Re: Allocating and accessing physically contiguous memory in user mode in WinCE 6.0 Beta

Tech-Archive recommends: Fix windows errors by optimizing your registry

I am not sure whether I am understanding about your talk.

CE 6 provides much robustness in their architecture, in terms of VM and
Device drivers.

However, does that need indeed?

I agree on the new VM architecture because it provides more strict
separation among the process address space, and more robustness when
failure has occurred.

On the other hands, user mode device driver may not so appealing to me
because it still has too many restrictions and that as you pointed,
because most embedded systems are closed systems, there is little
portion to play a role for 3rd party.

Still, kernel mode device driver is the default option and I think it
is still far from reality.

Steve Maillet (eMVP) wrote:
ps. Still now, I see many restrictions on udevice architecture. e.g. no
usb support, no filesystem support, no network driver support, etc.
Those are in Kernel mode for performance reasons.

The original post was essentially looking for a way to call
HalAllocCommonBuffer() from user mode code in CE V6.0 since the security
architecture of the kernel blocks that from user mode code. There are a LOT
of Embedded systems out there that used to party on the hardware with out
concern for security because they were completely closed systems. Those
devices now need to change to work on CE6.0. There are several options for
how you go about porting that kind of application in V6.0. A few of them
have been proposed in this thread.



--
Steve Maillet
EmbeddedFusion
www.EmbeddedFusion.com
smaillet at EmbeddedFusion dot com

.

Relevant Pages

  • Re: memcpy to PCI Adapter Memory
    ... > restrictions should be hidden in the HAL and not forced upon device driver ... What if the architecture supports quadword and greater movement? ... >>> PCI by itself is flexible and allows all kinds of transfer cycles: ... >>> Using rep movsd at unaligned address does not guarantee anything, ...
    (microsoft.public.development.device.drivers)
  • Re: [PATCH 1/7] spufs: The SPU file system
    ... > I never really thought of it as a device driver but rather an architecture ... it is now in fs/spufs. ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)
  • Re: User Vs Kernel mode -mouse like - SendInput/CreateFile("COM1",...)
    ... I've written exactly such a program, and it works fine in user mode. ... Don't bother keeping it in memory. ... For USB, the OS's driver should be adequate if you set up your HID ... > I have to do a device driver for a TouchScreen monitor and it is supposed to ...
    (microsoft.public.development.device.drivers)
  • Re: memcpy to PCI Adapter Memory
    ... The point I am trying to make is that these type of architecture dependent ... It would just be so nice for the device driver writer just had to worry ... >> Using rep movsd at unaligned address does not guarantee anything, ...
    (microsoft.public.development.device.drivers)
  • Re: Whats the different things between WinCE5.0 and Windows embedded CE6.0 in the view of device dir
    ... is the way you have to deal with user buffers. ... to create a device driver that runs in user mode (as opposed to kernel ... CE6.0 and WinCE5.0 only in the view of device driver. ... Windows CE engineers. ...
    (microsoft.public.windowsce.embedded)