Re: VirtualLock Question / MTRRs
From: Slava M. Usov (stripit.slough_at_gmx.net)
Date: 03/01/04
- Next message: Slava M. Usov: "Re: Allocating console"
- Previous message: Andrea De Nardis: "Re: Threads in c++"
- In reply to: Phil Barila: "Re: VirtualLock Question / MTRRs"
- Next in thread: Kirk Ferdmann: "Re: VirtualLock Question / MTRRs"
- Reply: Kirk Ferdmann: "Re: VirtualLock Question / MTRRs"
- Messages sorted by: [ date ] [ thread ]
Date: Mon, 1 Mar 2004 14:54:27 +0100
"Phil Barila" <PBarila@Barila.com> wrote in message
news:fJSdndSIrutrNt_dRVn-hA@4dv.net...
> It doesn't happen very often (almost never!), but you've made a mistake,
> Slava. DIRECT_IO buffers are paged and locked by the IO Manager.
Yes, yes, indeed. However, I wanted to stress the point that the MDL must
be either converted to a system address space region [thus 'doubly' mapped]
with MmGetSystemAddressForMdl(Safe)() or set in the map registers with
MapTransfer(). Either way, the MDL cannot be used directly in most cases,
and when it can it is more efficient to use the 'neither' model.
S
- Next message: Slava M. Usov: "Re: Allocating console"
- Previous message: Andrea De Nardis: "Re: Threads in c++"
- In reply to: Phil Barila: "Re: VirtualLock Question / MTRRs"
- Next in thread: Kirk Ferdmann: "Re: VirtualLock Question / MTRRs"
- Reply: Kirk Ferdmann: "Re: VirtualLock Question / MTRRs"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|