Does anyone know if the D-Link DSL-100D PCI card will work in linux?
URL : http://www.dlink.com.au/products/broadband/dsl100/
Regards
--
Andrew Hatfield
SecureONE - http://www.secureone.com.au/
President - South East Brisbane Linux Users Group http://www.seblug.org/
Kernel work available at http://development.secureone.com.au/kernel/
Clarification,
I can find drivers for this card at
http://www.dlink.com.au/tech/drivers/files/modems/dsl100d.htm
but the note is for PPPoE
The service we use implements PPPoA. Is all I need to do is enable PPPoA in
the networking section?
--
Andrew Hatfield
SecureONE - http://www.secureone.com.au/
President - South East Brisbane Linux Users Group http://www.seblug.org/
Kernel work available at http://development.secureone.com.au/kernel/
----- Original Message -----
From: "Andrew Hatfield" <[email protected]>
To: "Linux Kernel" <[email protected]>
Sent: Wednesday, February 20, 2002 10:36 AM
Subject: Dlink DSL PCI Card
> Does anyone know if the D-Link DSL-100D PCI card will work in linux?
>
> URL : http://www.dlink.com.au/products/broadband/dsl100/
>
> Regards
>
> --
>
> Andrew Hatfield
> SecureONE - http://www.secureone.com.au/
> President - South East Brisbane Linux Users Group
http://www.seblug.org/
>
> Kernel work available at http://development.secureone.com.au/kernel/
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
Having observed Linux driver development on Efficient
3060 and Efficient 3061 DSL PCI products, I've also
managed to track other competitor's Linux DSL
developments as well. Now that I'm out of DSL
industry forever... I can speak my opinion on this
Linux DSL driver quagmire.
The basic problem of DSL (and Winmodem) PCI adapter
and getting to produce open source drivers is:
Manufacturer Non-Disclosure Agreement (aka N.D.A. or
NDA).
NDA basically prevents its resultant driver code from
becoming open-source (GNU, GPL, LGPL, much less public
domain).
I've struggled with various industry chipset
manufacturers in getting them to release us from such
NDAs for three years, tactfully0. It is not only
their marketing decision, but a legal one as well
(i.e., how to undo a group of already-signed NDAs).
The only clean break is to manufacture a DSL chipset
and distribute them without having a single NDA signed
(this prevents unlevel competitions and future
lawsuits from other NDA signers).
The closest I've seen to getting the DSL chipset to be
un-NDA'd is Alcatel Microelectronics. But Alcatel
legal team shut them down to protect Alcatel
Networking division (Sting Ray DSL modem) from losing
future sales (an oxymoron, if I've seen any).
So, to summarized it best. Don't buy PCI adapters
that have microcodes loaded to them NOR require
proprietary microcodes accessed to memory by DMA. In
other word, don't buy winmodem nor DSL PCI adapters,
until those chipset manufacturers publish those
datasheets.
It staggers my mind that chipset marketing groups are
missing a huge revenues streams by having those
product supported on multiple O/S platforms. I've
already computed (many times) the cost of development
as trivial to the sales of such products (even
low-margin PCI adapters). It is more than
self-sustaining. NDA is the primary blocker here.
Safest bet is to buy external DSL modem with Ethernet
(or ATM-25) interface(s).
Direct further threads to the ummm, oh... there isn't
a linux-driver thread...
http://vger.kernel.org/vger-lists.html
Steve Egbert
mailto:[email protected]
BTW: ENI 3060/3061 driver is in binary-only driver
and built only to work with Linux-2.3.99-something (snicker).
__________________________________________________
Do You Yahoo!?
Yahoo! Sports - Coverage of the 2002 Olympic Games
http://sports.yahoo.com
> that have microcodes loaded to them NOR require
> proprietary microcodes accessed to memory by DMA. In
> other word, don't buy winmodem nor DSL PCI adapters,
> until those chipset manufacturers publish those
> datasheets.
The same reasoning goes for another reason. Some of the newest DSL PCI cards are
in many respects winmodems at multimegabit speed levels, burning huge chunks
of CPU on a pentium III processor even.
Alan
> The same reasoning goes for another reason. Some of
> the newest DSL PCI cards are in many respects
> winmodems at multimegabit speed levels, burning
> huge chunks of CPU on a pentium III processor even.
3061 is a true winmodem whose PCI activity has been
unofficially reportedly known to have a burst-rate of
a staggering 23Mbps. (33-23=) 11 Mhz PCI bus anyone?
On other hand, Efficient SpeedStream 3060 is a
bon-fide true DSL PCI adapter card with
downstream-only PCI accesses where its microcode is
loaded into its adapter, but again, its microcode is
NDA'd and more expensive than 3061.
Neither (and all other DSL and Winmodem adapter cards)
are going to make a viable candidate for (if any)
Linux-adapter-of-the-year award EVER ... until we
successfully and collectively apply Linux-PR
guidelines of tactful campaigning the chipset
manufacturers into releasing their datasheets (oh, yes
Alan): and demand a downstream-PCI-only adapter.
Chicken and the egg dilemna for both sides.
__________________________________________________
Do You Yahoo!?
Yahoo! Sports - Coverage of the 2002 Olympic Games
http://sports.yahoo.com
On Wed, Feb 20, 2002 at 07:53:17PM +0000, Alan Cox wrote:
> > that have microcodes loaded to them NOR require
> > proprietary microcodes accessed to memory by DMA. In
> > other word, don't buy winmodem nor DSL PCI adapters,
> > until those chipset manufacturers publish those
> > datasheets.
>
> The same reasoning goes for another reason. Some of the newest DSL PCI cards are
> in many respects winmodems at multimegabit speed levels, burning huge chunks
> of CPU on a pentium III processor even.
I did some digging on the chipset used by the dlink card, and its made by the
folks at http://www.itexinc.com/ . They claim Linux support, but only in the
form of an infrequently updated binary only module that is only available
through OEMs. Unfortunately, they're uncooperative in providing documentation
for writing an open source driver. It would be Really Nice if the guidelines
on the use of the Linux trademark prevented claims of Linux support without
driver source (ie, forcing binary only module drivers to be marketed as
"partial Linux support through kernel specific binary modules").
-ben
On Fri, 22 Feb 2002 11:36:00 -0500
Benjamin LaHaise <[email protected]> wrote:
> I did some digging on the chipset used by the dlink card, and its made by the
> folks at http://www.itexinc.com/ . They claim Linux support, but only in the
> form of an infrequently updated binary only module that is only available
> through OEMs. Unfortunately, they're uncooperative in providing documentation
> for writing an open source driver. It would be Really Nice if the guidelines
> on the use of the Linux trademark prevented claims of Linux support without
> driver source (ie, forcing binary only module drivers to be marketed as
> "partial Linux support through kernel specific binary modules").
I guess I would prefer the hard line: if it states "linux support" there has to be a driver source - or at least full docs for _any_ requesting parties. Otherwise the trademark should not be useable at all. If you provide several "stages" of support, poor "Aunt Tilly" user (is this already tm'ed? :-) won't be able to understand the difference - and you distro-guys (this is not meant to be a BadName) want this type of user to a certain extent.
But this is a very purist point of view.
Beat me,
Stephan
Stephan,
I am not sure if you actually requesting anything but here are some
points on this matter. First and foremost we do not make these cards.
They are made by our customers that we supply chips for. We do make the
drivers for these boards but they can then be customized by our
customers. The point is that we do support Linux for our customers and
our customers are not really end-users. Now since end-users are
indirectly our customers we do supply Linux drivers on request. Also if
we cannot meet a request for any reason that information is added to our
Marketing data for future driver development. Due to previous licensing
agreements we cannot release our source code to anyone including our
direct customers and there is no way around that. Sorry. Now if you
have a request for a driver please let me know the kernel version being
used as well as the ADSL protocol that you have. I will see what I can
do to get you a usable driver.
Thank you for your interest in ITeX
Dave Rattay
Applications Engineer
Integrated Telecom Express, Inc
400 Race St.
San Jose, CA 95126
-----Original Message-----
From: Stephan von Krawczynski [mailto:[email protected]]
Sent: Friday, February 22, 2002 9:20 AM
To: Benjamin LaHaise
Cc: [email protected]; [email protected]; [email protected];
[email protected]; ITeX Tech Support
Subject: Re: Dlink DSL PCI Card
On Fri, 22 Feb 2002 11:36:00 -0500
Benjamin LaHaise <[email protected]> wrote:
> I did some digging on the chipset used by the dlink card, and its made
by the
> folks at http://www.itexinc.com/ . They claim Linux support, but only
in the
> form of an infrequently updated binary only module that is only
available
> through OEMs. Unfortunately, they're uncooperative in providing
documentation
> for writing an open source driver. It would be Really Nice if the
guidelines
> on the use of the Linux trademark prevented claims of Linux support
without
> driver source (ie, forcing binary only module drivers to be marketed
as
> "partial Linux support through kernel specific binary modules").
I guess I would prefer the hard line: if it states "linux support" there
has to be a driver source - or at least full docs for _any_ requesting
parties. Otherwise the trademark should not be useable at all. If you
provide several "stages" of support, poor "Aunt Tilly" user (is this
already tm'ed? :-) won't be able to understand the difference - and you
distro-guys (this is not meant to be a BadName) want this type of user
to a certain extent.
But this is a very purist point of view.
Beat me,
Stephan
On Fri, Feb 22, 2002 at 02:33:48PM -0800, Dave Rattay [ITeX] wrote:
> Stephan,
>
> I am not sure if you actually requesting anything but here are some
> points on this matter. First and foremost we do not make these cards.
> They are made by our customers that we supply chips for. We do make the
> drivers for these boards but they can then be customized by our
> customers. The point is that we do support Linux for our customers and
> our customers are not really end-users. Now since end-users are
> indirectly our customers we do supply Linux drivers on request. Also if
> we cannot meet a request for any reason that information is added to our
> Marketing data for future driver development. Due to previous licensing
> agreements we cannot release our source code to anyone including our
> direct customers and there is no way around that. Sorry. Now if you
> have a request for a driver please let me know the kernel version being
> used as well as the ADSL protocol that you have. I will see what I can
> do to get you a usable driver.
You're missing the point. We as developers want the specifications for
the hardware as we're completely willing to write our own driver from
scratch. Until Itex is able to provide this to developers, products
based on your chipsets will remain not recommended for Linux users as
it will continually result in a poor user experience.
-ben
Ben,
Well in that case you will have to talk to marketing and sales. Try
[email protected]
Dave
-----Original Message-----
From: Benjamin LaHaise [mailto:[email protected]]
Sent: Friday, February 22, 2002 2:50 PM
To: Dave Rattay [ITeX]
Cc: Stephan von Krawczynski; [email protected];
[email protected]; [email protected]; [email protected];
ITeX Tech Support
Subject: Re: Dlink DSL PCI Card
On Fri, Feb 22, 2002 at 02:33:48PM -0800, Dave Rattay [ITeX] wrote:
> Stephan,
>
> I am not sure if you actually requesting anything but here are
some
> points on this matter. First and foremost we do not make these cards.
> They are made by our customers that we supply chips for. We do make
the
> drivers for these boards but they can then be customized by our
> customers. The point is that we do support Linux for our customers
and
> our customers are not really end-users. Now since end-users are
> indirectly our customers we do supply Linux drivers on request. Also
if
> we cannot meet a request for any reason that information is added to
our
> Marketing data for future driver development. Due to previous
licensing
> agreements we cannot release our source code to anyone including our
> direct customers and there is no way around that. Sorry. Now if you
> have a request for a driver please let me know the kernel version
being
> used as well as the ADSL protocol that you have. I will see what I
can
> do to get you a usable driver.
You're missing the point. We as developers want the specifications for
the hardware as we're completely willing to write our own driver from
scratch. Until Itex is able to provide this to developers, products
based on your chipsets will remain not recommended for Linux users as
it will continually result in a poor user experience.
-ben
> Stephan,
> [...]
> Due to previous licensing
> agreements we cannot release our source code to anyone including our
> direct customers and there is no way around that. Sorry.
Hello Dave,
first of all: thank you for your clear statement.
Unfortunately your world and our (linux-) world do not match (in my
eyes). If you cannot provide either docs or source code chances will
be very high that your product line (chips) will be disliked by a big
percentage of linux using people. Most of them do not like w*ndows
because they cannot control what the system is doing, you are _never_
sure what it does to your (personal) data and where it will be spread.
If you - being a linux user - would be willing to use binary-only
drivers you are just about in the same situation - you lost control.
Your statement - for me - makes one thing absolutely clear: I will
never use your chips.
> Now if you
> have a request for a driver please let me know the kernel version
being
> used as well as the ADSL protocol that you have. I will see what I
can
> do to get you a usable driver.
I do not want a useable driver. I want to have full control over my
system where my personal data resides on. You deny this. I cannot
certify a customer system including your drivers either. Simply
because I cannot check out the code. This is why my customers will not
buy your chips either.
In fact I would not even take them as free gift. Their use is a severe
security issue.
I have no doubts though that you might sell a lot of pieces in the
w*ndows market.
Regards,
Stephan
Stephan,
I definitely agree that our worlds do not match. While I would hate
to lose any customers there are some things that can not be avoided.
There are many things to be taken into consideration when we make our
drivers. One the vast majority of our customers use Windows and
therefore we must devote most of our time there. Second when a new OS
comes out from Microsoft (such as XP) we are given beta copies 6 months
in advance so that our drivers hit the market along side the new OS
release. This just doesn't happen with Linux. I agree that this would
be much simpler if the source code was released and we had "help" in
driver development but as I said that just can't happen, end of story.
Now as to specs for the board itself you can check with sales because I
am not even sure what our policy is on that and I wish you luck in those
regards.
Dave
-----Original Message-----
From: Stephan von Krawczynski [mailto:[email protected]]
Sent: Friday, February 22, 2002 3:15 PM
To: Dave Rattay [ITeX]
Cc: Benjamin LaHaise; [email protected]; [email protected];
[email protected]; [email protected]; ITeX Tech Support
Subject: RE: Dlink DSL PCI Card
> Stephan,
> [...]
> Due to previous licensing
> agreements we cannot release our source code to anyone including our
> direct customers and there is no way around that. Sorry.
Hello Dave,
first of all: thank you for your clear statement.
Unfortunately your world and our (linux-) world do not match (in my
eyes). If you cannot provide either docs or source code chances will
be very high that your product line (chips) will be disliked by a big
percentage of linux using people. Most of them do not like w*ndows
because they cannot control what the system is doing, you are _never_
sure what it does to your (personal) data and where it will be spread.
If you - being a linux user - would be willing to use binary-only
drivers you are just about in the same situation - you lost control.
Your statement - for me - makes one thing absolutely clear: I will
never use your chips.
> Now if you
> have a request for a driver please let me know the kernel version
being
> used as well as the ADSL protocol that you have. I will see what I
can
> do to get you a usable driver.
I do not want a useable driver. I want to have full control over my
system where my personal data resides on. You deny this. I cannot
certify a customer system including your drivers either. Simply
because I cannot check out the code. This is why my customers will not
buy your chips either.
In fact I would not even take them as free gift. Their use is a severe
security issue.
I have no doubts though that you might sell a lot of pieces in the
w*ndows market.
Regards,
Stephan
On Fri Feb 22, 2002 at 03:27:56PM -0800, Dave Rattay [ITeX] wrote:
> release. This just doesn't happen with Linux. I agree that this would
You have _always_ had access to Linux. You have had the entire
OS along with full source code to it for the last 10 years....
So 6 months of binary only access beats 10 years of full source?
> Now as to specs for the board itself you can check with sales because I
> am not even sure what our policy is on that and I wish you luck in those
> regards.
If you were able to release full specs for your DSL PCI Card
(including specs on DSPs and the microcode needed to drive them),
I would gladly buy one tomorrow, even without a driver, because I
would know that even if nobody else cared about that card under
Linux, I could read the specs myself and fix the driver, or I
could pay somebody to fix the driver for me. Without that
information, the card dies when Dlink loses interest...
-Erik
--
Erik B. Andersen http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--
On Fri, Feb 22, 2002 at 05:49:59PM -0500, Benjamin LaHaise wrote:
> On Fri, Feb 22, 2002 at 02:33:48PM -0800, Dave Rattay [ITeX] wrote:
>
> > Now if you
> > have a request for a driver please let me know the kernel version being
> > used as well as the ADSL protocol that you have. I will see what I can
> > do to get you a usable driver.
>
> You're missing the point.
Well, with such kind offer maybe we should start by asking about a
hundred of different binary drivers for various kernel versions,
"mainline", testing and various vendors, with symbol versioning in
modules and without and multiply that by all supported architectures -
of course. Next multiplier will be variants of ADSL protocols around.
This is just a beginning as these requests will be constantly renewed
with each new kernel. So, the offer still stands?
This is still not very close to what is really expected.
If this sounds riduculous then because it is as all people around
here are trying to explain.
Michal
There oughta be a page somewhere on "How to Sell your Product in the Linux
Market".
On Fri, 22 Feb 2002, Dave Rattay [ITeX] wrote:
> I definitely agree that our worlds do not match. While I would hate
> to lose any customers there are some things that can not be avoided.
> There are many things to be taken into consideration when we make our
> drivers. One the vast majority of our customers use Windows and
> therefore we must devote most of our time there.
Understood.
> Second when a new OS
> comes out from Microsoft (such as XP) we are given beta copies 6 months
> in advance so that our drivers hit the market along side the new OS
> release. This just doesn't happen with Linux.
Sure it does. It happens right here on LKML and some other lists. You
can get early-release images of Linux 2.5 right now. The distribution
builders won't pick up 2.5 for many months.
> I agree that this would
> be much simpler if the source code was released and we had "help" in
> driver development but as I said that just can't happen, end of story.
> Now as to specs for the board itself you can check with sales because I
> am not even sure what our policy is on that and I wish you luck in those
> regards.
Board spec.s are largely irrelevant. It's *chip* spec.s that driver
writers need most. There's no such thing as board drivers for Linux, but
there are a number of chipset drivers. For example, under Windows there's
a separate driver for each company's NE2000 clone, but under Linux the ne
driver handles them all. Board-level differences are usually dismissed as
"quirks", if they are significant at all.
(Stores need to understand this too. A lot of sales have gone to one
rather than another because store B told me what chipset was used in a
board-level product, and store A would not.)
--
Mark H. Wood, Lead System Programmer [email protected]
Our lives are forever changed. But *that* is exactly as it always was.
On Sat, 23 Feb 2002 07:42:10 -0500 (EST), Mark H. Wood wrote:
>> comes out from Microsoft (such as XP) we are given beta copies 6 months
>> in advance so that our drivers hit the market along side the new OS
>> release. This just doesn't happen with Linux.
>
>Sure it does. It happens right here on LKML and some other lists. You
>can get early-release images of Linux 2.5 right now. The distribution
>builders won't pick up 2.5 for many months.
Isn't/wasn't 2.5 meant as a development release ? And 2.4 was meant as production,
but certainly had more iterations than you would have expected of such.
If 2.5 is a development-release, I would not expect distros to pick it up
any time soon - just like they never shipped anything with 2.3. Even when
SuSE started shipping 2.4.0, the distro carried a clear recommendation of
using 2.2.18.
There are significant differences in the way Linux development happens and
the way in which commercial software is release-managed. It's a fact.
If Linux kernel development were to stick with the initial idea that odd releases
are development, even are production, then it might be easier to keep up. In
which case itexcinc would perhaps now be looking at 2.5x with an aim of releasing
drivers for 2.6.
>
>(Stores need to understand this too. A lot of sales have gone to one
>rather than another because store B told me what chipset was used in a
>board-level product, and store A would not.)
Wishful thinking - people who understand the importance of which chipset on which
board are usually not employed as shop-assistants.
/Per
PS: I'm all for the main point that's being argued here - but I do understand
itexcinc's point of view.
regards,
Per Jessen, Zurich
http://www.enidan.com - home of the J1 serial console.
Windows 2001: "I'm sorry Dave ... I'm afraid I can't do that."