<?xml version="1.0" encoding="utf-8"?>

<!-- Copyright (c) 2007-2010 Nokia Corporation and/or its subsidiary(-ies) All rights reserved. -->

<!-- This component and the accompanying materials are made available under the terms of the License 

"Eclipse Public License v1.0" which accompanies this distribution, 

and is available at the URL "http://www.eclipse.org/legal/epl-v10.html". -->

<!-- Initial Contributors:

    Nokia Corporation - initial contribution.

Contributors: 

-->

<!DOCTYPE concept

  PUBLIC "-//OASIS//DTD DITA Concept//EN" "concept.dtd">

<concept id="GUID-7C533836-0D27-5519-BC1D-7153AC8BE4C0" xml:lang="en"><title>Media

Driver Guide</title><shortdesc>Describes the issues that need to be considered, when writing a

media device driver for a writable data demand paging environment. </shortdesc><prolog><metadata><keywords/></metadata></prolog><conbody>

<section id="GUID-46C98F8A-1486-4FAD-918D-3FB0E7CA372E"><title>Purpose</title> <p>This

document explains the points that have to be considered when writing a media

driver in an environment that uses writable data paging. </p> </section>

<section id="GUID-1A89843E-B174-49F0-87D8-A8ADBE933A54"><title>Issues to consider</title> <p>The

main issue to consider when writing a media device driver for a writable demand

paging environment is to avoid page faults from occurring in DFCs, since this

can lead to a deadlock condition between the driver and the client process. </p> <p>This

can be avoided using the following methods: </p> <ul>

<li id="GUID-76EF65C9-63E8-5C9B-99CD-A8BFBBA9B68C"><p>Use shared chunks. </p> <p>Shared

chunks are memory areas that are accessible by both kernel-side and user-side

and they are never paged. </p> <p>This is the best solution for drivers that

involve fast throughput such as media drivers. </p> </li>

<li id="GUID-32198DFA-CCF9-5260-A677-88F5B9FBF315"><p>Use synchronous rather

than asynchronous data transfer </p> <p>This could be done by implementing

the following steps: </p> <ol id="GUID-8C359B63-D166-5A54-8772-811FC2656E66">

<li id="GUID-018F2121-A883-5B2F-B70E-C77069B3F1A1"><p>The client requests

a notification when data is available. </p> </li>

<li id="GUID-0FC1D31E-4937-572B-BD80-99DDFABCC0BF"><p>The data arrives. </p> </li>

<li id="GUID-6B47B0FA-8921-5B12-A71C-0E66DD286FD1"><p>The driver writes data

into an internal buffer and completes the client request. </p> </li>

<li id="GUID-14E4D83B-7361-5DE4-90BF-B9CEB658FF1D"><p>The client makes a read

request. </p> </li>

<li id="GUID-FF2A912E-3FBB-5671-ADBE-090FED1CD873"><p>The driver writes the

data back to the client in the client thread context. </p> </li>

</ol> <p>This approach is easy to implement, however it requires the buffering

of data. </p> </li>

<li id="GUID-08F4A4BD-699B-5800-BD2D-78A5406C19EB"><p>Use the <xref href="GUID-066868C5-12F8-59A5-A2DE-FEDC4F459771.dita">flexible

memory model</xref>  </p> <p>This provides the ability for the memory to be

mapped into a drive's address space as unpaged. </p> <p>This is an alternative

to the use of shared chunks. </p> <p>However, this is not supported on the

moving or multiple memory models. </p> </li>

</ul> </section>

</conbody></concept>