Re: Newbie Question About Device Drivers
From: TerryW (NOSPAM)
Date: 08/02/04
- Next message: Yavin: "RE: Client test"
- Previous message: Yavin: "Client test"
- In reply to: Don Burn: "Re: Newbie Question About Device Drivers"
- Next in thread: Phil Barila: "Re: Newbie Question About Device Drivers"
- Reply: Phil Barila: "Re: Newbie Question About Device Drivers"
- Messages sorted by: [ date ] [ thread ]
Date: Mon, 02 Aug 2004 09:48:04 -0500
Thanks for the Post Mr. Burn.
I guess we are trying to achieve a more "Real Time" operating system.
We build an XP embedded operating system and then deploy it. This
works great for a lot of our applications. But for some of our
applications, it is very critical that we monitor things flawlessly.
We have an advantech card that we put on a PCI slot that we use for
physical Digital IO (turning gates on/ checking for photo-eye states)
for some systems it is imperitive that we NEVER miss a state change
for an input. And some inputs cycle at 50 cycles per second. We
notice that during intensive screen activity we miss pulses when using
just C++ to read the address of the input card.
It would be ideal to have a piece of code that just counts how many
times the state has changed and return it to other pieces in shared
memory or on the socket.
Didn't mention the model number of the board (PCI-1751.)
TIA,
Terry W.
On Mon, 2 Aug 2004 09:32:05 -0400, "Don Burn" <burn@stopspam.acm.org>
wrote:
>In Windows NT/2000/XP/2003 you cannot access devices from a user space
>program. Additionally, there are some capabilities such as system wide
>monitoring of filesystem or network activities that need to be done in the
>kernel.
>
>Now the challenge is that the kernel is its own environment, with its own
>API's and requirements. Additionally, in the kernel you have access to the
>whole system space, so a mistake is likely to create a crash. So in general
>if you can achieve what you want in user space, it is probably better to do
>so. In fact Microsoft keeps talking about the possibility of some device
>drivers (such as slow USB devices) being supported in user space in the
>future.
>
>Overall there are a lot of differences in the kernel and user programming,
>give the group an idea of what you are thinking about and we can probably
>give you guidance.
- Next message: Yavin: "RE: Client test"
- Previous message: Yavin: "Client test"
- In reply to: Don Burn: "Re: Newbie Question About Device Drivers"
- Next in thread: Phil Barila: "Re: Newbie Question About Device Drivers"
- Reply: Phil Barila: "Re: Newbie Question About Device Drivers"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
