This paper provides preliminary information about the Windows Driver Foundation (WDF) for the Microsoft Windows family of operating systems. It describes limitations of the current driver models that have led to the development of WDF, explains how WDF solves these problems, and introduces the fundamental components of WDF.
Introduction to the Windows Driver Foundation
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
I don’t want to turn this into a full on Linux thread.
It is just a thought but could this be done with Linux too?
How easy is it to develop drivers under linux.
Could something be designed that would be a part of all distro’s so that you don’t have to write a driver for all distro’s.
Actually, writing Linux drivers is far more easy than writing Windows drivers.
I am currently writing an article in which I compare functionally equivalent Windows and Linux drivers. The WDM driver takes about 6 times the code of the Linux drivers and is far more complex.
Finally Microsoft seems to address this issue but I am afraid that they will drop support for older ( ie. Win2k ) systems with the new driver model which will render this model useless for many developers which still have to support Win2k.
I am currently writing an article in which I compare functionally equivalent Windows and Linux drivers
That would be an interesting read.
🙂
http://developer.apple.com/hardware/iokit/
Some similarities with IOKit on OS X. I suspect MS won’t be going down the embedded C++ route but the sensible ideas of providing as much consistancy as possible across driver models, the OS does the tedious common stuff and avoiding kernel space drivers if possible exists on OS X now.
Although the title states “porting drivers to kernel 2.6” it provides excellent guides writing a driver for linux kernel 2.6; where? At LWN.net
– http://lwn.net/Articles/driver-porting/ …
my .02 EUR
– Deucalion
PS: Yes, Microsoft is improving astonishingly fast – with Longhorn, we (*x/*bsd) will have sth to fear: advanced scripting via a new shell, a new driver-model, … what’s next?!
PPS: I have faith that we will improve just as fast as soon as we overcome social interaction problems with the charismatic leaders of F/OSS / OSS / FS / whatever
.
Finally Microsoft seems to address this issue but I am afraid that they will drop support for older ( ie. Win2k ) systems
The article says that the WDF would be supported on W2k and later systems… though it’s unclear whether that’s only true for the kernel-mode part or also true for the usermode drivers.
I wonder when this model will be compromised; will it be fast enough for graphics drivers?
Good link from gnu. Stallman is such a whiner. He is really stupid at times. And in this article he proved it once more. Mr Stallman, if you are reading this then my advice is get out of your cry cry baby mode and learn the realities of the world and yeah do some workout fatso because with big pot belly like yous…you ain’t gettin no chicks buddy. May be if you have a chick, you will be more in your senses
)
On the WDF, i guess it takes perfect sense to have a well-defined driver model because it brings consistency to the wohle system. Yes right now writing a smallest windows driver, you have to add so much redundant framework crap and that is what they are solving…so i bet WDF is going to provide same uniformity as provided by WDM but with tons of ease… Looking forward to it..Microsoft is really doing things in the right way moving forward.
–Wolf
I got the impression they only intend to use it for less time critical components, plz correct me if i’m wrong.
The Windows DDK has plenty of example code… if you are looking to write a network driver or a DX8 video driver or a floppy driver. It is next to useless otherwise. Most of the example code won’t even work if compiled in its present state.
Why do you think companies charge tens of thousands of dollars to write custom device drivers for Windows? Making a working WDM is something beyond all but the most experienced programmer, and MS is finally realizing this and addressing the problem.
just buy out http://www.scitechsoft.com
I think the most interesting point in the article is the WDF user-mode driver framework in Longhorn. Microsoft says that most bluescreens are caused by bad third-party drivers, yet Microsoft is the company that looks bad when Windows crashes. To protect the OS (and their reputation), it looks like Microsoft is pushing to support user-mode drivers! yeah!
I think the most interesting point in the article is the WDF user-mode driver framework in Longhorn.
Yeah, this will definately be a big improvement over earlier versions of Windows, and it will put to shame driver frameworks from most other OSs.
“eah, this will definately be a big improvement over earlier versions of Windows, and it will put to shame driver frameworks from most other OSs.”
provided that they stand still which will not happen
>I don’t want to turn this into a full on Linux thread.
>It is just a thought but could this be done with Linux too?
The concept of a kernal level driver matched with a user level one is hardly new. This is the model BeOS used I believe with excellent results.
This is more like Microsoft catches up with the rest of the world than Microsoft raising the bar.
Cheers
David
“This is more like Microsoft catches up with the rest of the world than Microsoft raising the bar. ”
I don’t see Linux raising the bar here…