Re: CPU Cost of KernelOControl

From: Dan McCarty (dmccarty_at_gmail.com)
Date: 01/25/05 

Date: 25 Jan 2005 12:05:47 -0800

We're using KernelIOControl to read I/O in kernel mode for a high-speed
device that has to have a polling loop of <50 microseconds.
Unfortunately, that's about as much as I can tell you about that
because I'm not handling that side of the project. (Earlier, we tried
using memory mapped files to share data between the UI app and the
real-time app but we gave up on that approach in favor of passing
structures in and out via KernelIOControl.)

Sorry, but does PSL stand for protected server library? I'm not sure
what the implications of that are. I'd venture to guess that you
implied that we should minimze our time in KernelIOControl, even at the
expense of getting too much (unneeded) data?

-Dan.

Relevant Pages

  • Re: CPU Cost of KernelOControl
    ... It's not the amount of time in the IOCTL, it's the call itself that is the ... > device that has to have a polling loop of <50 microseconds. ... > real-time app but we gave up on that approach in favor of passing ... > structures in and out via KernelIOControl.) ...
    (microsoft.public.windowsce.platbuilder)
  • Re: CPU Cost of KernelOControl
    ... It's not the amount of time in the IOCTL, it's the call itself that is the ... > device that has to have a polling loop of <50 microseconds. ... > real-time app but we gave up on that approach in favor of passing ... > structures in and out via KernelIOControl.) ...
    (microsoft.public.windowsce.embedded.vc)
  • Re: CPU Cost of KernelOControl
    ... We're using KernelIOControl to read I/O in kernel mode for a high-speed ... device that has to have a polling loop of <50 microseconds. ... real-time app but we gave up on that approach in favor of passing ...
    (microsoft.public.windowsce.embedded.vc)