
faster than the speed limit. However you lawyers may not agree :-)
>As a *company* we must abide very strict licenses.
>This situation is really different from the individual programmer's.
I understand.
Alex.
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Alan Cox <alan () redhat ! com>
Date: 2000-02-23 0:31:20
> Yes, if Sigma thinks that they will get into the Linux kernel with that solution,
> then I must say: never.
But then you don't speak for the kernel people.
The sigma stuff isnt a kernel candidate because the system as a whole is
not open source. Maybe in the future when the DVD forum and MPAA have been
destroyed by other forces. (and mp3.com is only the beginning)
Drivers that intelligently extract services into user space are a good thing.
Eg the draft 3d audio API proposals are putting most of the 3d compute stuff
in user space. Xvideo puts some of the video interfaces in X11 avoiding the
kernel and expensive kernel/X interlocks.
This can be good
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: "Alex Bennee" <trent_took () hotmail ! com>
Date: 2000-02-23 0:35:54
>From: Alan Cox <alan@redhat.com>
>Drivers that intelligently extract services into user space are a good
>thing.
>Eg the draft 3d audio API proposals are putting most of the 3d compute
>stuff
>in user space. Xvideo puts some of the video interfaces in X11 avoiding the
>kernel and expensive kernel/X interlocks.
But from my limited understanding of the way drivers are handled I thought
that having a standard interface (char or block) was part of the power. It
would be no use having a driver from sigma that gave a userspace binary
access to the cards HW registers. I can understand wanting to put all the
DVD navigation, stream selection etc.. stuff into user space but shouldn't a
driver allow you to do a cat x.mpg >/dev/mpeg from the command line because
it has a standard interface?
Alex.
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: sticke_m <sticke_m () informatik ! fh-hamburg ! de>
Date: 2000-02-23 12:04:38
On Tue, 22 Feb 2000, Alan Cox wrote:
> > Yes, if Sigma thinks that they will get into the Linux kernel with that solution,
> > then I must say: never.
>
> But then you don't speak for the kernel people.
Ok, I am by definition not a kernel hacker.
> The sigma stuff isnt a kernel candidate because the system as a whole is
> not open source. Maybe in the future when the DVD forum and MPAA have been
> destroyed by other forces. (and mp3.com is only the beginning)
Do you think they will be destroyed. I don't think so. Hollywood is
much too big and there is much to much money that will be made there.
Hey Alan nice that I reach you. I have send you some mails (dmx4linux)
that all seem to be ignored.
> Drivers that intelligently extract services into user space are a good thing.
> Eg the draft 3d audio API proposals are putting most of the 3d compute stuff
> in user space. Xvideo puts some of the video interfaces in X11 avoiding the
> kernel and expensive kernel/X interlocks.
Yes, but that kind of extraction is not yet supported in a structued union
fashion like it is done by QNX. The drivers should be accessed thru
the standard system calls even if the work is redirected to user space,
because in the other way you have lots of libraries that must be linked
to all applications that uses the driver. Otherwise you must have a
very very very very good library structure or you get a DLL salad like
under Win$.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Alan Cox <alan () redhat ! com>
Date: 2000-02-23 14:38:42
> > not open source. Maybe in the future when the DVD forum and MPAA have been
> > destroyed by other forces. (and mp3.com is only the beginning)
> Do you think they will be destroyed. I don't think so. Hollywood is
> much too big and there is much to much money that will be made there.
There is an astronomical amount of money to be made by whoever breaks their
little setup too. Ultimately they are fighting against the reality of the
internet era.
> Hey Alan nice that I reach you. I have send you some mails (dmx4linux)
> that all seem to be ignored.
[blank look] I dont recall seeing them - remind me
> because in the other way you have lots of libraries that must be linked
> to all applications that uses the driver. Otherwise you must have a
Why waste 1000 clocks on bouncing a request via kernel space ?
> very very very very good library structure or you get a DLL salad like
> under Win$.
DLL salad or kernel salad whats the difference.
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Alan Cox <alan () redhat ! com>
Date: 2000-02-23 14:41:23
> driver allow you to do a cat x.mpg >/dev/mpeg from the command line because
> it has a standard interface?
The reality of the situation is that the hardware doesnt have a standard enough
interface to abstract.
Alan
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.
List: linux-video
Subject: Re: Question about MPEG playback on HW
From: Alan Cox <alan () redhat ! com>
Date: 2000-02-23 14:44:05
> Sure, but you could make that sort of argument about a Car that could go
> faster than the speed limit. However you lawyers may not agree :-)
It has nothing to do with lawyers. US legal precedent is quite clear,
selling something which can be used to commit an offence is not an offence
if it has sensible non-offending uses. The betamax case made that point
quite clear in the supreme court decision
The law doesnt matter when a cartel can simply decide you wont be producing
any more DVD products because they say so.
Alan
--
To unsubscribe: mail video4linux-list-request@redhat.com with
"unsubscribe" as the Subject.