Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Codifex Maximu » Sun, 31 Dec 1899 09:00:00



Microsoft, in it's ongoing battle to get PnP and Advanced Power
Management working properly in NT5, has decided to put them in Kernel
Space.
In doing so, they have also bypassed most of the crossplatform and
portability benefits afforded by the HAL or Hardware Abstraction Layer.
They have also redesigned their driver model to include both PnP and APM
using a specially designed (and proprietary) new API.
The new PnP Kernel Space PnP Manager communicates across the Kernel/User
Space barrier to a User Space PnP Manager...

http://msdn.microsoft.com/isapi/msdnlib.idc?theURL=/library/backgrnd/...

I believe, bypassing the HAL, has been done to provide a boost in speed
to the OS at the expense of stability and portability; a standard
Microsoft Business Decision (tm).
--
Codifex Maximus
Murphy's law needs to be repealed.
Happiness is a static TCP/IP address...

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Craig Kelle » Sun, 31 Dec 1899 09:00:00



> http://msdn.microsoft.com/isapi/msdnlib.idc?theURL=/library/backgrnd/...

> I believe, bypassing the HAL, has been done to provide a boost in speed
> to the OS at the expense of stability and portability; a standard
> Microsoft Business Decision (tm).

Did you expect anything less from a bunch of programmers who couldn't
even come up with DOS on their own?

To be fair, they aren't the same people at all -- but lately, Microsoft
is it's own worst enemy.

--
The wheel is turning but the hamster is dead.



 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by david parso » Sun, 31 Dec 1899 09:00:00




Quote:>Microsoft, in it's ongoing battle to get PnP and Advanced Power
>Management working properly in NT5, has decided to put them in Kernel
>Space.
>In doing so, they have also bypassed most of the crossplatform and
>portability benefits afforded by the HAL or Hardware Abstraction Layer.

   What benefits are these?

   If you compare NT to, oh, Linux, you'd notice that the existance of
   the HAL doesn't seem to have helped MS port NT to any more platforms
   than Linux is ported to.  (And Linux, if anything, has an anti-HAL,
   with internal kernel interfaces are drastically changing every time
   you turn around.)

Quote:>I believe, bypassing the HAL, has been done to provide a boost in speed
>to the OS at the expense of stability and portability; a standard
>Microsoft Business Decision (tm).

   That's what Unix does all the time.   And Unix remains portable.
   It's the triumph of pragmatism over philisophical purity -- it
   might not be as nice to look at, but if it works it works.

                  ____
   david parsons  \bi/  It may be too late for NT, though.  Oh, darn.
                   \/

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Brian Hu » Sun, 31 Dec 1899 09:00:00



[ Talking about the NT HAL ]

Quote:>   What benefits are these?
>   If you compare NT to, oh, Linux, you'd notice that the existance of
>   the HAL doesn't seem to have helped MS port NT to any more platforms
>   than Linux is ported to.  (And Linux, if anything, has an anti-HAL,
>   with internal kernel interfaces are drastically changing every time
>   you turn around.)

The HAL was a classic case of "Good idea, bad implementation".  If you
look at the Linux sources (and the *BSD sources, etc), you'll find a
collection of routines and include files that differ from platform to
platform.  In Linux, take a look in the arch directory for what could be
called "Linux's HAL".  The "Linux HAL", however, is very low level, and
includes lots of macros, which makes the benefit of not using the HAL
small at worst.

Most portable programs do something like this.  Linux developed it's HAL
by actually being ported- mainly the m68K and Alpha ports, IIRC.  What
went into the HAL was determined by the minimal set of what needed to
change from one architecture to the next, and has evolved since then as
new architectures were added.  The NT HAL was laid down as an explicit
API up front- i.e. someone's "best guess" as to what was needed.

Brian

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Tim Smi » Sun, 31 Dec 1899 09:00:00



Quote:>   If you compare NT to, oh, Linux, you'd notice that the existance of
>   the HAL doesn't seem to have helped MS port NT to any more platforms
>   than Linux is ported to.  (And Linux, if anything, has an anti-HAL,

Your reasoning is based on the assumption that ease of porting implies
that there will be many ports.  That is a false assumption.

--Tim Smith

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Anthony O » Sun, 31 Dec 1899 09:00:00


On Sat, 27 Feb 1999 01:32:05 -0600, Codifex Maximus


>Microsoft, in it's ongoing battle to get PnP and Advanced Power
>Management working properly in NT5, has decided to put them in Kernel
>Space.

I tell you - Word is going in there next. You heard it here first.
It'll make Windows exactly like a GUI DOS...

Quote:>In doing so, they have also bypassed most of the crossplatform and
>portability benefits afforded by the HAL or Hardware Abstraction Layer.
>They have also redesigned their driver model to include both PnP and APM
>using a specially designed (and proprietary) new API.

You could have left the proprietary out. It is a given.

Quote:>The new PnP Kernel Space PnP Manager communicates across the Kernel/User
>Space barrier to a User Space PnP Manager...

>http://msdn.microsoft.com/isapi/msdnlib.idc?theURL=/library/backgrnd/...

>I believe, bypassing the HAL, has been done to provide a boost in speed
>to the OS at the expense of stability and portability; a standard
>Microsoft Business Decision (tm).

How can you boost speed on Plunge and Pray? It's hardly a speed
critical part of the OS. The same applies for APM. "This computer
achieves sleep mode 0.2321 milliseconds faster than Win98."???

Considering how many problems there are with the above on the
supposedly better supported (as far as PnP and APM go) Win9x platform,
I think M$ are creating a ticking bomb. (Again).

Regards

Anthony
--
-----------------------------------------
| And when our worlds                   |
| They fall apart                       |
| When the walls come tumbling in       |
| Though we may deserve it              |
| It will be worth it  - Depeche Mode   |
-----------------------------------------

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by jed » Sun, 31 Dec 1899 09:00:00




>>   If you compare NT to, oh, Linux, you'd notice that the existance of
>>   the HAL doesn't seem to have helped MS port NT to any more platforms
>>   than Linux is ported to.  (And Linux, if anything, has an anti-HAL,

>Your reasoning is based on the assumption that ease of porting implies
>that there will be many ports.  That is a false assumption.

        There's really not other good way to demonstrate it.
        Otherwise, it's nice talk that only slows the machine
        down.

--
                Herding Humans ~ Herding Cats

Neither will do a thing unless they really want to, or         |||
is coerced to the point where it will scratch your eyes out   / | \
as soon as your grip slips.

        In search of sane PPP docs? Try http://penguin.lvcm.com

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Mark Metso » Sun, 31 Dec 1899 09:00:00


On Sat, 27 Feb 1999 01:32:05 -0600, Codifex Maximus


>In doing so, they have also bypassed most of the crossplatform and
>portability benefits afforded by the HAL or Hardware Abstraction Layer.
>They have also redesigned their driver model to include both PnP and APM
>using a specially designed (and proprietary) new API.

Are you talking hardware drivers? They changed the API that
hardware manufacturers have to write hardware drivers to?
Or do you refer to drivers of another nature?

If hardware drivers then is this something that the
manufaturers will embrace because it will enable them
to sell a whole new version of their drivers?

Or will it appear to them more as if Microsoft has broken
an implicit or explicit agreement that it had with them?
If so its a wonder the hardware guys cooperate with them
at all instead of looking for more enduring standards to
write to.

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Anthony O » Sun, 31 Dec 1899 09:00:00




>>   If you compare NT to, oh, Linux, you'd notice that the existance of
>>   the HAL doesn't seem to have helped MS port NT to any more platforms
>>   than Linux is ported to.  (And Linux, if anything, has an anti-HAL,

>Your reasoning is based on the assumption that ease of porting implies
>that there will be many ports.  That is a false assumption.

Then what's the point? It's like setting something up for mass
production when you're only going to sell two of them.

Quote:>--Tim Smith

Regards

Anthony
--
-----------------------------------------
| And when our worlds                   |
| They fall apart                       |
| When the walls come tumbling in       |
| Though we may deserve it              |
| It will be worth it  - Depeche Mode   |
-----------------------------------------

 
 
 

Microsoft kills (H)ardware (A)bstraction (L)ayer in NT5/W2K

Post by Anthony O » Sun, 31 Dec 1899 09:00:00




>On Sat, 27 Feb 1999 01:32:05 -0600, Codifex Maximus

>>In doing so, they have also bypassed most of the crossplatform and
>>portability benefits afforded by the HAL or Hardware Abstraction Layer.
>>They have also redesigned their driver model to include both PnP and APM
>>using a specially designed (and proprietary) new API.

>Are you talking hardware drivers? They changed the API that
>hardware manufacturers have to write hardware drivers to?
>Or do you refer to drivers of another nature?

>If hardware drivers then is this something that the
>manufaturers will embrace because it will enable them
>to sell a whole new version of their drivers?

Not sell. The will have to write and give away a whole new version of
their drivers. Manufacturers that try to sell drivers get remembered
at upgrade time.

Quote:>Or will it appear to them more as if Microsoft has broken
>an implicit or explicit agreement that it had with them?
>If so its a wonder the hardware guys cooperate with them
>at all instead of looking for more enduring standards to
>write to.

If they want to sell their hardware in any sort of numbers, then they
write M$ drivers. There is not an alternative.

Yet.

Regards

Anthony
--
-----------------------------------------
| And when our worlds                   |
| They fall apart                       |
| When the walls come tumbling in       |
| Though we may deserve it              |
| It will be worth it  - Depeche Mode   |
-----------------------------------------