<?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-F395C299-6C79-5A0E-8FCF-7931D8405E8A" xml:lang="en"><title>Description</title><prolog><metadata><keywords/></metadata></prolog><conbody>

<p>Server applications have been introduced to the Symbian platform

since v9.0 for <xref href="GUID-4BFEDD79-9502-526A-BA7B-97550A6F0601.dita">Platform

Security</xref>. Principally, they offer a more secure solution to embedded

applications. </p>

<p>A server application is an application that contains a server to which

other applications can connect. Importantly, the server application runs in

a separate <xref href="GUID-E6F08B67-8DBC-5896-946D-BD0D27F82FE2.dita#GUID-E6F08B67-8DBC-5896-946D-BD0D27F82FE2/GUID-AC0E13F3-5DFE-5060-BE09-B76A08632B5E">process</xref> than

that of the client (e.g. another application) that launched it. This lets

each party protect their own memory and storage/disk space and have flexible

security policies (which is in contrast to interactions between DLL-based

applications that allow complete access to the memory and disk space of their

hosts). The server application system permits applications to establish powerful

inter-process communications links. This allows applications and other UI

components to observe the principle of least privilege by only requiring <xref href="GUID-F395C299-6C79-5A0E-8FCF-7931D8405E8A.dita#GUID-F395C299-6C79-5A0E-8FCF-7931D8405E8A/GUID-049F04BC-546E-53F9-8212-C0DC6D5B94C2">capabilities</xref> necessary

to do their job, and not the capabilities of applications that use them or

of components that are used by them. It also allows for better compartmentalisation

of data, by restricting access to <xref href="GUID-F395C299-6C79-5A0E-8FCF-7931D8405E8A.dita#GUID-F395C299-6C79-5A0E-8FCF-7931D8405E8A/GUID-8CACFFD5-51FF-5902-B4B5-2DBA506A3CD9">data

caged files</xref> to a client/sever interface. Correct use of the server

application system can improve the security, stability, robustness, usability

and functionality of a platform. </p>

<p>A client can launch one or many server applications; server applications

can have multiple clients, though they have a one-to-one relationship when

they're running. </p>

<section id="GUID-A3712A05-A6A1-549F-98E6-5E9DEC37006C"><title>Services</title> <p>Clients

and server applications interact through common services, which adhere to

a protocol defining the communication channel between the two. This inter-process

communication channel also provides a security checkpoint for application

capability checking, etc.. </p> <p>Services offer useful functionality to

clients, for example, applications often use other applications to send messages,

view, edit and create files, these can be considered services, which may be

implemented in a number of ways (e.g. a printing service). The client asks

the server (via the application containing it) to perform such tasks. </p> <p>Server

applications and clients can each have one or many services. Each service

type is identified by a UID, see the <xref href="GUID-F7CC81CE-E823-5196-944A-5F7524411A08.dita">Registration

and discovery</xref> section below. </p> </section>

<section id="GUID-049F04BC-546E-53F9-8212-C0DC6D5B94C2"><title>Capabilities</title> <p>The

Symbian <xref href="GUID-4BFEDD79-9502-526A-BA7B-97550A6F0601.dita">Platform Security</xref> system

uses capabilities to police who is allowed to do what in the system. </p> <p>Processes

are assigned capabilities in their executables at build time. These capabilities

indicate the things that the process is allowed to do, for example, <codeph>ReadUserData</codeph> (see <codeph>TCapability::ECapabilityReadUserData</codeph>). </p> <p>Servers can check that their clients have particular capabilities

before they carry out client requests. For example, a contacts server may

check that a client process has the <codeph>ReadUserData</codeph> capability. </p> <p>DLLs

may also be assigned capabilities. This means that the DLL is trusted to be

used in a process which has these capabilities. You cannot load a DLL into

a process when the process has capabilities that the DLL does not have. DLLs

cannot add capabilities to processes, a DLL having a capability does not mean

that the DLL can use that capability. The capabilities that a DLL can use

depend on the process that it is executing in. </p> <p>Before the introduction

of platform security, application and other UI component interaction was achieved

with the following techniques. Embedding allowed one application to use another

inside its process. View switching allowed one application to control which

views are displayed by another. DLLs allowed static functionality to be added

to an application. Plug-in DLLs allowed dynamic addition of functionality

to an application. </p> <p>These techniques are all still available, to some

extent, in the application frameworks with platform security. However, those

that rely on DLLs for UI component interaction are now limited, by the capability

rules described above. To achieve fully flexible relationships between UI

components a client/server or other cross process link between components

is now required. </p> <p>For example, consider a server that allows clients

to send an SMS message. Clearly this will require its clients to have a network

capability. However, it is reasonable for a platform to want the user to decide

when SMS messages are sent, no matter which application (with or without capabilities)

wants to send them. Therefore, the platform will want to supply a UI component

that has the ability to send SMS messages on behalf of other applications,

asking the user for confirmation. If such a component were supplied as a DLL,

it would have to run in the client applications process and run with the capabilities

of that process. This may prevent it from being able to send SMS messages

if the client application does not already have that capability. If, however,

the component is supplied as a separate application running in its own process

with a client/server interface allowing other applications to use it, then

it can always ask the user and always send messages, no matter which application

originates the request. Note that such cases where the user is allowed to

permit actions without capabilities are only valid for "user capabilities". </p> </section>

<section id="GUID-8CACFFD5-51FF-5902-B4B5-2DBA506A3CD9"><title>Data caging</title> <p>The

Symbian <xref href="GUID-4BFEDD79-9502-526A-BA7B-97550A6F0601.dita">Platform Security</xref> system

allows processes to have private files, which are only accessible by that

process. Any code that is executing within a process, including code loaded

in DLLs, can access the files in the process's data cage. Code that is executing

in other processes cannot access the files in the data cage, except where

the owning process explicitly supports it. A process can allow other processes

access to individual files in its data cage by file handle sharing, or by

providing an API which encapsulates access to data caged files. Such access

to data caged files and data must be granted with care according to the sensitivity

of the data. </p> <p>Consider a memo-pad application that stores a user's

memos as files in its data cage. This application wants to be able to send

memos in messages, so it links to a "send-as" library. If this send-as library

loads the message transports as DLLs in the memo-pad application's process,

an evil message transport DLL could access all the memo files, deleting, copying

or altering them. If however the send-as library uses IPC connections to the

message transports, the memo pad application controls exactly how much these

message transports can access, limiting the amount of damage that an evil

message transport can do. Similarly, the memo-pad application may offer an

API to allow other processes to access the memos. Since the memos are considered

to be user data, the access should be limited to processes with <codeph>ReadUserData</codeph> capability. </p> </section>

<section id="GUID-ADF237F8-7E92-4A2A-BD5B-73DA54E1CD0B"><title>See also</title><ul>

<li><p><xref href="GUID-940F3F6E-BA9C-5E19-9AC5-D848B5E175FB.dita">Application

Architecture Overview</xref></p></li>

</ul></section>