Xext: geext: simplify dispatcher

[xserver.git] / doc / Xserver-spec.xml

<?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE article PUBLIC "-//OASIS//DTD DocBook XML V4.3//EN"
 "http://www.oasis-open.org/docbook/xml/4.3/docbookx.dtd" [
 <!ENTITY % xorg-defs SYSTEM "defs.ent"> %xorg-defs;
 <!ENTITY % defs SYSTEM "/xserver/doc/xml/xserver.ent"> %defs;
]>

<article>
  <articleinfo>
    <title>Definition of the Porting Layer for the X v11 Sample Server</title>
    <titleabbrev>X Porting Layer</titleabbrev>
    <author>
      <firstname>Susan</firstname><surname>Angebranndt</surname>
      <affiliation><orgname>Digital Equipment Corporation</orgname></affiliation>
    </author>
    <author>
      <firstname>Raymond</firstname><surname>Drewry</surname>
      <affiliation><orgname>Digital Equipment Corporation</orgname></affiliation>
    </author>
    <author>
      <firstname>Philip</firstname><surname>Karlton</surname>
      <affiliation><orgname>Digital Equipment Corporation</orgname></affiliation>
    </author>
    <author>
      <firstname>Todd</firstname><surname>Newman</surname>
      <affiliation><orgname>Digital Equipment Corporation</orgname></affiliation>
    </author>
    <author>
      <firstname>Bob</firstname><surname>Scheifler</surname>
      <affiliation><orgname>Massachusetts Institute of Technology</orgname></affiliation>
    </author>
    <author>
      <firstname>Keith</firstname><surname>Packard</surname>
      <affiliation><orgname>MIT X Consortium</orgname></affiliation>
    </author>
    <author>
      <firstname>David</firstname><othername>P.</othername><surname>Wiggins</surname>
      <affiliation><orgname>X Consortium</orgname></affiliation>
    </author>
    <author>
      <firstname>Jim</firstname><surname>Gettys</surname>
      <affiliation><orgname>X.org Foundation and Hewlett Packard</orgname></affiliation>
    </author>
    <publisher><publishername>The X.Org Foundation</publishername></publisher>
    <releaseinfo>X Version 11, Release &fullrelvers;</releaseinfo>
    <releaseinfo>X Server Version &xserver.version;</releaseinfo>
    <copyright><year>1994</year><holder>X Consortium, Inc.</holder></copyright>
    <copyright><year>2004</year><holder>X.org Foundation, Inc.</holder></copyright>
    <legalnotice>
      <para>Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the ``Software''), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:</para>
      <para>The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.</para>
      <para>THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE X CONSORTIUM BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.</para>
      <para>LK201 and DEC are trademarks of Digital Equipment Corporation.  Macintosh and Apple are trademarks of Apple Computer, Inc.  PostScript is a trademark of Adobe Systems, Inc.  Ethernet is a trademark of Xerox Corporation.  X Window System is a trademark of the X.org Foundation, Inc.  Cray is a trademark of Cray Research, Inc.</para>
    </legalnotice>
    <pubdate>&xserver.reldate;</pubdate>
    <revhistory>
      <revision>
        <revnumber>1.0</revnumber>
        <date>27 Oct 2004</date>
        <authorinitials>sa</authorinitials>
        <revremark>Initial Version</revremark>
      </revision>
      <revision>
        <revnumber>1.1</revnumber>
        <date>27 Oct 2004</date>
        <authorinitials>bs</authorinitials>
        <revremark>Minor Revisions</revremark>
      </revision>
      <revision>
        <revnumber>2.0</revnumber>
        <date>27 Oct 2004</date>
        <authorinitials>kp</authorinitials>
        <revremark>Revised for Release 4 and 5</revremark>
      </revision>
      <revision>
        <revnumber>3.0</revnumber>
        <date>27 Oct 2004</date>
        <authorinitials>dpw</authorinitials>
        <revremark>Revised for Release 6</revremark>
      </revision>
      <revision>
        <revnumber>3.1</revnumber>
        <date>27 Oct 2004</date>
        <authorinitials>jg</authorinitials>
        <revremark>Revised for Release 6.8.2</revremark>
      </revision>
      <revision>
        <revnumber>3.2</revnumber>
        <date>17 Dec 2006</date>
        <authorinitials>efw</authorinitials>
        <revremark>DocBook conversion</revremark>
      </revision>
      <revision>
        <revnumber>3.3</revnumber>
        <date>17 Feb 2008</date>
        <authorinitials>aj</authorinitials>
        <revremark>Revised for backing store changes</revremark>
      </revision>
      <revision>
        <revnumber>3.4</revnumber>
        <date>31 Mar 2008</date>
        <authorinitials>efw</authorinitials>
        <revremark>Revised for devPrivates changes</revremark>
      </revision>
      <revision>
        <revnumber>3.5</revnumber>
        <date>July 2010</date>
        <authorinitials>ac</authorinitials>
        <revremark>Revised for Xorg 1.9 devPrivates changes
          and 1.8 CreateNewResourceType changes</revremark>
      </revision>
      <revision>
        <revnumber>3.6</revnumber>
        <date>July 2012</date>
        <authorinitials>kp</authorinitials>
        <revremark>Revised for X server 1.13 screen-specific devPrivates changes</revremark>
      </revision>
    </revhistory>
    <abstract>
      <para>The following document explains the structure of the X Window System display server and the interfaces among the larger pieces.  It is intended as a reference for programmers who are implementing an X Display Server on their workstation hardware.  It is included with the X Window System source tape, along with the document "Strategies for Porting the X v11 Sample Server."  The order in which you should read these documents is:
      <orderedlist>
        <listitem><para>Read the first section of the "Strategies for Porting" document (Overview of Porting Process).</para></listitem>
        <listitem><para>Skim over this document (the Definition document).</para></listitem>
        <listitem><para>Skim over the remainder of the Strategies document.</para></listitem>
        <listitem><para>Start planning and working, referring to the Strategies and Definition documents.</para></listitem>
      </orderedlist>
      You may also want to look at the following documents:
      <itemizedlist>
        <listitem><para>"The X Window System" for an overview of X.</para></listitem>
        <listitem><para>"Xlib - C Language X Interface" for a view of what the client programmer sees.</para></listitem>
        <listitem><para>"X Window System Protocol" for a terse description of the byte stream protocol between the client and server.</para></listitem>
      </itemizedlist>
      </para>
      <para>To understand this document and the accompanying source code, you should know the C language.  You should be familiar with 2D graphics and windowing concepts such as clipping, bitmaps, fonts, etc.  You should have a general knowledge of the X Window System.  To implement the server code on your hardware, you need to know a lot about your hardware, its graphic display device(s), and (possibly) its networking and multitasking facilities.  This document depends a lot on the source code, so you should have a listing of the code handy.</para>
      <para>Some source in the distribution is directly compilable on your machine.  Some of it will require modification.  Other parts may have to be completely written from scratch.  The distribution also includes source for a sample implementation of a display server which runs on a very wide variety of color and monochrome displays on Linux and *BSD which you will find useful for implementing any type of X server.</para>
      <para>Note to the 2008 edition: at this time this document must be considered incomplete, though improved over the 2004 edition.  In particular, the new Render extension is still lacking good documentation, and has become vital to high performance X implementations.  Modern applications and desktop environments are now much more sensitive to good implementation of the Render extension than in most operations of the old X graphics model.  The shadow frame buffer implementation is also very useful in many circumstances, and also needs documentation.  We hope to rectify these shortcomings in our documentation in the future.  Help would be greatly appreciated.</para>
    </abstract>
  </articleinfo>

<!-- Original authorship information:

.OF 'Porting Layer Definition'- % -'October 27, 2004'
Definition of the Porting Layer
for the X v11 Sample Server
Susan Angebranndt
Raymond Drewry
Philip Karlton
Todd Newman
Digital Equipment Corporation

minor revisions by
Bob Scheifler
Massachusetts Institute of Technology

Revised for Release 4 and Release 5 by
Keith Packard
MIT X Consortium

Revised for Release 6 by
David P. Wiggins
X Consortium

Minor Revisions for Release 6.8.2 by
Jim Gettys
X.org Foundation and Hewlett Packard
-->

<section>
  <title>The X Window System</title>
<para>
The X Window System, or simply "X," is a
windowing system that provides high-performance, high-level,
device-independent graphics.
</para>
<para>
X is a windowing system designed for bitmapped graphic displays.
The display can have a
simple, monochrome display or it can have a color display with up to 32 bits
per pixel with a special graphics processor doing the work.  (In this
document, monochrome means a black and white display with one bit per pixel.
Even though the usual meaning of monochrome is more general, this special
case is so common that we decided to reserve the word for this purpose.)
In practice, monochrome displays are now almost unheard of, with 4 bit
gray scale displays being the low end.
</para>
<para>
X is designed for a networking environment where
users can run applications on machines other than their own workstations.
Sometimes, the connection is over an Ethernet network with a protocol such as TCP/IP;
but, any "reliable" byte stream is allowable.
A high-bandwidth byte stream is preferable; RS-232 at
9600 baud would be slow without compression techniques.
</para>
<para>
X by itself allows great freedom of design.
For instance, it does not include any user interface standard.
Its intent is to "provide mechanism, not policy."
By making it general, it can be the foundation for a wide
variety of interactive software.
</para>
<para>
For a more detailed overview, see the document "The X Window System."
For details on the byte stream protocol, see "X Window System protocol."
</para>
</section>
<section>
<title>Overview of the Server</title>
<para>
The display server
manages windows and simple graphics requests
for the user on behalf of different client applications.
The client applications can be running on any machine on the network.
The server mainly does three things:
<itemizedlist>
  <listitem><para>Responds to protocol requests from existing clients (mostly graphic and text drawing commands)</para></listitem>
  <listitem><para>Sends device input (keystrokes and mouse actions) and other events to existing clients</para></listitem>
  <listitem><para>Maintains client connections</para></listitem>
</itemizedlist>
</para>
<para>
The server code is organized into four major pieces:
<itemizedlist>
  <listitem><para>Device Independent (DIX) layer - code shared among all implementations</para></listitem>
  <listitem><para>Operating System (OS) layer - code that is different for each operating system but is shared among all graphic devices for this operating system</para></listitem>
  <listitem><para>Device Dependent (DDX) layer - code that is (potentially) different for each combination of operating system and graphic device</para></listitem>
  <listitem><para>Extension Interface - a standard way to add features to the X server</para></listitem>
</itemizedlist>
</para>
<para>
The "porting layer" consists of the OS and DDX layers; these are
actually parallel and neither one is on top of the other.
The DIX layer is intended to be portable
without change to target systems and is not
detailed here, although several routines
in DIX that are called by DDX are
documented.
Extensions incorporate new functionality into the server; and require
additional functionality over a simple DDX.
</para>
<para>
The following sections outline the functions of the layers.
Section 3 briefly tells what you need to know about the DIX layer.
The OS layer is explained in Section 4.
Section 5 gives the theory of operation and procedural interface for the
DDX layer.
Section 6 describes the functions which exist for the extension writer.
</para>
</section>

<section>
  <title>DIX Layer</title>
<para>
The DIX layer is the machine and device independent part of X.
The source should be common to all operating systems and devices.
The port process should not include changes to this part, therefore internal interfaces to DIX
modules are not discussed, except for public interfaces to the DDX and the OS layers.
The functions described in this section are available for extension writers to use.
</para>
<para>
In the process of getting your server to work, if
you think that DIX must be modified for purposes other than bug fixes,
you may be doing something wrong.
Keep looking for a more compatible solution.
When the next release of the X server code is available,
you should be able to just drop in the new DIX code and compile it.
If you change DIX,
you will have to remember what changes you made and will have
to change the new sources before you can update to the new version.
</para>
<para>
The heart of the DIX code is a loop called the dispatch loop.
Each time the processor goes around the loop, it sends off accumulated input events
from the input devices to the clients, and it processes requests from the clients.
This loop is the most organized way for the server to
process the asynchronous requests that
it needs to process.
Most of these operations are performed by OS and DDX routines that you must supply.
</para>
<section>
  <title>Server Resource System</title>
<para>
X resources are C structs inside the server.
Client applications create and manipulate these objects
according to the rules of the X byte stream protocol.
Client applications refer to resources with resource IDs,
which are 32-bit integers that are sent over the network.
Within the server, of course, they are just C structs, and we refer to them
by pointers.
</para>
<section>
  <title>Pre-Defined Resource Types</title>
<para>
The DDX layer has several kinds of resources:
<itemizedlist>
<listitem><para>Window</para></listitem>
<listitem><para>Pixmap</para></listitem>
<listitem><para>Screen</para></listitem>
<listitem><para>Device</para></listitem>
<listitem><para>Colormap</para></listitem>
<listitem><para>Font</para></listitem>
<listitem><para>Cursor</para></listitem>
<listitem><para>Graphics Contexts</para></listitem>
</itemizedlist>
</para>
<para>
The type names of the more
important server
structs usually end in "Rec," such as "DeviceRec;"
the pointer types usually end in "Ptr," such as "DevicePtr."
</para>
<para>
The structs and
important defined constants are declared
in .h files that have names that suggest the name of the object.
For instance, there are two .h files for windows,
window.h and windowstr.h.
window.h defines only what needs to be defined in order to use windows
without peeking inside of them;
windowstr.h defines the structs with all of their components in great detail
for those who need it.
</para>
<para>
Three kinds of fields are in these structs:
<itemizedlist>
<listitem><para>Attribute fields - struct fields that contain values like normal structs</para></listitem>
<listitem><para>Pointers to procedures, or structures of procedures, that operate on the object</para></listitem>
<listitem><para>A single private field or a devPrivates list (see <xref linkend="wrappers_and_privates"/>)
used by your DDX code to store private data.</para></listitem>
</itemizedlist>
</para>
<para>
DIX calls through
the struct's procedure pointers to do its tasks.
These procedures are set either directly or indirectly by DDX procedures.
Most of
the procedures described in the remainder of this
document are accessed through one of these structs.
For example, the procedure to create a pixmap
is attached to a ScreenRec and might be called by using the expression
</para>
<para>
<blockquote>
<programlisting>(* pScreen->CreatePixmap)(pScreen, width, height, depth).</programlisting>
</blockquote>
</para>
<para>
All procedure pointers must be set to some routine unless noted otherwise;
a null pointer will have unfortunate consequences.
</para>
<para>
Procedure routines will be indicated in the documentation by this convention:
<blockquote>
<programlisting>void pScreen->MyScreenRoutine(arg, arg, ...)</programlisting>
</blockquote>
as opposed to a free routine, not in a data structure:
<blockquote>
<programlisting>void MyFreeRoutine(arg, arg, ...)</programlisting>
</blockquote>
</para>
<para>
The attribute fields are mostly set by DIX; DDX should not modify them
unless noted otherwise.
</para>
</section>
<section>
  <title>Creating Resources and Resource Types</title>
<para>
These functions should also be called from your extensionInitProc to
allocate all of the various resource classes and types required for
the extension.  Each time the server resets, these types must be reallocated
as the old allocations will have been discarded.
Resource types are integer values starting at 1.  Get
a resource type by calling
<blockquote><programlisting>

    RESTYPE CreateNewResourceType(deleteFunc, char *name)

</programlisting></blockquote>
deleteFunc will be called to destroy all resources with this
type.   name will be used to identify this type of resource
to clients using the X-Resource extension, to security
extensions such as SELinux, and to tracing frameworks such as DTrace.
[The name argument was added in xorg-server 1.8.]
</para>
<para>
Resource classes are masks starting at 1 &lt;&lt; 31 which can
be or'ed with any resource type to provide attributes for the
type.  To allocate a new class bit, call
<blockquote><programlisting>

    RESTYPE CreateNewResourceClass()

</programlisting></blockquote>
</para>
<para>
There are two ways of looking up resources, by type or
by class.  Classes are non-exclusive subsets of the space of
all resources, so you can lookup the union of multiple classes.
(RC_ANY is the union of all classes).</para>
<para>
Note that the appropriate class bits must be or'ed into the value returned
by CreateNewResourceType when calling resource lookup functions.</para>
<para>
If you need to create a ``private'' resource ID for internal use, you
can call FakeClientID.
<blockquote><programlisting>

        XID FakeClientID(client)
            int client;

</programlisting></blockquote>
This allocates from ID space reserved for the server.</para>
<para>
To associate a resource value with an ID, use AddResource.
<blockquote><programlisting>

        Bool AddResource(id, type, value)
            XID id;
            RESTYPE type;
            pointer value;

</programlisting></blockquote>
The type should be the full type of the resource, including any class
bits.  If AddResource fails to allocate memory to store the resource,
it will call the deleteFunc for the type, and then return False.</para>
<para>
To free a resource, use one of the following.
<blockquote><programlisting>

        void FreeResource(id, skipDeleteFuncType)
            XID id;
            RESTYPE skipDeleteFuncType;

        void FreeResourceByType(id, type, skipFree)
            XID id;
            RESTYPE type;
            Bool    skipFree;

</programlisting></blockquote>
FreeResource frees all resources matching the given id, regardless of
type; the type's deleteFunc will be called on each matching resource,
except that skipDeleteFuncType can be set to a single type for which
the deleteFunc should not be called (otherwise pass RT_NONE / X11_RESTYPE_NONE).
FreeResourceByType frees a specific resource matching a given id
and type; if skipFree is true, then the deleteFunc is not called.
</para>
</section>
<section>
  <title>Looking Up Resources</title>
<para>
To look up a resource, use one of the following.
<blockquote><programlisting>

        int dixLookupResourceByType(
            pointer *result,
            XID id,
            RESTYPE rtype,
            ClientPtr client,
            Mask access_mode);

        int dixLookupResourceByClass(
            pointer *result,
            XID id,
            RESTYPE rclass,
            ClientPtr client,
            Mask access_mode);

</programlisting></blockquote>
dixLookupResourceByType finds a resource with the given id and exact type.
dixLookupResourceByClass finds a resource with the given id whose type is
included in any one of the specified classes.
The client and access_mode must be provided to allow security extensions to
check if the client has the right privileges for the requested access.
The bitmask values defined in the dixaccess.h header are or'ed together
to define the requested access_mode.
</para>
</section>
</section>
<section>
  <title>Callback Manager</title>
<para>
To satisfy a growing number of requests for the introduction of ad hoc
notification style hooks in the server, a generic callback manager was
introduced in R6.  A callback list object can be introduced for each
new hook that is desired, and other modules in the server can register
interest in the new callback list.  The following functions support
these operations.</para>
<para>
Before getting bogged down in the interface details, an typical usage
example should establish the framework.  Let's look at the
ClientStateCallback in dix/dispatch.c.  The purpose of this particular
callback is to notify interested parties when a client's state
(initial, running, gone) changes.  The callback is "created" in this
case by simply declaring a variable:
<blockquote><programlisting>
        CallbackListPtr ClientStateCallback;
</programlisting></blockquote>
</para>
<para>
Whenever the client's state changes, the following code appears, which notifies
all interested parties of the change:
<blockquote><programlisting>
        if (ClientStateCallback) CallCallbacks(&amp;ClientStateCallback, (pointer)client);
</programlisting></blockquote>
</para>
<para>
Interested parties subscribe to the ClientStateCallback list by saying:
<blockquote><programlisting>
        AddCallback(&amp;ClientStateCallback, func, data);
</programlisting></blockquote>
</para>
<para>
When CallCallbacks is invoked on the list, func will be called thusly:
<blockquote><programlisting>
        (*func)(&amp;ClientStateCallback, data, client)
</programlisting></blockquote>
</para>
<para>
Now for the details.
<blockquote><programlisting>

        Bool AddCallback(pcbl, callback, subscriber_data)
            CallbackListPtr *pcbl;
            CallbackProcPtr callback;
            pointer         subscriber_data;

</programlisting></blockquote>
Adds the (callback, subscriber_data) pair to the given callback list.  Creates the callback
list if it doesn't exist.  Returns TRUE if successful.</para>
<para>
<blockquote><programlisting>

        Bool DeleteCallback(pcbl, callback, subscriber_data)
            CallbackListPtr *pcbl;
            CallbackProcPtr callback;
            pointer         subscriber_data;

</programlisting></blockquote>
Removes the (callback, data) pair to the given callback list if present.
Returns TRUE if (callback, data) was found.</para>
<para>
<blockquote><programlisting>

        void CallCallbacks(pcbl, call_data)
            CallbackListPtr    *pcbl;
            pointer         call_data;

</programlisting></blockquote>
For each callback currently registered on the given callback list, call
it as follows:
<blockquote><programlisting>

        (*callback)(pcbl, subscriber_data, call_data);
</programlisting></blockquote>
</para>
<para>
<blockquote><programlisting>
        void DeleteCallbackList(pcbl)
            CallbackListPtr    *pcbl;

</programlisting></blockquote>
Destroys the given callback list.</para>
</section>
<section>
  <title>Extension Interfaces</title>
<para>
This function should be called from your extensionInitProc which
should be called by InitExtensions.
<blockquote><programlisting>

        ExtensionEntry *AddExtension(name, NumEvents,NumErrors,
                MainProc, SwappedMainProc, CloseDownProc, MinorOpcodeProc)

                const char *name;  /*Null terminate string; case matters*/
                int NumEvents;
                int NumErrors;
                int (* MainProc)(ClientPtr);/*Called if client matches server order*/
                int (* SwappedMainProc)(ClientPtr);/*Called if client differs from server*/
                void (* CloseDownProc)(ExtensionEntry *);
                unsigned short (*MinorOpcodeProc)(ClientPtr);

</programlisting></blockquote>
name is the name used by clients to refer to the extension.  NumEvents is the
number of event types used by the extension, NumErrors is the number of
error codes needed by the extension.  MainProc is called whenever a client
accesses the major opcode assigned to the extension.  SwappedMainProc is
identical, except the client using the extension has reversed byte-sex.
CloseDownProc is called at server reset time to deallocate any private
storage used by the extension.  MinorOpcodeProc is used by DIX to place the
appropriate value into errors.  The DIX routine StandardMinorOpcode can be
used here which takes the minor opcode from the normal place in the request
(i.e. just after the major opcode).</para>
</section>
<section>
  <title>Macros and Other Helpers</title>
<para>
There are a number of macros in Xserver/include/dix.h which
are useful to the extension writer.  Ones of particular interest
are: REQUEST, REQUEST_SIZE_MATCH, REQUEST_AT_LEAST_SIZE,
REQUEST_FIXED_SIZE, LEGAL_NEW_RESOURCE, and
VALIDATE_DRAWABLE_AND_GC. Useful byte swapping macros can be found
in Xserver/include/dix.h: WriteReplyToClient and WriteSwappedDataToClient; and
in Xserver/include/misc.h: bswap_64, bswap_32, bswap_16, LengthRestB, LengthRestS,
LengthRestL, SwapRestS, SwapRestL, swapl, swaps, cpswapl, and cpswaps.</para>
</section>
</section>

<section>
  <title>OS Layer</title>
<para>
This part of the source consists of a few routines that you have to rewrite
for each operating system.
These OS functions maintain the client connections and schedule work
to be done for clients.
They also provide an interface to font files,
font name to file name translation, and
low level memory management.
<blockquote>
<programlisting>void OsInit()</programlisting>
</blockquote>
OsInit initializes your OS code, performing whatever tasks need to be done.
Frequently there is not much to be done.
The sample server implementation is in Xserver/os/osinit.c.
</para>
<section>
  <title>Scheduling and Request Delivery</title>
<para>
The main dispatch loop in DIX creates the illusion of multitasking between
different windows, while the server is itself but a single process.
The dispatch loop breaks up the work for each client into small digestible parts.
Some parts are requests from a client, such as individual graphic commands.
Some parts are events delivered to the client, such as keystrokes from the user.
The processing of events and requests for different
clients can be interleaved with one another so true multitasking
is not needed in the server.
</para>
<para>
You must supply some of the pieces for proper scheduling between clients.
<blockquote>
<programlisting>
        int WaitForSomething(pClientReady)
                int *pClientReady;
</programlisting>
</blockquote>
</para>
<para>
WaitForSomething is the scheduler procedure you must write that will
suspend your server process until something needs to be done.
This call should
make the server suspend until one or more of the following occurs:
<itemizedlist>
<listitem><para>There is an input event from the user or hardware (see SetInputCheck())</para></listitem>
<listitem><para>There are requests waiting from known clients, in which case you should return a count of clients stored in pClientReady</para></listitem>
<listitem><para>A new client tries to connect, in which case you should create the client and then continue waiting</para></listitem>
</itemizedlist>
</para>
<para>
Before WaitForSomething() computes the masks to pass to select, poll or
similar operating system interface, it needs to
see if there is anything to do on the work queue; it must call a DIX
routine called ProcessWorkQueue().
<blockquote>
<programlisting>
        ProcessWorkQueue ();
</programlisting>
</blockquote>
</para>
<para>
If WaitForSomething() decides it is about to do something that might block
(in the sample server,  before it calls select() or poll) it must call a DIX
routine called BlockHandler().
<blockquote>
<programlisting>
        void BlockHandler(void *pTimeout)
</programlisting>
</blockquote>
The types of the arguments are for agreement between the OS and DDX
implementations,  but the pTimeout is a pointer to the information
determining how long the block is allowed to last.
</para>
<para>
In the sample server,  pTimeout is a pointer.
</para>
<para>
The DIX BlockHandler() iterates through the Screens,  for each one calling
its BlockHandler.  A BlockHandler is declared thus:
<blockquote>
<programlisting>
        void xxxBlockHandler(ScreenPtr pScreen, void *pTimeout)
</programlisting>
</blockquote>
The arguments are a pointer to the Screen, and the arguments to the
DIX BlockHandler().
</para>
<para>
Immediately after WaitForSomething returns from the
block,  even if it didn't actually block,  it must call the DIX routine
WakeupHandler().
<blockquote>
<programlisting>
        void WakeupHandler(int result)
</programlisting>
</blockquote>
Once again,  the types are not specified by DIX.  The result is the
success indicator for the thing that (may have) blocked.
In the sample server, result is the result from select() (or equivalent
operating system function).
</para>
<para>
The DIX WakeupHandler() calls each Screen's
WakeupHandler.  A WakeupHandler is declared thus:
<blockquote>
<programlisting>
        void xxxWakeupHandler(ScreenPtr pScreen, int result)
</programlisting>
</blockquote>
The arguments are the Screen, of the Screen, and the arguments to
the DIX WakeupHandler().
</para>
<para>
In addition to the per-screen BlockHandlers, any module may register
block and wakeup handlers (only together) using:
<blockquote>
<programlisting>
        Bool RegisterBlockAndWakeupHandlers (blockHandler, wakeupHandler, blockData)
                ServerBlockHandlerProcPtr    blockHandler;
                ServerWakeupHandlerProcPtr   wakeupHandler;
                pointer blockData;
</programlisting>
</blockquote>
A FALSE return code indicates that the registration failed for lack of
memory.  To remove a registered Block handler at other than server reset time
(when they are all removed automatically), use:
<blockquote>
<programlisting>
        RemoveBlockAndWakeupHandlers (blockHandler, wakeupHandler, blockData)
                ServerBlockHandlerProcPtr   blockHandler;
                ServerWakeupHandlerProcPtr  wakeupHandler;
                pointer blockData;
</programlisting>
</blockquote>
All three arguments must match the values passed to
RegisterBlockAndWakeupHandlers.
</para>
<para>
These registered block handlers are called before the per-screen handlers:
<blockquote>
<programlisting>
        void (*ServerBlockHandler) (void *blockData, void *pTimeout)
</programlisting>
</blockquote>
</para>
<para>
Sometimes block handlers need to adjust the time referenced by pTimeout,
which on UNIX family systems is generally represented by a struct timeval
consisting of seconds and microseconds in 32 bit values.
As a convenience to reduce error prone struct timeval computations which
require modulus arithmetic and correct overflow behavior in the face of
millisecond wrapping through 32 bits,
<blockquote><programlisting>

        void AdjustWaitForDelay(void *pTimeout, unsigned long newdelay)

</programlisting></blockquote>
has been provided.
</para>
<para>
Any wakeup handlers registered with RegisterBlockAndWakeupHandlers will
be called after the Screen handlers:
<blockquote><programlisting>

        void (*ServerWakeupHandler) (void *blockData, int result)
</programlisting></blockquote>
</para>
<para>
The WaitForSomething on the sample server also has a built
in screen saver that darkens the screen if no input happens for a period of time.
The sample server implementation is in Xserver/os/WaitFor.c.
</para>
<para>
Note that WaitForSomething() may be called when you already have several
outstanding things (events, requests, or new clients) queued up.
For instance, your server may have just done a large graphics request,
and it may have been a long time since WaitForSomething() was last called.
If many clients have lots of requests queued up, DIX will only service
some of them for a given client
before going on to the next client (see isItTimeToYield, below).
Therefore, WaitForSomething() will have to report that these same clients
still have requests queued up the next time around.
</para>
<para>
An implementation should return information on as
many outstanding things as it can.
For instance, if your implementation always checks for client data first and does not
report any input events until there is no client data left,
your mouse and keyboard might get locked out by an application that constantly
barrages the server with graphics drawing requests.
Therefore, as a general rule, input devices should always have priority over graphics
devices.
</para>
<para>
A list of indexes (client->index) for clients with data ready to be read or
processed should be returned in pClientReady, and the count of indexes
returned as the result value of the call.
These are not clients that have full requests ready, but any clients who have
any data ready to be read or processed.
The DIX dispatcher
will process requests from each client in turn by calling
ReadRequestFromClient(), below.
</para>
<para>
WaitForSomething() must create new clients as they are requested (by
whatever mechanism at the transport level).  A new client is created
by calling the DIX routine:
<blockquote><programlisting>

        ClientPtr NextAvailableClient(ospriv)
                pointer ospriv;
</programlisting></blockquote>
This routine returns NULL if a new client cannot be allocated (e.g. maximum
number of clients reached).  The ospriv argument will be stored into the OS
private field (pClient->osPrivate), to store OS private information about the
client.  In the sample server, the osPrivate field contains the
number of the socket for this client. See also "New Client Connections."
NextAvailableClient() will call InsertFakeRequest(), so you must be
prepared for this.
</para>
<para>
If there are outstanding input events,
you should make sure that the two SetInputCheck() locations are unequal.
The DIX dispatcher will call your implementation of ProcessInputEvents()
until the SetInputCheck() locations are equal.
</para>
<para>
The sample server contains an implementation of WaitForSomething().
The
following two routines indicate to WaitForSomething() what devices should
be waited for.   fd is an OS dependent type; in the sample server
it is an open file descriptor.
<blockquote><programlisting>

        int AddEnabledDevice(fd)
                int fd;

        int RemoveEnabledDevice(fd)
                int fd;
</programlisting></blockquote>
These two routines are
usually called by DDX from the initialize cases of the
Input Procedures that are stored in the DeviceRec (the
routine passed to AddInputDevice()).
The sample server implementation of AddEnabledDevice
and RemoveEnabledDevice are in Xserver/os/connection.c.
</para>
<section>
  <title>Timer Facilities</title>
<para>
Similarly, the X server or an extension may need to wait for some timeout.
Early X releases implemented this functionality using block and wakeup handlers,
but this has been rewritten to use a general timer facilty, and the
internal screen saver facilities reimplemented to use Timers.
These functions are TimerInit, TimerForce, TimerSet, TimerCheck, TimerCancel,
and TimerFree, as defined in Xserver/include/os.h. A callback function will be called
when the timer fires, along with the current time, and a user provided argument.
<blockquote><programlisting>
        typedef struct _OsTimerRec *OsTimerPtr;

        typedef CARD32 (*OsTimerCallback)(
                OsTimerPtr /* timer */,
                CARD32 /* time */,
                pointer /* arg */);

         OsTimerPtr TimerSet( OsTimerPtr /* timer */,
                int /* flags */,
                CARD32 /* millis */,
                OsTimerCallback /* func */,
                pointer /* arg */);

</programlisting></blockquote>
</para>
<para>
TimerSet returns a pointer to a timer structure and sets a timer to the specified time
with the specified argument.  The flags can be TimerAbsolute and TimerForceOld.
The TimerSetOld flag controls whether if the timer is reset and the timer is pending, the
whether the callback function will get called.
The TimerAbsolute flag sets the callback time to an absolute time in the future rather
than a time relative to when TimerSet is called.
TimerFree should be called to free the memory allocated
for the timer entry.
<blockquote><programlisting>
        void TimerInit(void)

        Bool TimerForce(OsTimerPtr /* pTimer */)

        void TimerCheck(void);

        void TimerCancel(OsTimerPtr /* pTimer */)

        void TimerFree(OsTimerPtr /* pTimer */)
</programlisting></blockquote>
</para>
<para>
TimerInit frees any existing timer entries. TimerForce forces a call to the timer's
callback function and returns true if the timer entry existed, else it returns false and
does not call the callback function. TimerCancel will cancel the specified timer.
TimerFree calls TimerCancel and frees the specified timer.
Calling TimerCheck will force the server to see if any timer callbacks should be called.
</para>
</section>
</section>
<section>
  <title>New Client Connections</title>
<para>
The process whereby a new client-server connection starts up is
very dependent upon what your byte stream mechanism.
This section describes byte stream initiation using examples from the TCP/IP
implementation on the sample server.
</para>
<para>
The first thing that happens is a client initiates a connection with the server.
How a client knows to do this depends upon your network facilities and the
Xlib implementation.
In a typical scenario, a user named Fred
on his X workstation is logged onto a Cray
supercomputer running a command shell in an X window.  Fred can type shell
commands and have the Cray respond as though the X server were a dumb terminal.
Fred types in a command to run an X client application that was linked with Xlib.
Xlib looks at the shell environment variable DISPLAY, which has the
value "fredsbittube:0.0."
The host name of Fred's workstation is "fredsbittube," and the 0s are
for multiple screens and multiple X server processes.
(Precisely what
happens on your system depends upon how X and Xlib are implemented.)
</para>
<para>
The client application calls a TCP routine on the
Cray to open a TCP connection for X
to communicate with the network node "fredsbittube."
The TCP software on the Cray does this by looking up the TCP
address of "fredsbittube" and sending an open request to TCP port 6000
on fredsbittube.
</para>
<para>
All X servers on TCP listen for new clients on port 6000 by default;
this is known as a "well-known port" in IP terminology.
</para>
<para>
The server receives this request from its port 6000
and checks where it came from to see if it is on the server's list
of "trustworthy" hosts to talk to.
Then, it opens another port for communications with the client.
This is the byte stream that all X communications will go over.
</para>
<para>
Actually, it is a bit more complicated than that.
Each X server process running on the host machine is called a "display."
Each display can have more than one screen that it manages.
"corporatehydra:3.2" represents screen 2 on display 3 on
the multi-screened network node corporatehydra.
The open request would be sent on well-known port number 6003.
</para>
<para>
Once the byte stream is set up, what goes on does not depend very much
upon whether or not it is TCP.
The client sends an xConnClientPrefix struct (see Xproto.h) that has the
version numbers for the version of Xlib it is running, some byte-ordering information,
and two character strings used for authorization.
If the server does not like the authorization strings
or the version numbers do not match within the rules,
or if anything else is wrong, it sends a failure
response with a reason string.
</para>
<para>
If the information never comes, or comes much too slowly, the connection
should be broken off.  You must implement the connection timeout.  The
sample server implements this by keeping a timestamp for each still-connecting
client and, each time just before it attempts to accept new connections, it
closes any connection that are too old.
The connection timeout can be set from the command line.
</para>
<para>
You must implement whatever authorization schemes you want to support.
The sample server on the distribution tape supports a simple authorization
scheme.  The only interface seen by DIX is:
<blockquote><programlisting>

        char *
        ClientAuthorized(client, proto_n, auth_proto, string_n, auth_string)
            ClientPtr client;
            unsigned int proto_n;
            char *auth_proto;
            unsigned int string_n;
            char *auth_string;
</programlisting></blockquote>
DIX will only call this once per client, once it has read the full initial
connection data from the client.  If the connection should be
accepted ClientAuthorized() should return NULL, and otherwise should
return an error message string.
</para>
<para>
Accepting new connections happens internally to WaitForSomething().
WaitForSomething() must call the DIX routine NextAvailableClient()
to create a client object.
Processing of the initial connection data will be handled by DIX.
Your OS layer must be able to map from a client
to whatever information your OS code needs to communicate
on the given byte stream to the client.
DIX uses this ClientPtr to refer to
the client from now on.   The sample server uses the osPrivate field in
the ClientPtr to store the file descriptor for the socket, the
input and output buffers, and authorization information.
</para>
<para>
To initialize the methods you choose to allow clients to connect to
your server, main() calls the routine
<blockquote><programlisting>

        void CreateWellKnownSockets()
</programlisting></blockquote>
This routine is called only once, and not called when the server
is reset.  To recreate any sockets during server resets, the following
routine is called from the main loop:
<blockquote><programlisting>

        void ResetWellKnownSockets()
</programlisting></blockquote>
Sample implementations of both of these routines are found in
Xserver/os/connection.c.
</para>
<para>
For more details, see the section called "Connection Setup" in the X protocol specification.
</para>
</section>
<section>
  <title>Reading Data from Clients</title>
<para>
Requests from the client are read in as a byte stream by the OS layer.
They may be in the form of several blocks of bytes delivered in sequence; requests may
be broken up over block boundaries or there may be many requests per block.
Each request carries with it length information.
It is the responsibility of the following routine to break it up into request blocks.
<blockquote><programlisting>

        int ReadRequestFromClient(who)
                ClientPtr who;
</programlisting></blockquote>
</para>
<para>
You must write
the routine ReadRequestFromClient() to get one request from the byte stream
belonging to client "who."
You must swap the third and fourth bytes (the second 16-bit word) according to the
byte-swap rules of
the protocol to determine the length of the
request.
This length is measured in 32-bit words, not in bytes.  Therefore, the
theoretical maximum request is 256K.
(However, the maximum length allowed is dependent upon the server's input
buffer.  This size is sent to the client upon connection.  The maximum
size is the constant MAX_REQUEST_SIZE in Xserver/include/os.h)
The rest of the request you return is
assumed NOT to be correctly swapped for internal
use, because that is the responsibility of DIX.
</para>
<para>
The 'who' argument is the ClientPtr returned from WaitForSomething.
The return value indicating status should be set to the (positive) byte count if the read is successful,
0 if the read was blocked, or a negative error code if an error happened.
</para>
<para>
You must then store a pointer to
the bytes of the request in the client request buffer field;
who->requestBuffer.  This can simply be a pointer into your buffer;
DIX may modify it in place but will not otherwise cause damage.
Of course, the request must be contiguous; you must
shuffle it around in your buffers if not.
</para>
<para>
The sample server implementation is in Xserver/os/io.c.
</para>
<section><title>Inserting Data for Clients</title>
<para>
DIX can insert data into the client stream, and can cause a "replay" of
the current request.
<blockquote><programlisting>

        Bool InsertFakeRequest(client, data, count)
            ClientPtr client;
            char *data;
            int count;

        int ResetCurrentRequest(client)
            ClientPtr client;
</programlisting></blockquote>
</para>
<para>
InsertFakeRequest() must insert the specified number of bytes of data
into the head of the input buffer for the client.  This may be a
complete request, or it might be a partial request.  For example,
NextAvailableCient() will insert a partial request in order to read
the initial connection data sent by the client.  The routine returns FALSE
if memory could not be allocated.  ResetCurrentRequest()
should "back up" the input buffer so that the currently executing request
will be reexecuted.  DIX may have altered some values (e.g. the overall
request length), so you must recheck to see if you still have a complete
request.  ResetCurrentRequest() should always cause a yield (isItTimeToYield).
</para>
</section>
</section>

<section>
  <title>Sending Events, Errors And Replies To Clients</title>
<para>
<blockquote><programlisting>

        int WriteToClient(who, n, buf)
                ClientPtr who;
                int n;
                char *buf;
</programlisting></blockquote>
WriteToClient should write n bytes starting at buf to the
ClientPtr "who".
It returns the number of bytes written, but for simplicity,
the number returned must be either the same value as the number
requested, or -1, signaling an error.
The sample server implementation is in Xserver/os/io.c.
</para>
<para>
<blockquote><programlisting>
        void SendErrorToClient(client, majorCode, minorCode, resId, errorCode)
            ClientPtr client;
            unsigned int majorCode;
            unsigned int minorCode;
            XID resId;
            int errorCode;
</programlisting></blockquote>
SendErrorToClient can be used to send errors back to clients,
although in most cases your request function should simply return
the error code, having set client->errorValue to the appropriate
error value to return to the client, and DIX will call this
function with the correct opcodes for you.
</para>
<para>
<blockquote><programlisting>

        void FlushAllOutput()

        void FlushIfCriticalOutputPending()

        void SetCriticalOutputPending()
</programlisting></blockquote>
These three routines may be implemented to support buffered or delayed
writes to clients, but at the very least, the stubs must exist.
FlushAllOutput() unconditionally flushes all output to clients;
FlushIfCriticalOutputPending() flushes output only if
SetCriticalOutputPending() has be called since the last time output
was flushed.
The sample server implementation is in Xserver/os/io.c and
actually ignores requests to flush output on a per-client basis
if it knows that there
are requests in that client's input queue.
</para>
</section>
<section>
  <title>Font Support</title>
<para>
In the sample server, fonts are encoded in disk files or fetched from the
font server.   The two fonts required by the server, <quote>fixed</quote>
and <quote>cursor</quote> are commonly compiled into the font library.
For disk fonts, there is one file per font, with a file name like
"fixed.pcf".  Font server fonts are read over the network using the
X Font Server Protocol.  The disk directories containing disk fonts and
the names of the font servers are listed together in the current "font path."
</para>
<para>
In principle, you can put all your fonts in ROM or in RAM in your server.
You can put them all in one library file on disk.
You could generate them on the fly from stroke descriptions.  By placing the
appropriate code in the Font Library, you will automatically export fonts in
that format both through the X server and the Font server.
</para>
<para>
The code for processing fonts in different formats, as well as handling the
metadata files for them on disk (such as <filename>fonts.dir</filename>) is
located in the libXfont library, which is provided as a separately compiled
module.  These routines are
shared between the X server and the Font server, so instead of this document
specifying what you must implement, simply refer to the font
library interface specification for the details.  All of the interface code to the Font
library is contained in dix/dixfonts.c
</para>
</section>
<section>
  <title>Memory Management</title>
<para>
Memory management is based on functions in the C runtime library, malloc(),
realloc(), and free(), and you should simply call the C library functions
directly.  Consult a C runtime library reference manual for more details.
</para>
<para>
Treat memory allocation carefully in your implementation.  Memory
leaks can be very hard to find and are frustrating to a user.  An X
server could be running for days or weeks without being reset, just
like a regular terminal.  If you leak a few dozen k per day, that will
add up and will cause problems for users that leave their workstations
on.
</para>
</section>
<section>
  <title>Client Scheduling</title>
<para>
The X server
has the ability to schedule clients much like an operating system would,
suspending and restarting them without regard for the state of their input
buffers.  This functionality allows the X server to suspend one client and
continue processing requests from other clients while waiting for a
long-term network activity (like loading a font) before continuing with the
first client.
<blockquote><programlisting>
        Bool isItTimeToYield;
</programlisting></blockquote>
isItTimeToYield is a global variable you can set
if you want to tell
DIX to end the client's "time slice" and start paying attention to the next client.
After the current request is finished, DIX will move to the next client.
</para>
<para>
In the sample
server, ReadRequestFromClient() sets isItTimeToYield after
10 requests packets in a row are read from the same client.
</para>
<para>
This scheduling algorithm can have a serious effect upon performance when two
clients are drawing into their windows simultaneously.
If it allows one client to run until its request
queue is empty by ignoring isItTimeToYield, the client's queue may
in fact never empty and other clients will be blocked out.
On the other hand, if it switches between different clients too quickly,
performance may suffer due to too much switching between contexts.
For example, if a graphics processor needs to be set up with drawing modes
before drawing, and two different clients are drawing with
different modes into two different windows, you may
switch your graphics processor modes so often that performance is impacted.
</para>
<para>
See the Strategies document for
heuristics on setting isItTimeToYield.
</para>
<para>
The following functions provide the ability to suspend request
processing on a particular client, resuming it at some later time:
<blockquote><programlisting>

        int IgnoreClient (who)
                ClientPtr who;

        int AttendClient (who)
                ClientPtr who;
</programlisting></blockquote>
Ignore client is responsible for pretending that the given client doesn't
exist.  WaitForSomething should not return this client as ready for reading
and should not return if only this client is ready.  AttendClient undoes
whatever IgnoreClient did, setting it up for input again.
</para>
<para>
Three functions support "process control" for X clients:
<blockquote><programlisting>

        Bool ClientSleep (client, function, closure)
                ClientPtr       client;
                Bool            (*function)();
                pointer         closure;

</programlisting></blockquote>
This suspends the current client (the calling routine is responsible for
making its way back to Dispatch()).  No more X requests will be processed
for this client until ClientWakeup is called.
<blockquote><programlisting>

        Bool ClientSignal (client)
                ClientPtr       client;

</programlisting></blockquote>
This function causes a call to the (*function) parameter passed to
ClientSleep to be queued on the work queue.  This does not automatically
"wakeup" the client, but the function called is free to do so by calling:
<blockquote><programlisting>

        ClientWakeup (client)
                ClientPtr       client;

</programlisting></blockquote>
This re-enables X request processing for the specified client.
</para>
</section>
<section>
  <title>Other OS Functions</title>
<para>
<blockquote><programlisting>
        void
        ErrorF(char *f, ...)

        void
        FatalError(char *f, ...)
</programlisting></blockquote>
You should write these three routines to provide for diagnostic output
from the dix and ddx layers, although implementing them to produce no
output will not affect the correctness of your server.  ErrorF() and
FatalError() take a printf() type of format specification in the first
argument and an implementation-dependent number of arguments following
that.  Normally, the formats passed to ErrorF() and FatalError()
should be terminated with a newline.
</para>
<para>
After printing the message arguments, FatalError() must be implemented
such that the server will call ddxGiveUp(EXIT_ERR_ABORT) to give the ddx layer
a chance to reset the hardware, and then
terminate the server; it must not return.
</para>
<para>
The sample server implementation for these routines
is in Xserver/os/log.c along with other routines for logging messages.
</para>
</section>
</section>

<section>
  <title>DDX Layer</title>
<para>
This section describes the
interface between DIX and DDX.
While there may be an OS-dependent driver interface between DDX
and the physical device, that interface is left to the DDX
implementor and is not specified here.
</para>
<para>
The DDX layer does most of its work through procedures that are
pointed to by different structs.
As previously described, the behavior of these resources is largely determined by
these procedure pointers.
Most of these routines are for graphic display on the screen or support functions thereof.
The rest are for user input from input devices.
</para>
<section>
  <title>Input</title>
<para>
In this document "input" refers to input from the user,
such as mouse, keyboard, and
bar code readers.
X input devices are of several types: keyboard, pointing device, and
many others.  The core server has support for extension devices as
described by the X Input Extension document; the interfaces used by
that extension are described elsewhere.  The core devices are actually
implemented as two collections of devices, the mouse is a ButtonDevice,
a ValuatorDevice and a PtrFeedbackDevice while the keyboard is a KeyDevice,
a FocusDevice and a KbdFeedbackDevice.  Each part implements a portion of
the functionality of the device.  This abstraction is hidden from view for
core devices by DIX.
</para>
<para>
You, the DDX programmer, are
responsible for some of the routines in this section.
Others are DIX routines that you should call to do the things you need to do in these DDX routines.
Pay attention to which is which.
</para>
<section>
  <title>Input Device Data Structures</title>
<para>
DIX keeps a global directory of devices in a central data structure
called InputInfo.
For each device there is a device structure called a DeviceRec.
DIX can locate any DeviceRec through InputInfo.
In addition, it has a special pointer to identify the main pointing device
and a special pointer to identify the main keyboard.
</para>
<para>
The DeviceRec (Xserver/include/input.h) is a device-independent
structure that contains the state of an input device.
A DevicePtr is simply a pointer to a DeviceRec.
</para>
<para>
An xEvent describes an event the server reports to a client.
Defined in Xproto.h, it is a huge struct of union of structs that have fields for
all kinds of events.
All of the variants overlap, so that the struct is actually very small in memory.
</para>
</section>
<section>
  <title>Processing Events</title>
<para>
The main DDX input interface is the following routine:
<blockquote><programlisting>

        void ProcessInputEvents()
</programlisting></blockquote>
You must write this routine to deliver input events from the user.
DIX calls it when input is pending (see next section), and possibly
even when it is not.
You should write it to get events from each device and deliver
the events to DIX.
To deliver the events to DIX, DDX should call the following
routine:
<blockquote><programlisting>

        void DevicePtr->processInputProc(pEvent, device, count)
                    xEventPtr events;
                    DeviceIntPtr device;
                    int count;
</programlisting></blockquote>
This is the "input proc" for the device, a DIX procedure.
DIX will fill in this procedure pointer to one of its own routines by
the time ProcessInputEvents() is called the first time.
Call this input proc routine as many times as needed to
deliver as many events as should be delivered.
DIX will buffer them up and send them out as needed.  Count is set
to the number of event records which make up one atomic device event and
is always 1 for the core devices (see the X Input Extension for descriptions
of devices which may use count &#x3E; 1).
</para>
<para>
For example, your ProcessInputEvents() routine might check the mouse and the
keyboard.
If the keyboard had several keystrokes queued up, it could just call
the keyboard's processInputProc as many times as needed to flush its internal queue.
</para>
<para>
event is an xEvent struct you pass to the input proc.
When the input proc returns, it is finished with the event rec, and you can fill
in new values and call the input proc again with it.
</para>
<para>
You should deliver the events in the same order that they were generated.
</para>
<para>
For keyboard and pointing devices the xEvent variant should be keyButtonPointer.
Fill in the following fields in the xEvent record:
<itemizedlist>

<listitem><para>type - is one of the following: KeyPress, KeyRelease, ButtonPress,
                                        ButtonRelease, or MotionNotify</para></listitem>
<listitem><para>detail - for KeyPress or KeyRelease fields, this should be the
                                        key number (not the ASCII code); otherwise unused</para></listitem>
<listitem><para>time - is the time that the event happened (32-bits, in milliseconds, arbitrary origin)</para></listitem>
<listitem><para>rootX - is the x coordinate of cursor</para></listitem>
<listitem><para>rootY - is the y coordinate of cursor</para></listitem>

</itemizedlist>
The rest of the fields are filled in by DIX.
</para>
<para>
The time stamp is maintained by your code in the DDX layer, and it is your responsibility to
stamp all events correctly.
</para>
<para>
The x and y coordinates of the pointing device and the time must be filled in for all event types
including keyboard events.
</para>
<para>
The pointing device must report all button press and release events.
In addition, it should report a MotionNotify event every time it gets called
if the pointing device has moved since the last notify.
Intermediate pointing device moves are stored in a special GetMotionEvents buffer,
because most client programs are not interested in them.
</para>
<para>
There are quite a collection of sample implementations of this routine,
one for each supported device.
</para>
</section>
<section>
<title>Telling DIX When Input is Pending</title>
<para>
In the server's dispatch loop, DIX checks to see
if there is any device input pending whenever WaitForSomething() returns.
If the check says that input is pending, DIX calls the
DDX routine ProcessInputEvents().
</para>
<para>
This check for pending input must be very quick; a procedure call
is too slow.
The code that does the check is a hardwired IF
statement in DIX code that simply compares the values
pointed to by two pointers.
If the values are different, then it assumes that input is pending and
ProcessInputEvents() is called by DIX.
</para>
<para>
You must pass pointers to DIX to tell it what values to compare.
The following procedure
is used to set these pointers:
<blockquote><programlisting>

        void SetInputCheck(p1, p2)
                long *p1, *p2;
</programlisting></blockquote>
You should call it sometime during initialization to indicate to DIX the
correct locations to check.
You should
pay special attention to the size of what they actually point to,
because the locations are assumed to be longs.
</para>
<para>
These two pointers are initialized by DIX
to point to arbitrary values that
are different.
In other words, if you forget to call this routine during initialization,
the worst thing that will happen is that
ProcessInputEvents will be called when
there are no events to process.
</para>
<para>
p1 and p2 might
point at the head and tail of some shared
memory queue.
Another use would be to have one point at a constant 0, with the
other pointing at some mask containing 1s
for each input device that has
something pending.
</para>
<para>
The DDX layer of the sample server calls SetInputCheck()
once when the
server's private internal queue is initialized.
It passes pointers to the queue's head and tail.  See Xserver/mi/mieq.c.
</para>
<para>
<blockquote><programlisting>
        int TimeSinceLastInputEvent()
</programlisting></blockquote>
DDX must time stamp all hardware input
events.  But DIX sometimes needs to know the
time and the OS layer needs to know the time since the last hardware
input event in
order for the screen saver to work.   TimeSinceLastInputEvent() returns
the this time in milliseconds.
</para>
</section>
<section>
  <title>Controlling Input Devices</title>
<para>
You must write four routines to do various device-specific
things with the keyboard and pointing device.
They can have any name you wish because
you pass the procedure pointers to DIX routines.
</para>
<para>
<blockquote><programlisting>

        int pInternalDevice->valuator->GetMotionProc(pdevice, coords, start, stop, pScreen)
                DeviceIntPtr pdevice;
                xTimecoord * coords;
                unsigned long start;
                unsigned long stop;
                ScreenPtr pScreen;
</programlisting></blockquote>
You write this DDX routine to fill in coords with all the motion
events that have times (32-bit count of milliseconds) between time
start and time stop.  It should return the number of motion events
returned.  If there is no motion events support, this routine should
do nothing and return zero.  The maximum number of coords to return is
set in InitPointerDeviceStruct(), below.
</para>
<para>
When the user drags the pointing device, the cursor position
theoretically sweeps through an infinite number of points.  Normally,
a client that is concerned with points other than the starting and
ending points will receive a pointer-move event only as often as the
server generates them. (Move events do not queue up; each new one
replaces the last in the queue.)  A server, if desired, can implement
a scheme to save these intermediate events in a motion buffer.  A
client application, like a paint program, may then request that these
events be delivered to it through the GetMotionProc routine.
</para>
<para>
<blockquote><programlisting>

        void pInternalDevice->bell->BellProc(percent, pDevice, ctrl, unknown)
                int percent;
                DeviceIntPtr pDevice;
                pointer ctrl;
                int class;
</programlisting></blockquote>
You need to write this routine to ring the bell on the keyboard.
loud is a number from 0 to 100, with 100 being the loudest.
Class is either BellFeedbackClass or KbdFeedbackClass (from XI.h).
</para>
<para>
<blockquote><programlisting>

        void pInternalDevice->somedevice->CtrlProc(device, ctrl)
                DevicePtr device;
                SomethingCtrl *ctrl;

</programlisting></blockquote>
You write two versions of this procedure, one for the keyboard and one for the pointing device.
DIX calls it to inform DDX when a client has requested changes in the current
settings for the particular device.
For a keyboard, this might be the repeat threshold and rate.
For a pointing device, this might be a scaling factor (coarse or fine) for position reporting.
See input.h for the ctrl structures.
</para>
</section>
<section>
  <title>Input Initialization</title>
<para>
Input initialization is a bit complicated.
It all starts with InitInput(), a routine that you write to call
AddInputDevice() twice
(once for pointing device and once for keyboard.)
</para>
<para>
When you Add the devices, a routine you supply for each device
gets called to initialize them.
Your individual initialize routines must call InitKeyboardDeviceStruct()
or InitPointerDeviceStruct(), depending upon which it is.
In other words, you indicate twice that the keyboard is the keyboard and
the pointer is the pointer.
</para>
<para>
<blockquote><programlisting>

        void InitInput(argc, argv)
            int argc;
            char **argv;
</programlisting></blockquote>
InitInput is a DDX routine you must write to initialize the
input subsystem in DDX.
It must call AddInputDevice() for each device that might generate events.
</para>
<para>
<blockquote><programlisting>

        DevicePtr AddInputDevice(deviceProc, autoStart)
                DeviceProc deviceProc;
                Bool autoStart;
</programlisting></blockquote>
AddInputDevice is a DIX routine you call to create a device object.
deviceProc is a DDX routine that is called by DIX to do various operations.
AutoStart should be TRUE for devices that need to be turned on at
initialization time with a special call, as opposed to waiting for some
client application to
turn them on.
This routine returns NULL if sufficient memory cannot be allocated to
install the device.
</para>
<para>
Note also that except for the main keyboard and pointing device,
an extension is needed to provide for a client interface to a device.
</para>
<para>
The following DIX
procedures return the specified DevicePtr. They may or may not be useful
to DDX implementors.
</para>
<para>
<blockquote><programlisting>

        DevicePtr LookupKeyboardDevice()
</programlisting></blockquote>
LookupKeyboardDevice returns pointer for current main keyboard device.
</para>
<para>
<blockquote><programlisting>

        DevicePtr LookupPointerDevice()
</programlisting></blockquote>
LookupPointerDevice returns pointer for current main pointing device.
</para>
<para>
A DeviceProc (the kind passed to AddInputDevice()) in the following form:
<blockquote><programlisting>

        Bool pInternalDevice->DeviceProc(device, action);
                DeviceIntPtr device;
                int action;
</programlisting></blockquote>
You must write a DeviceProc for each device.
device points to the device record.
action tells what action to take;
it will be one of  these defined constants  (defined in input.h):
<itemizedlist>
<listitem><para>
DEVICE_INIT -
At DEVICE_INIT time, the device should initialize itself by calling
InitPointerDeviceStruct(), InitKeyboardDeviceStruct(), or a similar
routine (see below)
and "opening" the device if necessary.
If you return a non-zero (i.e., != Success) value from the DEVICE_INIT
call, that device will be considered unavailable. If either the main keyboard
or main pointing device cannot be initialized, the DIX code will refuse
to continue booting up.</para></listitem>
<listitem><para>
DEVICE_ON - If the DeviceProc is called with DEVICE_ON, then it is
allowed to start
putting events into the client stream by calling through the ProcessInputProc
in the device.</para></listitem>
<listitem><para>
DEVICE_OFF - If the DeviceProc is called with DEVICE_OFF, no further
events from that
device should be given to the DIX layer.
The device will appear to be dead to the user.</para></listitem>
<listitem><para>
DEVICE_CLOSE - At DEVICE_CLOSE (terminate or reset) time, the device should
be totally closed down.</para></listitem>
</itemizedlist>
</para>
<para>
<blockquote><programlisting>

        void InitPointerDeviceStruct(device, map, mapLength,
                        GetMotionEvents, ControlProc, numMotionEvents)
                DevicePtr device;
                CARD8 *map;
                int mapLength;
                ValuatorMotionProcPtr ControlProc;
                PtrCtrlProcPtr GetMotionEvents;
                int numMotionEvents;
</programlisting></blockquote>
InitPointerDeviceStruct is a DIX routine you call at DEVICE_INIT time to declare
some operating routines and data structures for a pointing device.
map and mapLength are as described in the X Window
System protocol specification.
ControlProc and GetMotionEvents are DDX routines, see above.
</para>
<para>
numMotionEvents is for the motion-buffer-size for the GetMotionEvents
request.
A typical length for a motion buffer would be 100 events.
A server that does not implement this capability should set
numMotionEvents to zero.
</para>
<para>
<blockquote><programlisting>

        void InitKeyboardDeviceStruct(device, pKeySyms, pModifiers, Bell, ControlProc)
                DevicePtr device;
                KeySymsPtr pKeySyms;
                CARD8 *pModifiers;
                BellProcPtr Bell;
                KbdCtrlProcPtr ControlProc;

</programlisting></blockquote>
You call this DIX routine when a keyboard device is initialized and
its device procedure is called with
DEVICE_INIT.
The formats of the keysyms and modifier maps are defined in
Xserver/include/input.h.
They describe the layout of keys on the keyboards, and the glyphs
associated with them.  ( See the next section for information on
setting up the modifier map and the keysym map.)
ControlProc and Bell are DDX routines, see above.
</para>
</section>
<section>
  <title>Keyboard Mapping and Keycodes</title>
<para>
When you send a keyboard event, you send a report that a given key has
either been pressed or has been released.  There must be a keycode for
each key that identifies the key; the keycode-to-key mapping can be
any mapping you desire, because you specify the mapping in a table you
set up for DIX.  However, you are restricted by the protocol
specification to keycode values in the range 8 to 255 inclusive.
</para>
<para>
The keycode mapping information that you set up consists of the following:
<itemizedlist>
<listitem><para>
A minimum and maximum keycode number</para></listitem>
<listitem><para>
An array of sets of keysyms for each key, that is of length
maxkeycode - minkeycode + 1.
Each element of this array is a list of codes for symbols that are on that key.
There is no limit to the number of symbols that can be on a key.</para></listitem>
</itemizedlist>
Once the map is set up, DIX keeps and
maintains the client's changes to it.
</para>
<para>
The X protocol defines standard names to indicate the symbol(s)
printed on each keycap. (See X11/keysym.h)
</para>
</section>
</section>
<section>
<title>Screens</title>
<para>
Different computer graphics
displays have different capabilities.
Some are simple monochrome
frame buffers that are just lying
there in memory, waiting to be written into.
Others are color displays with many bits per pixel using some color lookup table.
Still others have high-speed graphic processors that prefer to do all of the work
themselves,
including maintaining their own high-level, graphic data structures.
</para>
<section>
  <title>Screen Hardware Requirements</title>
<para>
The only requirement on screens is that you be able to both read
and write locations in the frame buffer.
All screens must have a depth of 32 or less (unless you use
an X extension to allow a greater depth).
All screens must fit into one of the classes listed in the section
in this document on Visuals and Depths.
</para>
<para>
X uses the pixel as its fundamental unit of distance on the screen.
Therefore, most programs will measure everything in pixels.</para>
<para>
The sample server assumes square pixels.
Serious WYSIWYG (what you see is what you get) applications for
publishing and drawing programs will adjust for
different screen resolutions automatically.
Considerable work
is involved in compensating for non-square pixels (a bit in the DDX
code for the sample server but quite a bit in the client applications).</para>
</section>
<section>
  <title>Data Structures</title>
<para>
X supports multiple screens that are connected to the same
server.  Therefore, all the per-screen information is bundled into one data
structure of attributes and procedures, which is the ScreenRec (see
Xserver/include/scrnintstr.h).
The procedure entry points in a ScreenRec operate on
regions, colormaps, cursors, and fonts, because these resources
can differ in format from one screen to another.</para>
<para>
Windows are areas on the screen that can be drawn into by graphic
routines.  "Pixmaps" are off-screen graphic areas that can be drawn
into.  They are both considered drawables and are described in the
section on Drawables.  All graphic operations work on drawables, and
operations are available to copy patches from one drawable to another.</para>
<para>
The pixel image data in all drawables is in a format that is private
to DDX.  In fact, each instance of a drawable is associated with a
given screen.  Presumably, the pixel image data for pixmaps is chosen
to be conveniently understood by the hardware.  All screens in a
single server must be able to handle all pixmaps depths declared in
the connection setup information.</para>
<para>
Pixmap images are transferred to the server in one of two ways:
XYPixmap or ZPimap.  XYPixmaps are a series of bitmaps, one for each
bit plane of the image, using the bitmap padding rules from the
connection setup.  ZPixmaps are a series of bits, nibbles, bytes or
words, one for each pixel, using the format rules (padding and so on)
for the appropriate depth.</para>
<para>
All screens in a given server must agree on a set of pixmap image
formats (PixmapFormat) to support (depth, number of bits per pixel,
etc.).</para>
<para>
There is no color interpretation of bits in the pixmap.  Pixmaps
do not contain pixel values.  The interpretation is made only when
the bits are transferred onto the screen.</para>
<para>
The screenInfo structure (in scrnintstr.h) is a global data structure
that has a pointer to an array of ScreenRecs, one for each screen on
the server.  (These constitute the one and only description of each
screen in the server.)  Each screen has an identifying index (0, 1, 2, ...).
In addition, the screenInfo struct contains global server-wide
details, such as the bit- and byte- order in all bit images, and the
list of pixmap image formats that are supported.  The X protocol
insists that these must be the same for all screens on the server.</para>
</section>
<section>
  <title>Output Initialization</title>
<para>
<blockquote><programlisting>

        InitOutput(pScreenInfo, argc, argv)
                ScreenInfo *pScreenInfo;
                int argc;
                char **argv;
</programlisting></blockquote>
Upon initialization, your DDX routine InitOutput() is called by DIX.
It is passed a pointer to screenInfo to initialize.  It is also passed
the argc and argv from main() for your server for the command-line
arguments.  These arguments may indicate what or how many screen
device(s) to use or in what way to use them.  For instance, your
server command line may allow a "-D" flag followed by the name of the
screen device to use.</para>
<para>
Your InitOutput() routine should initialize each screen you wish to
use by calling AddScreen(), and then it should initialize the pixmap
formats that you support by storing values directly into the
screenInfo data structure.  You should also set certain
implementation-dependent numbers and procedures in your screenInfo,
which determines the pixmap and scanline padding rules for all screens
in the server.</para>
<para>
<blockquote><programlisting>

        int AddScreen(scrInitProc, argc, argv)
                Bool (*scrInitProc)();
                int argc;
                char **argv;
</programlisting></blockquote>
You should call AddScreen(), a DIX procedure, in InitOutput() once for
each screen to add it to the screenInfo database.  The first argument
is an initialization procedure for the screen that you supply.  The
second and third are the argc and argv from main().  It returns the
screen number of the screen installed, or -1 if there is either
insufficient memory to add the screen, or (*scrInitProc) returned
FALSE.</para>
<para>
The scrInitProc should be of the following form:
<blockquote><programlisting>

        Bool scrInitProc(pScreen, argc, argv)
                ScreenPtr pScreen;
                int argc;
                char **argv;
</programlisting></blockquote>
pScreen is the pointer to the screen's new ScreenRec. argc and argv
are as before.  Your screen initialize procedure should return TRUE
upon success or FALSE if the screen cannot be initialized (for
 instance, if the screen hardware does not exist on this machine).</para>
<para>
This procedure must determine what actual device it is supposed to initialize.
If you have a different procedure for each screen, then it is no problem.
If you have the same procedure for multiple screens, it may have trouble
figuring out which screen to initialize each time around, especially if
InitOutput() does not initialize all of the screens.
It is probably easiest to have one procedure for each screen.</para>
<para>
The initialization procedure should fill in all the screen procedures
for that screen (windowing functions, region functions, etc.) and certain
screen attributes for that screen.</para>
</section>
<section>
  <title>Region Routines in the ScreenRec</title>
<para>
A region is a dynamically allocated data structure that describes an
irregularly shaped piece of real estate in XY pixel space.  You can
think of it as a set of pixels on the screen to be operated upon with
set operations such as AND and OR.</para>
<para>
A region is frequently implemented as a list of rectangles or bitmaps
that enclose the selected pixels.  Region operators control the
"clipping policy," or the operations that work on regions.  (The
sample server uses YX-banded rectangles.  Unless you have something
already implemented for your graphics system, you should keep that
implementation.)  The procedure pointers to the region operators are
located in the ScreenRec data structure.  The definition of a region
can be found in the file Xserver/include/regionstr.h.  The region code
is found in Xserver/mi/miregion.c.  DDX implementations using other
region formats will need to supply different versions of the region
operators.</para>
<para>
Since the list of rectangles is unbounded in size, part of the region
data structure is usually a large, dynamically allocated chunk of
memory.  As your region operators calculate logical combinations of
regions, these blocks may need to be reallocated by your region
software.  For instance, in the sample server, a RegionRec has some
header information and a pointer to a dynamically allocated rectangle
list.  Periodically, the rectangle list needs to be expanded with
realloc(), whereupon the new pointer is remembered in the RegionRec.</para>
<para>
Most of the region operations come in two forms: a function pointer in
the Screen structure, and a macro.  The server can be compiled so that
the macros make direct calls to the appropriate functions (instead of
indirecting through a screen function pointer), or it can be compiled
so that the macros are identical to the function pointer forms.
Making direct calls is faster on many architectures.</para>
<para>
<blockquote><programlisting>

        RegionPtr pScreen->RegionCreate( rect, size)
                BoxPtr rect;
                int size;

        macro: RegionPtr RegionCreate(rect, size)

</programlisting></blockquote>
RegionCreate creates a region that describes ONE rectangle.  The
caller can avoid unnecessary reallocation and copying by declaring the
probable maximum number of rectangles that this region will need to
describe itself.  Your region routines, though, cannot fail just
because the region grows beyond this size.  The caller of this routine
can pass almost anything as the size; the value is merely a good guess
as to the maximum size until it is proven wrong by subsequent use.
Your region procedures are then on their own in estimating how big the
region will get.  Your implementation might ignore size, if
applicable.</para>
<para>
<blockquote><programlisting>

        void pScreen->RegionInit (pRegion, rect, size)
                RegionPtr       pRegion;
                BoxPtr          rect;
                int             size;

        macro: RegionInit(pRegion, rect, size)

</programlisting></blockquote>
Given an existing raw region structure (such as an local variable), this
routine fills in the appropriate fields to make this region as usable as
one returned from RegionCreate.  This avoids the additional dynamic memory
allocation overhead for the region structure itself.
</para>
<para>
<blockquote><programlisting>

        Bool pScreen->RegionCopy(dstrgn, srcrgn)
                RegionPtr dstrgn, srcrgn;

        macro: Bool RegionCopy(dstrgn, srcrgn)

</programlisting></blockquote>
RegionCopy copies the description of one region, srcrgn, to another
already-created region,
dstrgn; returning TRUE if the copy succeeded, and FALSE otherwise.</para>
<para>
<blockquote><programlisting>

        void pScreen->RegionDestroy( pRegion)
                RegionPtr pRegion;

        macro: RegionDestroy(pRegion)

</programlisting></blockquote>
RegionDestroy destroys a region and frees all allocated memory.</para>
<para>
<blockquote><programlisting>

        void pScreen->RegionUninit (pRegion)
                RegionPtr pRegion;

        macro: RegionUninit(pRegion)

</programlisting></blockquote>
Frees everything except the region structure itself, useful when the
region was originally passed to RegionInit instead of received from
RegionCreate.  When this call returns, pRegion must not be reused until
it has been RegionInit'ed again.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->Intersect(newReg, reg1, reg2)
                RegionPtr newReg, reg1, reg2;

        macro: Bool RegionIntersect(newReg, reg1, reg2)

        Bool  pScreen->Union(newReg, reg1, reg2)
                RegionPtr newReg, reg1, reg2;

        macro: Bool RegionUnion(newReg, reg1, reg2)

        Bool  pScreen->Subtract(newReg, regMinuend, regSubtrahend)
                RegionPtr newReg, regMinuend, regSubtrahend;

        macro: Bool RegionUnion(newReg, regMinuend, regSubtrahend)

        Bool pScreen->Inverse(newReg, pReg,  pBox)
                RegionPtr newReg, pReg;
                BoxPtr pBox;

        macro: Bool RegionInverse(newReg, pReg,  pBox)

</programlisting></blockquote>
The above four calls all do basic logical operations on regions.  They
set the new region (which already exists) to describe the logical
intersection, union, set difference, or inverse of the region(s) that
were passed in.  Your routines must be able to handle a situation
where the newReg is the same region as one of the other region
arguments.</para>
<para>
The subtract function removes the Subtrahend from the Minuend and
puts the result in newReg.</para>
<para>
The inverse function returns a region that is the pBox minus the
region passed in.  (A true "inverse" would make a region that extends
to infinity in all directions but has holes in the middle.)  It is
undefined for situations where the region extends beyond the box.</para>
<para>
Each routine must return the value TRUE for success.</para>
<para>
<blockquote><programlisting>

        void pScreen->RegionReset(pRegion, pBox)
                RegionPtr pRegion;
                BoxPtr pBox;

        macro: RegionReset(pRegion, pBox)

</programlisting></blockquote>
RegionReset sets the region to describe
one rectangle and reallocates it to a size of one rectangle, if applicable.</para>
<para>
<blockquote><programlisting>

        void  pScreen->TranslateRegion(pRegion, x, y)
                RegionPtr pRegion;
                int x, y;

        macro: RegionTranslate(pRegion, x, y)

</programlisting></blockquote>
TranslateRegion simply moves a region +x in the x direction and +y in the y
direction.</para>
<para>
<blockquote><programlisting>

        int  pScreen->RectIn(pRegion, pBox)
                RegionPtr pRegion;
                BoxPtr pBox;

        macro: int RegionContainsRect(pRegion, pBox)

</programlisting></blockquote>
RectIn returns one of the defined constants rgnIN, rgnOUT, or rgnPART,
depending upon whether the box is entirely inside the region, entirely
outside of the region, or partly in and partly out of the region.
These constants are defined in Xserver/include/region.h.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->PointInRegion(pRegion, x, y, pBox)
                RegionPtr pRegion;
                int x, y;
                BoxPtr pBox;

        macro: Bool RegionContainsPoint(pRegion, x, y, pBox)

</programlisting></blockquote>
PointInRegion returns true if the point x, y is in the region.  In
addition, it fills the rectangle pBox with coordinates of a rectangle
that is entirely inside of pRegion and encloses the point.  In the mi
implementation, it is the largest such rectangle.  (Due to the sample
server implementation, this comes cheaply.)</para>
<para>
This routine used by DIX when tracking the pointing device and
deciding whether to report mouse events or change the cursor.  For
instance, DIX needs to change the cursor when it moves from one window
to another.  Due to overlapping windows, the shape to check may be
irregular.  A PointInRegion() call for every pointing device movement
may be too expensive.  The pBox is a kind of wake-up box; DIX need not
call PointInRegion() again until the cursor wanders outside of the
returned box.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->RegionNotEmpty(pRegion)
                RegionPtr pRegion;

        macro: Bool RegionNotEmpty(pRegion)

</programlisting></blockquote>
RegionNotEmpty is a boolean function that returns
true or false depending upon whether the region encloses any pixels.</para>
<para>
<blockquote><programlisting>

        void pScreen->RegionEmpty(pRegion)
                RegionPtr pRegion;

        macro: RegionEmpty(pRegion)

</programlisting></blockquote>
RegionEmpty sets the region to be empty.</para>
<para>
<blockquote><programlisting>

        BoxPtr pScreen->RegionExtents(pRegion)
                RegionPtr pRegion;

        macro: RegionExtents(pRegion)

</programlisting></blockquote>
RegionExtents returns a rectangle that is the smallest
possible superset of the entire region.
The caller will not modify this rectangle, so it can be the one
in your region struct.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->RegionAppend (pDstRgn, pRegion)
                RegionPtr pDstRgn;
                RegionPtr pRegion;

        macro: Bool RegionAppend(pDstRgn, pRegion)

        Bool pScreen->RegionValidate (pRegion, pOverlap)
                RegionPtr pRegion;
                Bool *pOverlap;

        macro: Bool RegionValidate(pRegion, pOverlap)

</programlisting></blockquote>
These functions provide an optimization for clip list generation and
must be used in conjunction.  The combined effect is to produce the
union of a collection of regions, by using RegionAppend several times,
and finally calling RegionValidate which takes the intermediate
representation (which needn't be a valid region) and produces the
desired union.  pOverlap is set to TRUE if any of the original
regions overlap; FALSE otherwise.</para>
<para>
<blockquote><programlisting>

        RegionPtr pScreen->BitmapToRegion (pPixmap)
                PixmapPtr pPixmap;

        macro: RegionPtr BitmapToRegion(pScreen, pPixmap)

</programlisting></blockquote>
Given a depth-1 pixmap, this routine must create a valid region which
includes all the areas of the pixmap filled with 1's and excludes the
areas filled with 0's.  This routine returns NULL if out of memory.</para>
<para>
<blockquote><programlisting>

        RegionPtr pScreen->RectsToRegion (nrects, pRects, ordering)
                int nrects;
                xRectangle *pRects;
                int ordering;

        macro: RegionPtr RegionFromRects(nrects, pRects, ordering)

</programlisting></blockquote>
Given a client-supplied list of rectangles, produces a region which includes
the union of all the rectangles.  Ordering may be used as a hint which
describes how the rectangles are sorted.  As the hint is provided by a
client, it must not be required to be correct, but the results when it is
not correct are not defined (core dump is not an option here).</para>
<para>
<blockquote><programlisting>

        void pScreen->SendGraphicsExpose(client,pRegion,drawable,major,minor)
                ClientPtr client;
                RegionPtr pRegion;
                XID drawable;
                int major;
                int minor;

</programlisting></blockquote>
SendGraphicsExpose dispatches a list of GraphicsExposure events which
span the region to the specified client.  If the region is empty, or
a NULL pointer, a NoExpose event is sent instead.</para>
</section>
<section>
  <title>Cursor Routines for a Screen</title>
<para>
A cursor is the visual form tied to the pointing device.  The default
cursor is an "X" shape, but the cursor can have any shape.  When a
client creates a window, it declares what shape the cursor will be
when it strays into that window on the screen.</para>
<para>
For each possible shape the cursor assumes, there is a CursorRec data
structure.  This data structure contains a pointer to a CursorBits
data structure which contains a bitmap for the image of the cursor and
a bitmap for a mask behind the cursor, in addition, the CursorRec data
structure contains foreground and background colors for the cursor.
The CursorBits data structure is shared among multiple CursorRec
structures which use the same font and glyph to describe both source
and mask.  The cursor image is applied to the screen by applying the
mask first, clearing 1 bits in its form to the background color, and
then overwriting on the source image, in the foreground color.  (One
bits of the source image that fall on top of zero bits of the mask
image are undefined.)  This way, a cursor can have transparent parts,
and opaque parts in two colors.  X allows any cursor size, but some
hardware cursor schemes allow a maximum of N pixels by M pixels.
Therefore, you are allowed to transform the cursor to a smaller size,
but be sure to include the hot-spot.</para>
<para>
CursorBits in Xserver/include/cursorstr.h is a device-independent
structure containing a device-independent representation of the bits
for the source and mask.  (This is possible because the bitmap
representation is the same for all screens.)</para>
<para>
When a cursor is created, it is "realized" for each screen.  At
realization time, each screen has the chance to convert the bits into
some other representation that may be more convenient (for instance,
putting the cursor into off-screen memory) and set up its
device-private area in either the CursorRec data structure or
CursorBits data structure as appropriate to possibly point to whatever
data structures are needed.  It is more memory-conservative to share
realizations by using the CursorBits private field, but this makes the
assumption that the realization is independent of the colors used
(which is typically true).  For instance, the following are the device
private entries for a particular screen and cursor:
<blockquote><programlisting>

        pCursor->devPriv[pScreen->myNum]
        pCursor->bits->devPriv[pScreen->myNum]

</programlisting></blockquote>
This is done because the change from one cursor shape to another must
be fast and responsive; the cursor image should be able to flutter as
fast as the user moves it across the screen.</para>
<para>
You must implement the following routines for your hardware:
<blockquote><programlisting>

        Bool pScreen->RealizeCursor( pScr, pCurs)
                ScreenPtr pScr;
                CursorPtr pCurs;

        Bool pScreen->UnrealizeCursor( pScr, pCurs)
                ScreenPtr pScr;
                CursorPtr pCurs;

</programlisting></blockquote>
</para>
<para>
RealizeCursor and UnrealizeCursor should realize (allocate and
calculate all data needed) and unrealize (free the dynamically
allocated data) a given cursor when DIX needs them.  They are called
whenever a device-independent cursor is created or destroyed.  The
source and mask bits pointed to by fields in pCurs are undefined for
bits beyond the right edge of the cursor.  This is so because the bits
are in Bitmap format, which may have pad bits on the right edge.  You
should inhibit UnrealizeCursor() if the cursor is currently in use;
this happens when the system is reset.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->DisplayCursor( pScr, pCurs)
                ScreenPtr pScr;
                CursorPtr pCurs;

</programlisting></blockquote>
DisplayCursor should change the cursor on the given screen to the one
passed in.  It is called by DIX when the user moves the pointing
device into a different window with a different cursor.  The hotspot
in the cursor should be aligned with the current cursor position.</para>
<para>
<blockquote><programlisting>

        void pScreen->RecolorCursor( pScr, pCurs, displayed)
                ScreenPtr pScr;
                CursorPtr pCurs;
                Bool displayed;
</programlisting></blockquote>
RecolorCursor notifies DDX that the colors in pCurs have changed and
indicates whether this is the cursor currently being displayed.  If it
is, the cursor hardware state may have to be updated.  Whether
displayed or not, state created at RealizeCursor time may have to be
updated.  A generic version, miRecolorCursor, may be used that
does an unrealize, a realize, and possibly a display (in micursor.c);
however this constrains UnrealizeCursor and RealizeCursor to always return
TRUE as no error indication is returned here.</para>
<para>
<blockquote><programlisting>

        void pScreen->ConstrainCursor( pScr, pBox)
                ScreenPtr pScr;
                BoxPtr pBox;

</programlisting></blockquote>
ConstrainCursor should cause the cursor to restrict its motion to the
rectangle pBox.  DIX code is capable of enforcing this constraint by
forcefully moving the cursor if it strays out of the rectangle, but
ConstrainCursor offers a way to send a hint to the driver or hardware
if such support is available.  This can prevent the cursor from
wandering out of the box, then jumping back, as DIX forces it back.</para>
<para>
<blockquote><programlisting>

        void pScreen->PointerNonInterestBox( pScr, pBox)
                ScreenPtr pScr;
                BoxPtr pBox;

</programlisting></blockquote>
PointerNonInterestBox is DIX's way of telling the pointing device code
not to report motion events while the cursor is inside a given
rectangle on the given screen.  It is optional and, if not
implemented, it should do nothing.  This routine is called only when
the client has declared that it is not interested in motion events in
a given window.  The rectangle you get may be a subset of that window.
It saves DIX code the time required to discard uninteresting mouse
motion events.  This is only a hint, which may speed performance.
Nothing in DIX currently calls PointerNonInterestBox.</para>
<para>
<blockquote><programlisting>

        void pScreen->CursorLimits( pScr, pCurs, pHotBox, pTopLeftBox)
                ScreenPtr pScr;
                CursorPtr pCurs;
                BoxPtr pHotBox;
                BoxPtr pTopLeftBox;     /* return value */

</programlisting></blockquote>
CursorLimits should calculate the box that the cursor hot spot is
physically capable of moving within, as a function of the screen pScr,
the device-independent cursor pCurs, and a box that DIX hypothetically
would want the hot spot confined within, pHotBox.  This routine is for
informing DIX only; it alters no state within DDX.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->SetCursorPosition( pScr, newx, newy, generateEvent)
                ScreenPtr pScr;
                int newx;
                int newy;
                Bool generateEvent;

</programlisting></blockquote>
SetCursorPosition should artificially move the cursor as though the
user had jerked the pointing device very quickly.  This is called in
response to the WarpPointer request from the client, and at other
times.  If generateEvent is True, the device should decide whether or
not to call ProcessInputEvents() and then it must call
DevicePtr->processInputProc.  Its effects are, of course, limited in
value for absolute pointing devices such as a tablet.</para>
<para>
<blockquote><programlisting>

        void NewCurrentScreen(newScreen, x, y)
            ScreenPtr newScreen;
            int x,y;

</programlisting></blockquote>
If your ddx provides some mechanism for the user to magically move the
pointer between multiple screens, you need to inform DIX when this
occurs.  You should call NewCurrentScreen to accomplish this, specifying
the new screen and the new x and y coordinates of the pointer on that screen.</para>
</section>
<section>
  <title>Visuals, Depths and Pixmap Formats for Screens</title>
<para>
The "depth" of a image is the number of bits that are used per pixel to display it.</para>
<para>
The "bits per pixel" of a pixmap image that is sent over the client
byte stream is a number that is either 4, 8, 16, 24 or 32.  It is the
number of bits used per pixel in Z format.  For instance, a pixmap
image that has a depth of six is best sent in Z format as 8 bits per
pixel.</para>
<para>
A "pixmap image format" or a "pixmap format" is a description of the
format of a pixmap image as it is sent over the byte stream.  For each
depth available on a server, there is one and only one pixmap format.
This pixmap image format gives the bits per pixel and the scanline
padding unit. (For instance, are pixel rows padded to bytes, 16-bit
words, or 32-bit words?)</para>
<para>
For each screen, you must decide upon what depth(s) it supports.  You
should only count the number of bits used for the actual image.  Some
displays store additional bits to indicate what window this pixel is
in, how close this object is to a viewer, transparency, and other
data; do not count these bits.</para>
<para>
A "display class" tells whether the display is monochrome or color,
whether there is a lookup table, and how the lookup table works.</para>
<para>
A "visual" is a combination of depth, display class, and a description
of how the pixel values result in a color on the screen.  Each visual
has a set of masks and offsets that are used to separate a pixel value
into its red, green, and blue components and a count of the number of
colormap entries.  Some of these fields are only meaningful when the
class dictates so.  Each visual also has a screen ID telling which
screen it is usable on.  Note that the depth does not imply the number
of map_entries; for instance, a display can have 8 bits per pixel but
only 254 colormap entries for use by applications (the other two being
reserved by hardware for the cursor).</para>
<para>
Each visual is identified by a 32-bit visual ID which the client uses
to choose what visual is desired on a given window.  Clients can be
using more than one visual on the same screen at the same time.</para>
<para>
The class of a display describes how this translation takes place.
There are three ways to do the translation.
<itemizedlist>
<listitem><para>
Pseudo - The pixel value, as a whole, is looked up
in a table of length map_entries to
determine the color to display.</para></listitem>
<listitem><para>
True - The
pixel value is broken up into red, green, and blue fields, each of which
are looked up in separate red, green, and blue lookup tables,
each of length map_entries.</para></listitem>
<listitem><para>
Gray - The pixel value is looked up in a table of length map_entries to
determine a gray level to display.</para></listitem>
</itemizedlist>
</para>
<para>
In addition, the lookup table can be static (resulting colors are fixed for each
pixel value)
or dynamic (lookup entries are under control of the client program).
This leads to a total of six classes:
<itemizedlist>
<listitem><para>
Static Gray - The pixel value (of however many bits) determines directly the
level of gray
that the pixel assumes.</para></listitem>
<listitem><para>
Gray Scale - The pixel value is fed through a lookup table to arrive at the level
of gray to display
for the given pixel.</para></listitem>
<listitem><para>
Static Color - The pixel value is fed through a fixed lookup table that yields the
color to display
for that pixel.</para></listitem>
<listitem><para>
PseudoColor - The whole pixel value is fed through a programmable lookup
table that has one
color (including red, green, and blue intensities) for each possible pixel value,
and that color is displayed.</para></listitem>
<listitem><para>
True Color - Each pixel value consists of one or more bits
that directly determine each primary color intensity after being fed through
a fixed table.</para></listitem>
<listitem><para>
Direct Color - Each pixel value consists of one or more bits for each primary color.
Each primary color value is individually looked up in a table for that primary
color, yielding
an intensity for that primary color.
For each pixel, the red value is looked up in the
red table, the green value in the green table, and
the blue value in the blue table.</para></listitem>
</itemizedlist>
</para>
<para>
Here are some examples:
<itemizedlist>
<listitem><para>
A simple monochrome 1 bit per pixel display is Static Gray.</para></listitem>
<listitem><para>
A display that has 2 bits per pixel for a choice
between the colors of black, white, green and violet is Static Color.</para></listitem>
<listitem><para>
A display that has three bits per pixel, where
each bit turns on or off one of the red, green or
blue guns, is in the True Color class.</para></listitem>
<listitem><para>
If you take the last example and scramble the
correspondence between pixel values and colors
it becomes a Static Color display.</para></listitem>
</itemizedlist></para>
<para>
A display has 8 bits per pixel.  The 8 bits select one entry out of 256 entries
in a lookup table, each entry consisting of 24 bits (8bits each for red, green,
and blue).
The display can show any 256 of 16 million colors on the screen at once.
This is a pseudocolor display.
The client application gets to fill the lookup table in this class of display.</para>
<para>
Imagine the same hardware from the last example.
Your server software allows the user, on the
command line that starts up the server
program,
to fill the lookup table to his liking once and for all.
From then on, the server software would not change the lookup table
until it exits.
For instance, the default might be a lookup table with a reasonable sample of
colors from throughout the color space.
But the user could specify that the table be filled with 256 steps of gray scale
because he knew ahead of time he would be manipulating a lot of black-and-white
scanned photographs
and not very many color things.
Clients would be presented with this unchangeable lookup table.
Although the hardware qualifies as a PseudoColor display,
the facade presented to the X client is that this is a Static Color display.</para>
<para>
You have to decide what kind of display you have or want
to pretend you have.
When you initialize the screen(s), this class value must be set in the
VisualRec data structure along with other display characteristics like the
depth and other numbers.</para>
<para>
The allowable DepthRec's and VisualRec's are pointed to by fields in the ScreenRec.
These are set up when InitOutput() is called; you should malloc() appropriate blocks
or use static variables initialized to the correct values.</para>
</section>
<section>
<title>Colormaps for Screens</title>
<para>
A colormap is a device-independent
mapping between pixel values and colors displayed on the screen.</para>
<para>
Different windows on the same screen can have different
colormaps at the same time.
At any given time, the most recently installed
colormap(s) will be in use in the server
so that its (their) windows' colors will be guaranteed to be correct.
Other windows may be off-color.
Although this may seem to be chaotic, in practice most clients
use the default colormap for the screen.</para>
<para>
The default colormap for a screen is initialized when the screen is initialized.
It always remains in existence and is not owned by any regular client.  It
is owned by client 0 (the server itself).
Many clients will simply use this default colormap for their drawing.
Depending upon the class of the screen, the entries in this colormap may
be modifiable by client applications.</para>
</section>
<section>
  <title>Colormap Routines</title>
<para>
You need to implement the following routines to handle the device-dependent
aspects of color maps.  You will end up placing pointers to these procedures
in your ScreenRec data structure(s).  The sample server implementations of
many of these routines are in fbcmap.c.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->CreateColormap(pColormap)
                ColormapPtr pColormap;

</programlisting></blockquote>
This routine is called by the DIX CreateColormap routine after it has allocated
all the data for the new colormap and just before it returns to the dispatcher.
It is the DDX layer's chance to initialize the colormap, particularly if it is
a static map.  See the following
section for more details on initializing colormaps.
The routine returns FALSE if creation failed, such as due to memory
limitations.
Notice that the colormap has a devPriv field from which you can hang any
colormap specific storage you need.  Since each colormap might need special
information, we attached the field to the colormap and not the visual.</para>
<para>
<blockquote><programlisting>

        void pScreen->DestroyColormap(pColormap)
                ColormapPtr pColormap;

</programlisting></blockquote>
This routine is called by the DIX FreeColormap routine after it has uninstalled
the colormap and notified all interested parties, and before it has freed
any of the colormap storage.
It is the DDX layer's chance to free any data it added to the colormap.</para>
<para>
<blockquote><programlisting>

        void pScreen->InstallColormap(pColormap)
                ColormapPtr pColormap;

</programlisting></blockquote>
InstallColormap should
fill a lookup table on the screen with which the colormap is associated with
the colors in pColormap.
If there is only one hardware lookup table for the screen, then all colors on
the screen may change simultaneously.</para>
<para>
In the more general case of multiple hardware lookup tables,
this may cause some other colormap to be
uninstalled, meaning that windows that subscribed to the colormap
that was uninstalled may end up being off-color.
See the note, below, about uninstalling maps.</para>
<para>
<blockquote><programlisting>

        void pScreen->UninstallColormap(pColormap)
                ColormapPtr pColormap;

</programlisting></blockquote>
UninstallColormap should
remove pColormap from screen pColormap->pScreen.
Some other map, such as the default map if possible,
should be installed in place of pColormap if applicable.
If
pColormap is the default map, do nothing.
If any client has requested ColormapNotify events, the DDX layer must notify the client.
(The routine WalkTree() is
be used to find such windows.  The DIX routines TellNoMap(),
TellNewMap()  and TellGainedMap() are provided to be used as
the procedure parameter to WalkTree.  These procedures are in
Xserver/dix/colormap.c.)</para>
<para>
<blockquote><programlisting>

        int pScreen->ListInstalledColormaps(pScreen, pCmapList)
                ScreenPtr pScreen;
                XID *pCmapList;


</programlisting></blockquote>
ListInstalledColormaps fills the pCmapList in with the resource ids
of the installed maps and returns a count of installed maps.
pCmapList will point to an array of size MaxInstalledMaps that was allocated
by the caller.</para>
<para>
<blockquote><programlisting>

        void pScreen->StoreColors (pmap, ndef, pdefs)
                ColormapPtr pmap;
                int ndef;
                xColorItem *pdefs;

</programlisting></blockquote>
StoreColors changes some of the entries in the colormap pmap.
The number of entries to change are ndef, and pdefs points to the information
describing what to change.
Note that partial changes of entries in the colormap are allowed.
Only the colors
indicated in the flags field of each xColorItem need to be changed.
However, all three color fields will be sent with the proper value for the
benefit of screens that may not be able to set part of a colormap value.
If the screen is a static class, this routine does nothing.
The structure of colormap entries is nontrivial; see colormapst.h
and the definition of xColorItem in Xproto.h for
more details.</para>
<para>
<blockquote><programlisting>

        void pScreen->ResolveColor(pRed, pGreen, pBlue, pVisual)
                unsigned short *pRed, *pGreen, *pBlue;
                VisualPtr pVisual;


</programlisting></blockquote>
Given a requested color, ResolveColor returns the nearest color that this hardware is
capable of displaying on this visual.
In other words, this rounds off each value, in place, to the number of bits
per primary color that your screen can use.
Remember that each screen has one of these routines.
The level of roundoff should be what you would expect from the value
you put in the bits_per_rgb field of the pVisual.</para>
<para>
Each value is an unsigned value ranging from 0 to 65535.
The bits least likely to be used are the lowest ones.</para>
<para>
For example, if you had a pseudocolor display
with any number of bits per pixel
that had a lookup table supplying 6 bits for each color gun
(a total of 256K different colors), you would
round off each value to 6 bits.  Please don't simply truncate these values
to the upper 6 bits, scale the result so that the maximum value seen
by the client will be 65535 for each primary.  This makes color values
more portable between different depth displays (a 6-bit truncated white
will not look white on an 8-bit display).</para>
<section>
<title>Initializing a Colormap</title>
<para>
When a client requests a new colormap and when the server creates the default
colormap, the procedure CreateColormap in the DIX layer is invoked.
That procedure allocates memory for the colormap and related storage such as
the lists of which client owns which pixels.
It then sets a bit, BeingCreated, in the flags field of the ColormapRec
and calls the DDX layer's CreateColormap routine.
This is your chance to initialize the colormap.
If the colormap is static, which you can tell by looking at the class field,
you will want to fill in each color cell to match the hardwares notion of the
color for that pixel.
If the colormap is the default for the screen, which you can tell by looking
at the CM_IsDefault bit in the flags field, you should allocate BlackPixel
and WhitePixel to match the values you set in the pScreen structure.
(Of course, you picked those values to begin with.)</para>
<para>
You can also wait and use AllocColor() to allocate blackPixel
and whitePixel after the default colormap has been created.
If the default colormap is static and you initialized it in
pScreen->CreateColormap, then use can use AllocColor afterwards
to choose pixel values with the closest rgb values to those
desired for blackPixel and whitePixel.
If the default colormap is dynamic and uninitialized, then
the rgb values you request will be obeyed, and AllocColor will
again choose pixel values for you.
These pixel values can then be stored into the screen.</para>
<para>
There are two ways to fill in the colormap.
The simplest way is to use the DIX function AllocColor.
<blockquote><programlisting>

int AllocColor (pmap, pred, pgreen, pblue, pPix, client)
    ColormapPtr         pmap;
    unsigned short      *pred, *pgreen, *pblue;
    Pixel               *pPix;
    int                 client;

</programlisting></blockquote>
This takes three pointers to 16 bit color values and a pointer to a suggested
pixel value.  The pixel value is either an index into one colormap or a
combination of three indices depending on the type of pmap.
If your colormap starts out empty, and you don't deliberately pick the same
value twice, you will always get your suggested pixel.
The truly nervous could check that the value returned in *pPix is the one
AllocColor was called with.
If you don't care which pixel is used, or would like them sequentially
allocated from entry 0, set *pPix to 0.  This will find the first free
pixel and use that.</para>
<para>
AllocColor will take care of all the  bookkeeping  and  will
call StoreColors to get the colormap rgb values initialized.
The hardware colormap will be changed whenever this colormap
is installed.</para>
<para>
If for some reason AllocColor doesn't do what you want, you can do your
own bookkeeping and call StoreColors yourself.  This is much more difficult
and shouldn't be necessary for most devices.</para>
</section>
</section>
<section>
  <title>Fonts for Screens</title>
<para>
A font is a set of bitmaps that depict the symbols in a character set.
Each font is for only one typeface in a given size, in other words,
just one bitmap for each character.  Parallel fonts may be available
in a variety of sizes and variations, including "bold" and "italic."
X supports fonts for 8-bit and 16-bit character codes (for oriental
languages that have more than 256 characters in the font).  Glyphs are
bitmaps for individual characters.</para>
<para>
The source comes with some useful font files in an ASCII, plain-text
format that should be comprehensible on a wide variety of operating
systems.  The text format, referred to as BDF, is a slight extension
of the current Adobe 2.1 Bitmap Distribution Format (Adobe Systems,
Inc.).</para>
<para>
A short paper in PostScript format is included with the sample server
that defines BDF.  It includes helpful pictures, which is why it is
done in PostScript and is not included in this document.</para>
<para>
Your implementation should include some sort of font compiler to read
these files and generate binary files that are directly usable by your
server implementation.  The sample server comes with the source for a
font compiler.</para>
<para>
It is important the font properties contained in the BDF files are
preserved across any font compilation. In particular, copyright
information cannot be casually tossed aside without legal
ramifications. Other properties will be important to some
sophisticated applications.</para>
<para>
All clients get font information from the server.  Therefore, your
server can support any fonts it wants to.  It should probably support
at least the fonts supplied with the X11 tape.  In principle, you can
convert fonts from other sources or dream up your own fonts for use on
your server.</para>
<section>
<title>Portable Compiled Format</title>
<para>
A font compiler is supplied with the sample server.  It has
compile-time switches to convert the BDF files into a portable binary
form, called Portable Compiled Format or PCF.  This allows for an
arbitrary data format inside the file, and by describing the details
of the format in the header of the file, any PCF file can be read by
any PCF reading client.  By selecting the format which matches the
required internal format for your renderer, the PCF reader can avoid
reformatting the data each time it is read in.  The font compiler
should be quite portable.</para>
<para>
The fonts included with the tape are stored in fonts/bdf.  The
font compiler is found in fonts/tools/bdftopcf.</para>
</section>
<section>
  <title>Font Realization</title>
<para>
Each screen configured into the server
has an opportunity at font-load time
to "realize" a font into some internal format if necessary.
This happens every time the font is loaded into memory.</para>
<para>
A font (FontRec in Xserver/include/dixfontstr.h) is
a device-independent structure containing a device-independent
representation of the font.  When a font is created, it is "realized"
for each screen.  At this point, the screen has the chance to convert
the font into some other format.  The DDX layer can also put information
in the devPrivate storage.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->RealizeFont(pScr, pFont)
                ScreenPtr pScr;
                FontPtr pFont;

        Bool pScreen->UnrealizeFont(pScr, pFont)
                ScreenPtr pScr;
                FontPtr pFont;

</programlisting></blockquote>
RealizeFont and UnrealizeFont should calculate and allocate these extra data structures and
dispose of them when no longer needed.
These are called in response to OpenFont and CloseFont requests from
the client.
The sample server implementation is in fbscreen.c (which does very little).</para>
</section>
</section>
<section>
  <title>Other Screen Routines</title>
<para>
You must supply several other screen-specific routines for
your X server implementation.
Some of these are described in other sections:
<itemizedlist>
<listitem><para>
GetImage() is described in the Drawing Primitives section.</para></listitem>
<listitem><para>
GetSpans() is described in the Pixblit routine section.</para></listitem>
<listitem><para>
Several window and pixmap manipulation procedures are
described in the Window section under Drawables.</para></listitem>
<listitem><para>
The CreateGC() routine is described under Graphics Contexts.</para></listitem>
</itemizedlist>
</para>
<para>
<blockquote><programlisting>

        void pScreen->QueryBestSize(kind, pWidth, pHeight)
                int kind;
                unsigned short *pWidth, *pHeight;
                ScreenPtr pScreen;

</programlisting></blockquote>
QueryBestSize() returns the best sizes for cursors, tiles, and stipples
in response to client requests.
kind is one of the defined constants CursorShape, TileShape, or StippleShape
(defined in X.h).
For CursorShape, return the maximum width and
height for cursors that you can handle.
For TileShape and StippleShape, start with the suggested values in pWidth
and pHeight and modify them in place to be optimal values that are
greater than or equal to the suggested values.
The sample server implementation is in Xserver/fb/fbscreen.c.</para>
<para>
<blockquote><programlisting>

        pScreen->SourceValidate(pDrawable, x, y, width, height)
                DrawablePtr pDrawable;
                int x, y, width, height;
                unsigned int subWindowMode;

</programlisting></blockquote>
SourceValidate should be called by any primitive that reads from pDrawable.
If you know that
you will never need SourceValidate, you can avoid this check.  Currently,
SourceValidate is used by the mi software cursor code to remove the cursor
from the screen when the source rectangle overlaps the cursor position.
x,y,width,height describe the source rectangle (source relative, that is)
for the copy operation.  subWindowMode comes from the GC or source Picture.
</para>
<para>
<blockquote><programlisting>

        Bool pScreen->SaveScreen(pScreen, on)
                ScreenPtr pScreen;
                int on;

</programlisting></blockquote>
SaveScreen() is used for Screen Saver support (see WaitForSomething()).
pScreen is the screen to save.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->CloseScreen(pScreen)
            ScreenPtr pScreen;

</programlisting></blockquote>
When the server is reset, it calls this routine for each screen.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->CreateScreenResources(pScreen)
            ScreenPtr pScreen;

</programlisting></blockquote>
If this routine is not NULL, it will be called once per screen per
server initialization/reset after all modules have had a chance to
request private space on all structures that support them (see
<xref linkend="wrappers_and_privates"/> below).  You may create resources
in this function instead of in the
screen init function passed to AddScreen in order to guarantee that
all pre-allocated space requests have been registered first.  With the
new devPrivates mechanism, this is not strictly necessary, however.
This routine returns TRUE if successful.</para>
</section>
</section>
<section>
<title>Drawables</title>
<para>
A drawable is a descriptor of a surface that graphics are drawn into, either
a window on the screen or a pixmap in memory.</para>
<para>
Each drawable has a type, class,
ScreenPtr for the screen it is associated with, depth, position, size,
and serial number.
The type is one of the defined constants DRAWABLE_PIXMAP,
DRAWABLE_WINDOW and UNDRAWABLE_WINDOW.
(An undrawable window is used for window class InputOnly.)
The serial number is guaranteed to be unique across drawables, and
is used in determining
the validity of the clipping information in a GC.
The screen selects the set of procedures used to manipulate and draw into the
drawable.  Position is used (currently) only by windows; pixmaps must
set these fields to 0,0 as this reduces the amount of conditional code
executed throughout the mi code.  Size indicates the actual client-specified
size of the drawable.
There are, in fact, no other fields that a window drawable and pixmap
drawable have in common besides those mentioned here.</para>
<para>
Both PixmapRecs and WindowRecs are structs that start with a drawable
and continue on with more fields.  Pixmaps have a single pointer field
named devPrivate which usually points to the pixmap data but could conceivably be
used for anything that DDX wants.  Both windows and pixmaps also have a
devPrivates field which can be used for DDX specific data (see <xref linkend="wrappers_and_privates"/>
below).  This is done because different graphics hardware has
different requirements for management; if the graphics is always
handled by a processor with an independent address space, there is no
point having a pointer to the bit image itself.</para>
<para>
The definition of a drawable and a pixmap can be found in the file
Xserver/include/pixmapstr.h.
The definition of a window can be found in the file Xserver/include/windowstr.h.</para>
<section>
  <title>Pixmaps</title>
<para>
A pixmap is a three-dimensional array of bits stored somewhere offscreen,
rather than in the visible portion of the screen's display frame buffer.  It
can be used as a source or destination in graphics operations.  There is no
implied interpretation of the pixel values in a pixmap, because it has no
associated visual or colormap.  There is only a depth that indicates the
number of significant bits per pixel.  Also, there is no implied physical
size for each pixel; all graphic units are in numbers of pixels.  Therefore,
a pixmap alone does not constitute a complete image; it represents only a
rectangular array of pixel values.</para>
<para>
Note that the pixmap data structure is reference-counted.</para>
<para>
The server implementation is free to put the pixmap data
anywhere it sees fit, according to its graphics hardware setup.  Many
implementations will simply have the data dynamically allocated in the
server's address space.  More sophisticated implementations may put the
data in undisplayed framebuffer storage.</para>
<para>
In addition to dynamic devPrivates (see <xref linkend="wrappers_and_privates"/>
below), the pixmap data structure has two fields that are private to
the device.  Although you can use them for anything you want, they
have intended purposes.  devKind is intended to be a device specific
indication of the pixmap location (host memory, off-screen, etc.).  In
the sample server, since all pixmaps are in memory, devKind stores the
width of the pixmap in bitmap scanline units.  devPrivate is usually
a pointer to the bits in the pixmap.</para>
<para>
A bitmap is a pixmap that is one bit deep.</para>
<para>
<blockquote><programlisting>

        PixmapPtr pScreen->CreatePixmap(pScreen, width, height, depth)
                ScreenPtr pScreen;
                int width, height, depth;

</programlisting></blockquote>
This ScreenRec procedure must create a pixmap of the size
requested.
It must allocate a PixmapRec and fill in all of the fields.
The reference count field must be set to 1.
If width or height are zero, no space should be allocated
for the pixmap data, and if the implementation is using the
devPrivate field as a pointer to the pixmap data, it should be
set to NULL.
If successful, it returns a pointer to the new pixmap; if not, it returns NULL.
See Xserver/fb/fbpixmap.c for the sample server implementation.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->DestroyPixmap(pPixmap)
                PixmapPtr pPixmap;

</programlisting></blockquote>
This ScreenRec procedure must "destroy" a pixmap.
It should decrement the reference count and, if zero, it
must deallocate the PixmapRec and all attached devPrivate blocks.
If successful, it returns TRUE.
See Xserver/fb/fbpixmap.c for the sample server implementation.</para>
<para>
<blockquote><programlisting>

        Bool
        pScreen->ModifyPixmapHeader(pPixmap, width, height, depth, bitsPerPixel, devKind, pPixData)
                PixmapPtr   pPixmap;
                int         width;
                int         height;
                int         depth;
                int         bitsPerPixel;
                int         devKind;
                pointer     pPixData;

</programlisting></blockquote>
This routine takes a pixmap header and initializes the fields of the PixmapRec to the
parameters of the same name.  pPixmap must have been created via
pScreen->CreatePixmap with a zero width or height to avoid
allocating space for the pixmap data.  pPixData is assumed to be the
pixmap data; it will be stored in an implementation-dependent place
(usually pPixmap->devPrivate.ptr).  This routine returns
TRUE if successful.  See Xserver/mi/miscrinit.c for the sample
server implementation.</para>
<para>
<blockquote><programlisting>

        PixmapPtr
        GetScratchPixmapHeader(pScreen, width, height, depth, bitsPerPixel, devKind, pPixData)
                ScreenPtr   pScreen;
                int         width;
                int         height;
                int         depth;
                int         bitsPerPixel;
                int         devKind;
                pointer     pPixData;

        void FreeScratchPixmapHeader(pPixmap)
                PixmapPtr pPixmap;

</programlisting></blockquote>
DDX should use these two DIX routines when it has a buffer of raw
image data that it wants to manipulate as a pixmap temporarily,
usually so that some other part of the server can be leveraged to
perform some operation on the data.  The data should be passed in
pPixData, and will be stored in an implementation-dependent place
(usually pPixmap->devPrivate.ptr). The other
fields go into the corresponding PixmapRec fields.
If successful, GetScratchPixmapHeader returns a valid PixmapPtr which can
be used anywhere the server expects a pixmap, else
it returns NULL.  The pixmap should be released when no longer needed
(usually within the same function that allocated it)
with FreeScratchPixmapHeader.</para>
</section>
<section>
  <title>Windows</title>
<para>
A window is a visible, or potentially visible, rectangle on the screen.
DIX windowing functions maintain an internal n-ary tree data structure, which
represents the current relationships of the mapped windows.
Windows that are contained in another window are children of that window and
are clipped to the boundaries of the parent.
The root window in the tree is the window for the entire screen.
Sibling windows constitute a doubly-linked list; the parent window has a pointer
to the head and tail of this list.
Each child also has a pointer to its parent.</para>
<para>
The border of a window is drawn by a DDX procedure when DIX requests that it
be drawn.  The contents of the window is drawn by the client through
requests to the server.</para>
<para>
Window painting is orchestrated through an expose event system.
When a region is exposed,
DIX generates an expose event, telling the client to repaint the window and
passing the region that is the minimal area needed to be repainted.</para>
<para>
As a favor to clients, the server may retain
the output to the hidden parts of windows
in off-screen memory; this is called "backing store".
When a part of such a window becomes exposed, it
can quickly move pixels into place instead of
triggering an expose event and waiting for a client on the other
end of the network to respond.
Even if the network response is insignificant, the time to
intelligently paint a section of a window is usually more than
the time to just copy already-painted sections.
At best, the repainting involves blanking out the area to a background color,
which will take about the
same amount of time.
In this way, backing store can dramatically increase the
performance of window moves.</para>
<para>
On the other hand, backing store can be quite complex, because
all graphics drawn to hidden areas must be intercepted and redirected
to the off-screen window sections.
Not only can this be complicated for the server programmer,
but it can also impact window painting performance.
The backing store implementation can choose, at any time, to
forget pieces of backing that are written into, relying instead upon
expose events to repaint for simplicity.</para>
<para>
In X, the decision to use the backing-store scheme is made
by you, the server implementor.  The sample server implements
backing store "for free" by reusing the infrastructure for the Composite
extension.  As a side effect, it treats the WhenMapped and Always hints
as equivalent.  However, it will never forget pixel contents when the
window is mapped.</para>
<para>
When a window operation is requested by the client,
such as a window being created or moved,
a new state is computed.
During this transition, DIX informs DDX what rectangles in what windows are about to
become obscured and what rectangles in what windows have become exposed.
This provides a hook for the implementation of backing store.
If DDX is unable to restore exposed regions, DIX generates expose
events to the client.
It is then the client's responsibility to paint the
window parts that were exposed but not restored.</para>
<para>
If a window is resized, pixels sometimes need to be
moved, depending upon
the application.
The client can request "Gravity" so that
certain blocks of the window are
moved as a result of a resize.
For instance, if the window has controls or other items
that always hang on the edge of the
window, and that edge is moved as a result of the resize,
then those pixels should be moved
to avoid having the client repaint it.
If the client needs to repaint it anyway, such an operation takes
time, so it is desirable
for the server to approximate the appearance of the window as best
it can while waiting for the client
to do it perfectly.
Gravity is used for that, also.</para>
<para>
The window has several fields used in drawing
operations:
<itemizedlist>
<listitem><para>
clipList - This region, in conjunction with
the client clip region in the gc, is used to clip output.
clipList has the window's children subtracted from it, in addition to pieces of sibling windows
that overlap this window.  To get the list with the
children included (subwindow-mode is IncludeInferiors),
the routine NotClippedByChildren(pWin) returns the unclipped region.</para></listitem>
<listitem><para>
borderClip is the region used by CopyWindow and
includes the area of the window, its children, and the border, but with the
overlapping areas of sibling children removed.</para></listitem>
</itemizedlist>
Most of the other fields are for DIX use only.</para>
<section>
<title>Window Procedures in the ScreenRec</title>
<para>
You should implement
all of the following procedures and store pointers to them in the screen record.</para>
<para>
The device-independent portion of the server "owns" the window tree.
However, clever hardware might want to know the relationship of
mapped windows.  There are pointers to procedures
in the ScreenRec data structure that are called to give the hardware
a chance to update its internal state.  These are helpers and
hints to DDX only;
they do not change the window tree, which is only changed by DIX.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->CreateWindow(pWin)
                WindowPtr pWin;

</programlisting></blockquote>
This routine is a hook for when DIX creates a window.
It should fill in the "Window Procedures in the WindowRec" below
and also allocate the devPrivate block for it.</para>
<para>
See Xserver/fb/fbwindow.c for the sample server implementation.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->DestroyWindow(pWin);
                WindowPtr pWin;

</programlisting></blockquote>
This routine is a hook for when DIX destroys a window.
It should deallocate the devPrivate block for it and any other blocks that need
to be freed, besides doing other cleanup actions.</para>
<para>
See Xserver/fb/fbwindow.c for the sample server implementation.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->PositionWindow(pWin, x, y);
                WindowPtr pWin;
                int x, y;

</programlisting></blockquote>
This routine is a hook for when DIX moves or resizes a window.
It should do whatever private operations need to be done when a window is moved or resized.
For instance, if DDX keeps a pixmap tile used for drawing the background
or border, and it keeps the tile rotated such that it is longword
aligned to longword locations in the frame buffer, then you should rotate your tiles here.
The actual graphics involved in moving the pixels on the screen and drawing the
border are handled by CopyWindow(), below.</para>
<para>
See Xserver/fb/fbwindow.c for the sample server implementation.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->RealizeWindow(pWin);
                WindowPtr pWin;

        Bool  pScreen->UnrealizeWindow(pWin);
                WindowPtr pWin;

</programlisting></blockquote>
These routines are hooks for when DIX maps (makes visible) and unmaps
(makes invisible) a window.  It should do whatever private operations
need to be done when these happen, such as allocating or deallocating
structures that are only needed for visible windows.  RealizeWindow
does NOT draw the window border, background or contents;
UnrealizeWindow does NOT erase the window or generate exposure events
for underlying windows; this is taken care of by DIX.  DIX does,
however, call PaintWindowBackground() and PaintWindowBorder() to
perform some of these.</para>
<para>
<blockquote><programlisting>

        Bool pScreen->ChangeWindowAttributes(pWin, vmask)
                WindowPtr pWin;
                unsigned long vmask;

</programlisting></blockquote>
ChangeWindowAttributes is called whenever DIX changes window
attributes, such as the size, front-to-back ordering, title, or
anything of lesser severity that affects the window itself.  The
sample server implements this routine.  It computes accelerators for
quickly putting up background and border tiles.  (See description of
the set of routines stored in the WindowRec.)</para>
<para>
<blockquote><programlisting>

        int pScreen->ValidateTree(pParent,  pChild, kind)
                WindowPtr pParent, pChild;
                VTKind kind;

</programlisting></blockquote>
ValidateTree calculates the clipping region for the parent window and
all of its children.  This routine must be provided. The sample server
has a machine-independent version in Xserver/mi/mivaltree.c.  This is
a very difficult routine to replace.</para>
<para>
<blockquote><programlisting>

        void pScreen->PostValidateTree(pParent,  pChild, kind)
                WindowPtr pParent, pChild;
                VTKind kind;

</programlisting></blockquote>
If this routine is not NULL, DIX calls it shortly after calling
ValidateTree, passing it the same arguments.  This is useful for
managing multi-layered framebuffers.
The sample server sets this to NULL.</para>
<para>
<blockquote><programlisting>

        void pScreen->WindowExposures(pWin, pRegion, pBSRegion)
                WindowPtr pWin;
                RegionPtr pRegion;
                RegionPtr pBSRegion;

</programlisting></blockquote>
The WindowExposures() routine
paints the border and generates exposure events for the window.
pRegion is an unoccluded region of the window, and pBSRegion is an
occluded region that has backing store.
Since exposure events include a rectangle describing what was exposed,
this routine may have to send back a series of exposure events, one for
each rectangle of the region.
The count field in the expose event is a hint to the
client as to the number of
regions that are after this one.
This routine must be provided. The sample
server has a machine-independent version in Xserver/mi/miexpose.c.</para>
<para>
<blockquote><programlisting>

        void pScreen->ClipNotify (pWin, dx, dy)
                WindowPtr pWin;
                int dx, dy;

</programlisting></blockquote>
Whenever the cliplist for a window is changed, this function is called to
perform whatever hardware manipulations might be necessary.  When called,
the clip list and border clip regions in the window are set to the new
values.  dx,dy are the distance that the window has been moved (if at all).</para>
</section>
<section>
  <title>Window Painting Procedures</title>
<para>
In addition to the procedures listed above, there are two routines which
manipulate the actual window image directly.
In the sample server, mi implementations will work for
most purposes and fb routines speed up situations, such
as solid backgrounds/borders or tiles that are 8, 16 or 32 pixels square.</para>
<para>
<blockquote><programlisting>

        void pScreen->ClearToBackground(pWin, x, y, w, h, generateExposures);
                WindowPtr pWin;
                int x, y, w, h;
                Bool generateExposures;

</programlisting></blockquote>
This routine is called on a window in response to a ClearToBackground request
from the client.
This request has two different but related functions, depending upon generateExposures.</para>
<para>
If generateExposures is true, the client is declaring that the given rectangle
on the window is incorrectly painted and needs to be repainted.
The sample server implementation calculates the exposure region
and hands it to the DIX procedure HandleExposures(), which
calls the WindowExposures() routine, below, for the window
and all of its child windows.</para>
<para>
If generateExposures is false, the client is trying to simply erase part
of the window to the background fill style.
ClearToBackground should write the background color or tile to the
rectangle in question (probably using PaintWindowBackground).
If w or h is zero, it clears all the way to the right or lower edge of the window.</para>
<para>
The sample server implementation is in Xserver/mi/miwindow.c.</para>
<para>
<blockquote><programlisting>

        void pScreen->CopyWindow(pWin, oldpt, oldRegion);
                WindowPtr pWin;
                DDXPointRec oldpt;
                RegionPtr oldRegion;

</programlisting></blockquote>
CopyWindow is called when a window is moved, and graphically moves to
pixels of a window on the screen.  It should not change any other
state within DDX (see PositionWindow(), above).</para>
<para>
oldpt is the old location of the upper-left corner.  oldRegion is the
old region it is coming from.  The new location and new region is
stored in the WindowRec.  oldRegion might modified in place by this
routine (the sample implementation does this).</para>
<para>
CopyArea could be used, except that this operation has more
complications.  First of all, you do not want to copy a rectangle onto
a rectangle.  The original window may be obscured by other windows,
and the new window location may be similarly obscured.  Second, some
hardware supports multiple windows with multiple depths, and your
routine needs to take care of that.</para>
<para>
The pixels in oldRegion (with reference point oldpt) are copied to the
window's new region (pWin->borderClip).  pWin->borderClip is gotten
directly from the window, rather than passing it as a parameter.</para>
<para>
The sample server implementation is in Xserver/fb/fbwindow.c.</para>
</section>
<section>
<title>Screen Operations for Multi-Layered Framebuffers</title>
<para>
The following screen functions are useful if you have a framebuffer with
multiple sets of independent bit planes, e.g. overlays or underlays in
addition to the "main" planes.  If you have a simple single-layer
framebuffer, you should probably use the mi versions of these routines
in mi/miwindow.c.  This can be easily accomplished by calling miScreenInit.</para>
<para>
<blockquote><programlisting>

    void pScreen->MarkWindow(pWin)
        WindowPtr pWin;

</programlisting></blockquote>
This formerly dix function MarkWindow has moved to ddx and is accessed
via this screen function.  This function should store something,
usually a pointer to a device-dependent structure, in pWin->valdata so
that ValidateTree has the information it needs to validate the window.</para>
<para>
<blockquote><programlisting>

    Bool pScreen->MarkOverlappedWindows(parent, firstChild, ppLayerWin)
        WindowPtr parent;
        WindowPtr firstChild;
        WindowPtr * ppLayerWin;

</programlisting></blockquote>
This formerly dix function MarkWindow has moved to ddx and is accessed
via this screen function.  In the process, it has grown another
parameter: ppLayerWin, which is filled in with a pointer to the window
at which save under marking and ValidateTree should begin.  In the
single-layered framebuffer case, pLayerWin == pWin.</para>
<para>
<blockquote><programlisting>

    Bool pScreen->ChangeSaveUnder(pLayerWin, firstChild)
        WindowPtr pLayerWin;
        WindowPtr firstChild;

</programlisting></blockquote>
The dix functions ChangeSaveUnder and CheckSaveUnder have moved to ddx and
are accessed via this screen function.  pLayerWin should be the window
returned in the ppLayerWin parameter of MarkOverlappedWindows.  The function
may turn on backing store for windows that might be covered, and may partially
turn off backing store for windows.  It returns TRUE if PostChangeSaveUnder
needs to be called to finish turning off backing store.</para>
<para>
<blockquote><programlisting>

    void pScreen->PostChangeSaveUnder(pLayerWin, firstChild)
        WindowPtr pLayerWin;
        WindowPtr firstChild;

</programlisting></blockquote>
The dix function DoChangeSaveUnder has moved to ddx and is accessed via
this screen function.  This function completes the job of turning off
backing store that was started by ChangeSaveUnder.</para>
<para>
<blockquote><programlisting>

    void pScreen->MoveWindow(pWin, x, y, pSib, kind)
        WindowPtr pWin;
        int x;
        int y;
        WindowPtr pSib;
        VTKind kind;

</programlisting></blockquote>
The formerly dix function MoveWindow has moved to ddx and is accessed via
this screen function.  The new position of the window is given by
x,y.  kind is VTMove if the window is only moving, or VTOther if
the border is also changing.</para>
<para>
<blockquote><programlisting>

    void pScreen->ResizeWindow(pWin, x, y, w, h, pSib)
        WindowPtr pWin;
        int x;
        int y;
        unsigned int w;
        unsigned int h;
        WindowPtr pSib;

</programlisting></blockquote>
The formerly dix function SlideAndSizeWindow has moved to ddx and is accessed via
this screen function.  The new position is given by x,y.  The new size
is given by w,h.</para>
<para>
<blockquote><programlisting>

    WindowPtr pScreen->GetLayerWindow(pWin)
        WindowPtr pWin

</programlisting></blockquote>
This is a new function which returns a child of the layer parent of pWin.</para>
<para>
<blockquote><programlisting>

    void pScreen->HandleExposures(pWin)
        WindowPtr pWin;

</programlisting></blockquote>
The formerly dix function HandleExposures has moved to ddx and is accessed via
this screen function.  This function is called after ValidateTree and
uses the information contained in valdata to send exposures to windows.</para>
<para>
<blockquote><programlisting>

    void pScreen->ReparentWindow(pWin, pPriorParent)
        WindowPtr pWin;
        WindowPtr pPriorParent;

</programlisting></blockquote>
This function will be called when a window is reparented.  At the time of
the call, pWin will already be spliced into its new position in the
window tree, and pPriorParent is its previous parent.  This function
can be NULL.</para>
<para>
<blockquote><programlisting>

    void pScreen->SetShape(pWin)
        WindowPtr pWin;

</programlisting></blockquote>
The formerly dix function SetShape has moved to ddx and is accessed via
this screen function.  The window's new shape will have already been
stored in the window when this function is called.</para>
<para>
<blockquote><programlisting>

    void pScreen->ChangeBorderWidth(pWin, width)
        WindowPtr pWin;
        unsigned int width;

</programlisting></blockquote>
The formerly dix function ChangeBorderWidth has moved to ddx and is accessed via
this screen function.  The new border width is given by width.</para>
<para>
<blockquote><programlisting>

    void pScreen->MarkUnrealizedWindow(pChild, pWin, fromConfigure)
        WindowPtr pChild;
        WindowPtr pWin;
        Bool fromConfigure;

</programlisting></blockquote>
This function is called for windows that are being unrealized as part of
an UnrealizeTree.  pChild is the window being unrealized, pWin is an
ancestor, and the fromConfigure value is simply propagated from UnrealizeTree.</para>
</section>
</section>
</section>
<section>
<title>Graphics Contexts and Validation</title>
<para>
This graphics context (GC) contains state variables such as foreground and
background pixel value (color), the current line style and width,
the current tile or stipple for pattern generation, the current font for text
generation, and other similar attributes.</para>
<para>
In many graphics systems, the equivalent of the graphics context and the
drawable are combined as one entity.
The main distinction between the two kinds of status is that a drawable
describes a writing surface and the writings that may have already been done
on it, whereas a graphics context describes the drawing process.
A drawable is like a chalkboard.
A GC is like a piece of chalk.</para>
<para>
Unlike many similar systems, there is no "current pen location."
Every graphic operation is accompanied by the coordinates where it is to happen.</para>
<para>
The GC also includes two vectors of procedure pointers, the first
operate on the GC itself and are called GC funcs.  The second, called
GC ops,
contains the functions that carry out the fundamental graphic operations
such as drawing lines, polygons, arcs, text, and copying bitmaps.
The DDX graphic software can, if it
wants to be smart, change these two vectors of procedure pointers
to take advantage of hardware/firmware in the server machine, which can do
a better job under certain circumstances.  To reduce the amount of memory
consumed by each GC, it is wise to create a few "boilerplate" GC ops vectors
which can be shared by every GC which matches the constraints for that set.
Also, it is usually reasonable to have every GC created by a particular
module to share a common set of GC funcs.  Samples of this sort of
sharing can be seen in fb/fbgc.c.</para>
<para>
The DDX software is notified any time the client (or DIX) uses a changed GC.
For instance, if the hardware has special support for drawing fixed-width
fonts, DDX can intercept changes to the current font in a GC just before
drawing is done.  It can plug into either a fixed-width procedure that makes
the hardware draw characters, or a variable-width procedure that carefully
lays out glyphs by hand in software, depending upon the new font that is
selected.</para>
<para>
A definition of these structures can be found in the file
Xserver/include/gcstruct.h.</para>
<para>
Also included in each GC is support for dynamic devPrivates, which the
DDX can use for any purpose (see <xref linkend="wrappers_and_privates"/> below).</para>
<para>
The DIX routines available for manipulating GCs are
CreateGC, ChangeGC, ChangeGCXIDs, CopyGC, SetClipRects, SetDashes, and FreeGC.
<blockquote><programlisting>

        GCPtr CreateGC(pDrawable, mask, pval, pStatus)
            DrawablePtr pDrawable;
            BITS32 mask;
            XID *pval;
            int *pStatus;

        int ChangeGC(client, pGC, mask, pUnion)
            ClientPtr client;
            GCPtr pGC;
            BITS32 mask;
            ChangeGCValPtr pUnion;

        int ChangeGCXIDs(client, pGC, mask, pC32)
            ClientPtr client;
            GCPtr pGC;
            BITS32 mask;
            CARD32 *pC32;

        int CopyGC(pgcSrc, pgcDst, mask)
            GCPtr pgcSrc;
            GCPtr pgcDst;
            BITS32 mask;

        int SetClipRects(pGC, xOrigin, yOrigin, nrects, prects, ordering)
            GCPtr pGC;
            int xOrigin, yOrigin;
            int nrects;
            xRectangle *prects;
            int ordering;

        SetDashes(pGC, offset, ndash, pdash)
            GCPtr pGC;
            unsigned offset;
            unsigned ndash;
            unsigned char *pdash;

        int FreeGC(pGC, gid)
            GCPtr pGC;
            GContext gid;

</programlisting></blockquote>
</para>
<para>
As a convenience, each Screen structure contains an array of
GCs that are preallocated, one at each depth the screen supports.
These are particularly useful in the mi code.  Two DIX routines
must be used to get these GCs:
<blockquote><programlisting>

        GCPtr GetScratchGC(depth, pScreen)
            int depth;
            ScreenPtr pScreen;

        FreeScratchGC(pGC)
            GCPtr pGC;

</programlisting></blockquote>
Always use these two routines, don't try to extract the scratch
GC yourself -- someone else might be using it, so a new one must
be created on the fly.</para>
<para>
If you need a GC for a very long time, say until the server is restarted,
you should not take one from the pool used by GetScratchGC, but should
get your own using CreateGC or CreateScratchGC.
This leaves the ones in the pool free for routines that only need it for
a little while and don't want to pay a heavy cost to get it.
<blockquote><programlisting>

        GCPtr CreateScratchGC(pScreen, depth)
            ScreenPtr pScreen;
            int depth;

</programlisting></blockquote>
NULL is returned if the GC cannot be created.
The GC returned can be freed with FreeScratchGC.</para>
<section>
  <title>Details of Operation</title>
<para>
At screen initialization, a screen must supply a GC creation procedure.
At GC creation, the screen must fill in GC funcs and GC ops vectors
(Xserver/include/gcstruct.h).  For any particular GC, the func vector
must remain constant, while the op vector may vary.  This invariant is to
ensure that Wrappers work correctly.</para>
<para>
When a client request is processed that results in a change
to the GC, the device-independent state of the GC is updated.
This includes a record of the state that changed.
Then the ChangeGC GC func is called.
This is useful for graphics subsystems that are able to process
state changes in parallel with the server CPU.
DDX may opt not to take any action at GC-modify time.
This is more efficient if multiple GC-modify requests occur
between draws using a given GC.</para>
<para>
Validation occurs at the first draw operation that specifies the GC after
that GC was modified.  DIX calls then the ValidateGC GC func.  DDX should
then update its internal state.  DDX internal state may be stored as one or
more of the following:  1) device private block on the GC; 2) hardware
state; 3) changes to the GC ops.</para>
<para>
The GC contains a serial number, which is loaded with a number fetched from
the window that was drawn into the last time the GC was used.  The serial
number in the drawable is changed when the drawable's
clipList or absCorner changes.  Thus, by
comparing the GC serial number with the drawable serial number, DIX can
force a validate if the drawable has been changed since the last time it
was used with this GC.</para>
<para>
In addition, the drawable serial number is always guaranteed to have the
most significant bit set to 0.  Thus, the DDX layer can set the most
significant bit of the serial number to 1 in a GC to force a validate the next time
the GC is used.  DIX also uses this technique to indicate that a change has
been made to the GC by way of a SetGC, a SetDashes or a SetClip request.</para>
</section>
<section>
  <title>GC Handling Routines</title>
<para>
The ScreenRec data structure has a pointer for
CreateGC().
<blockquote><programlisting>

        Bool pScreen->CreateGC(pGC)
                GCPtr pGC;
</programlisting></blockquote>
This routine must fill in the fields of
a dynamically allocated GC that is passed in.
It does NOT allocate the GC record itself or fill
in the defaults; DIX does that.</para>
<para>
This must fill in both the GC funcs and ops; none of the drawing
functions will be called before the GC has been validated,
but the others (dealing with allocating of clip regions,
changing and destroying the GC, etc.) might be.</para>
<para>
The GC funcs vector contains pointers to 7
routines and a devPrivate field:
<blockquote><programlisting>

        pGC->funcs->ChangeGC(pGC, changes)
                GCPtr pGC;
                unsigned long changes;

</programlisting></blockquote>
This GC func is called immediately after a field in the GC is changed.
changes is a bit mask indicating the changed fields of the GC in this
request.</para>
<para>
The ChangeGC routine is useful if you have a system where
state-changes to the GC can be swallowed immediately by your graphics
system, and a validate is not necessary.</para>
<para>
<blockquote><programlisting>

        pGC->funcs->ValidateGC(pGC, changes, pDraw)
                GCPtr pGC;
                unsigned long changes;
                DrawablePtr pDraw;

</programlisting></blockquote>
ValidateGC is called by DIX just before the GC will be used when one
of many possible changes to the GC or the graphics system has
happened.  It can modify devPrivates data attached to the GC,
change the op vector, or change hardware according to the
values in the GC.  It may not change the device-independent portion of
the GC itself.</para>
<para>
In almost all cases, your ValidateGC() procedure should take the
regions that drawing needs to be clipped to and combine them into a
composite clip region, which you keep a pointer to in the private part
of the GC.  In this way, your drawing primitive routines (and whatever
is below them) can easily determine what to clip and where.  You
should combine the regions clientClip (the region that the client
desires to clip output to) and the region returned by
NotClippedByChildren(), in DIX.  An example is in Xserver/fb/fbgc.c.</para>
<para>
Some kinds of extension software may cause this routine to be called
more than originally intended; you should not rely on algorithms that
will break under such circumstances.</para>
<para>
See the Strategies document for more information on creatively using
this routine.</para>
<para>
<blockquote><programlisting>

        pGC->funcs->CopyGC(pGCSrc, mask, pGCDst)
                GCPtr pGCSrc;
                unsigned long mask;
                GCPtr pGCDst;

</programlisting></blockquote>
This routine is called by DIX when a GC is being copied to another GC.
This is for situations where dynamically allocated chunks of memory
are stored in the GC's dynamic devPrivates and need to be transferred to
the destination GC.</para>
<para>
<blockquote><programlisting>

        pGC->funcs->DestroyGC(pGC)
                GCPtr pGC;

</programlisting></blockquote>
This routine is called before the GC is destroyed for the
entity interested in this GC to clean up after itself.
This routine is responsible for freeing any auxiliary storage allocated.</para>
</section>
<section>
  <title>GC Clip Region Routines</title>
<para>
The GC clientClip field requires three procedures to manage it.  These
procedures are in the GC funcs vector.  The underlying principle is that dix
knows nothing about the internals of the clipping information, (except when
it has come from the client), and so calls ddX whenever it needs to copy,
set, or destroy such information.  It could have been possible for dix not
to allow ddX to touch the field in the GC, and require it to keep its own
copy in devPriv, but since clip masks can be very large, this seems like a
bad idea.  Thus, the server allows ddX to do whatever it wants to the
clientClip field of the GC, but requires it to do all manipulation itself.</para>
<para>
<blockquote><programlisting>

        void pGC->funcs->ChangeClip(pGC, type, pValue, nrects)
                GCPtr pGC;
                int type;
                char *pValue;
                int nrects;

</programlisting></blockquote>
This routine is called whenever the client changes the client clip
region.  The pGC points to the GC involved, the type tells what form
the region has been sent in.  If type is CT_NONE, then there is no
client clip.  If type is CT_UNSORTED, CT_YBANDED or CT_YXBANDED, then
pValue pointer to a list of rectangles, nrects long.  If type is
CT_REGION, then pValue pointer to a RegionRec from the mi region code.
If type is CT_PIXMAP pValue is a pointer to a pixmap.  (The defines
for CT_NONE, etc. are in Xserver/include/gc.h.)  This routine is
responsible for incrementing any necessary reference counts (e.g. for
a pixmap clip mask) for the new clipmask and freeing anything that
used to be in the GC's clipMask field.  The lists of rectangles passed
in can be freed with free(), the regions can be destroyed with the
RegionDestroy field in the screen, and pixmaps can be destroyed by
calling the screen's DestroyPixmap function.  DIX and MI code expect
what they pass in to this to be freed or otherwise inaccessible, and
will never look inside what's been put in the GC.  This is a good
place to be wary of storage leaks.</para>
<para>
In the sample server, this routine transforms either the bitmap or the
rectangle list into a region, so that future routines will have a more
predictable starting point to work from.  (The validate routine must
take this client clip region and merge it with other regions to arrive
at a composite clip region before any drawing is done.)</para>
<para>
<blockquote><programlisting>

        void pGC->funcs->DestroyClip(pGC)
                GCPtr pGC;

</programlisting></blockquote>
This routine is called whenever the client clip region must be destroyed.
The pGC points to the GC involved.  This call should set the clipType
field of the GC to CT_NONE.
In the sample server, the pointer to the client clip region is set to NULL
by this routine after destroying the region, so that other software
(including ChangeClip() above) will recognize that there is no client clip region.</para>
<para>
<blockquote><programlisting>

        void pGC->funcs->CopyClip(pgcDst, pgcSrc)
                GCPtr pgcDst, pgcSrc;

</programlisting></blockquote>
This routine makes a copy of the clipMask and clipType from pgcSrc
into pgcDst.  It is responsible for destroying any previous clipMask
in pgcDst.  The clip mask in the source can be the same as the
clip mask in the dst (clients do the strangest things), so care must
be taken when destroying things.  This call is required because dix
does not know how to copy the clip mask from pgcSrc.</para>
</section>
</section>
<section>
  <title>Drawing Primitives</title>
<para>
The X protocol (rules for the byte stream that goes between client and server)
does all graphics using primitive
operations, which are called Drawing Primitives.
These include line drawing, area filling, arcs, and text drawing.
Your implementation must supply 16 routines
to perform these on your hardware.
(The number 16 is arbitrary.)</para>
<para>
More specifically, 16 procedure pointers are in each
GC op vector.
At any given time, ALL of them MUST point to a valid procedure that
attempts to do the operation assigned, although
the procedure pointers may change and may
point to different procedures to carry out the same operation.
A simple server will leave them all pointing to the same 16 routines, while
a more optimized implementation will switch each from one
procedure to another, depending upon what is most optimal
for the current GC and drawable.</para>
<para>
The sample server contains a considerable chunk of code called the
mi (machine independent)
routines, which serve as drawing primitive routines.
Many server implementations will be able to use these as-is,
because they work for arbitrary depths.
They make no assumptions about the formats of pixmaps
and frame buffers, since they call a set of routines
known as the "Pixblit Routines" (see next section).
They do assume that the way to draw is
through these low-level routines that apply pixel values rows at a time.
If your hardware or firmware gives more performance when
things are done differently, you will want to take this fact into account
and rewrite some or all of the drawing primitives to fit your needs.</para>
<section>
  <title>GC Components</title>
<para>
This section describes the fields in the GC that affect each drawing primitive.
The only primitive that is not affected is GetImage, which does not use a GC
because its destination is a protocol-style bit image.
Since each drawing primitive mirrors exactly the X protocol request of the
same name, you should refer to the X protocol specification document
for more details.</para>
<para>
ALL of these routines MUST CLIP to the
appropriate regions in the drawable.
Since there are many regions to clip to simultaneously,
your ValidateGC routine should combine these into a unified
clip region to which your drawing routines can quickly refer.
This is exactly what the fb routines supplied with the sample server
do.
The mi implementation passes responsibility for clipping while drawing
down to the Pixblit routines.</para>
<para>
Also, all of them must adhere to the current plane mask.
The plane mask has one bit for every bit plane in the drawable;
only planes with 1 bits in the mask are affected by any drawing operation.</para>
<para>
All functions except for ImageText calls must obey the alu function.
This is usually Copy, but could be any of the allowable 16 raster-ops.</para>
<para>
All of the functions, except for CopyArea, might use the current
foreground and background pixel values.
Each pixel value is 32 bits.
These correspond to foreground and background colors, but you have
to run them through the colormap to find out what color the pixel values
represent.  Do not worry about the color, just apply the pixel value.</para>
<para>
The routines that draw lines (PolyLine, PolySegment, PolyRect, and PolyArc)
use the line width, line style, cap style, and join style.
Line width is in pixels.
The line style specifies whether it is solid or dashed, and what kind of dash.
The cap style specifies whether Rounded, Butt, etc.
The join style specifies whether joins between joined lines are Miter, Round or Beveled.
When lines cross as part of the same polyline, they are assumed to be drawn once.
(See the X protocol specification for more details.)</para>
<para>
Zero-width lines are NOT meant to be really zero width; this is the client's way
of telling you that you can optimize line drawing with little regard to
the end caps and joins.
They are called "thin" lines and are meant to be one pixel wide.
These are frequently done in hardware or in a streamlined assembly language
routine.</para>
<para>
Lines with widths greater than zero, though, must all be drawn with the same
algorithm, because client software assumes that every jag on every
line at an angle will come at the same place.
Two lines that should have
one pixel in the space between them
(because of their distance apart and their widths) should have such a one-pixel line
of space between them if drawn, regardless of angle.</para>
<para>
The solid area fill routines (FillPolygon, PolyFillRect, PolyFillArc)
all use the fill rule, which specifies subtle interpretations of
what points are inside and what are outside of a given polygon.
The PolyFillArc routine also uses the arc mode, which specifies
whether to fill pie segments or single-edge slices of an ellipse.</para>
<para>
The line drawing, area fill, and PolyText routines must all
apply the correct "fill style."
This can be either a solid foreground color, a transparent stipple,
an opaque stipple, or a tile.
Stipples are bitmaps where the 1 bits represent that the foreground color is written,
and 0 bits represent that either the pixel is left alone (transparent) or that
the background color is written (opaque).
A tile is a pixmap of the full depth of the GC that is applied in its full glory to all areas.
The stipple and tile patterns can be any rectangular size, although some implementations
will be faster for certain sizes such as 8x8 or 32x32.
The mi implementation passes this responsibility down to the Pixblit routines.</para>
<para>
See the X protocol document for full details.
The description of the CreateGC request has a very good, detailed description of these
attributes.</para>
</section>
<section>
<title>The Primitives</title>
<para>
The Drawing Primitives are as follows:

<blockquote><programlisting>

        RegionPtr pGC->ops->CopyArea(src, dst, pGC, srcx, srcy, w, h, dstx, dsty)
                DrawablePtr dst, src;
                GCPtr pGC;
                int srcx, srcy, w, h, dstx, dsty;

</programlisting></blockquote>
CopyArea copies a rectangle of pixels from one drawable to another of
the same depth.  To effect scrolling, this must be able to copy from
any drawable to itself, overlapped.  No squeezing or stretching is done
because the source and destination are the same size.  However,
everything is still clipped to the clip regions of the destination
drawable.</para>
<para>
If pGC->graphicsExposures is True, any portions of the destination which
were not valid in the source (either occluded by covering windows, or
outside the bounds of the drawable) should be collected together and
returned as a region (if this resultant region is empty, NULL can be
returned instead).  Furthermore, the invalid bits of the source are
not copied to the destination and (when the destination is a window)
are filled with the background tile.  The sample routine
miHandleExposures generates the appropriate return value and fills the
invalid area using pScreen->PaintWindowBackground.</para>
<para>
For instance, imagine a window that is partially obscured by other
windows in front of it.  As text is scrolled on your window, the pixels
that are scrolled out from under obscuring windows will not be
available on the screen to copy to the right places, and so an exposure
event must be sent for the client to correctly repaint them.  Of
course, if you implement backing store, you could do this without resorting
to exposure events.</para>
<para>
An example implementation is fbCopyArea() in Xserver/fb/fbcopy.c.</para>
<para>
<blockquote><programlisting>

        RegionPtr pGC->ops->CopyPlane(src, dst, pGC, srcx, srcy, w, h, dstx, dsty, plane)
                DrawablePtr dst, src;
                GCPtr pGC;
                int srcx, srcy, w, h, dstx, dsty;
                unsigned long plane;

</programlisting></blockquote>
CopyPlane must copy one plane of a rectangle from the source drawable
onto the destination drawable.  Because this routine only copies one
bit out of each pixel, it can copy between drawables of different
depths.  This is the only way of copying between drawables of
different depths, except for copying bitmaps to pixmaps and applying
foreground and background colors to it.  All other conditions of
CopyArea apply to CopyPlane too.</para>
<para>
An example implementation is fbCopyPlane() in
Xserver/fb/fbcopy.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolyPoint(dst, pGC, mode, n, pPoint)
                DrawablePtr dst;
                GCPtr pGC;
                int mode;
                int n;
                DDXPointPtr pPoint;

</programlisting></blockquote>
PolyPoint draws a set of one-pixel dots (foreground color)
at the locations given in the array.
mode is one of the defined constants Origin (absolute coordinates) or Previous
(each coordinate is relative to the last).
Note that this does not use the background color or any tiles or stipples.</para>
<para>
Example implementations are fbPolyPoint() in Xserver/fb/fbpoint.c and
miPolyPoint in Xserver/mi/mipolypnt.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->Polylines(dst, pGC, mode, n, pPoint)
                DrawablePtr dst;
                GCPtr pGC;
                int mode;
                int n;
                DDXPointPtr pPoint;

</programlisting></blockquote>
Similar to PolyPoint, Polylines draws lines between the locations given in the array.
Zero-width lines are NOT meant to be really zero width; this is the client's way of
telling you that you can maximally optimize line drawing with little regard to
the end caps and joins.
mode is one of the defined constants Previous or Origin, depending upon
whether the points are each relative to the last or are absolute.</para>
<para>
Example implementations are miWideLine() and miWideDash() in
mi/miwideline.c and miZeroLine() in mi/mizerline.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolySegment(dst, pGC, n, pPoint)
                DrawablePtr dst;
                GCPtr pGC;
                int n;
                xSegment *pSegments;

</programlisting></blockquote>
PolySegments draws unconnected
lines between pairs of points in the array; the array must be of
even size; no interconnecting lines are drawn.</para>
<para>
An example implementation is miPolySegment() in mipolyseg.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolyRectangle(dst, pGC, n, pRect)
                DrawablePtr dst;
                GCPtr pGC;
                int n;
                xRectangle *pRect;

</programlisting></blockquote>
PolyRectangle draws outlines of rectangles for each rectangle in the array.</para>
<para>
An example implementation is miPolyRectangle() in Xserver/mi/mipolyrect.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolyArc(dst, pGC, n, pArc)
                DrawablePtr dst;
                GCPtr pGC;
                int n;
                xArc*pArc;

</programlisting></blockquote>
PolyArc draws connected conic arcs according to the descriptions in the array.
See the protocol specification for more details.</para>
<para>
Example implementations are miZeroPolyArc in Xserver/mi/mizerarc. and
miPolyArc() in Xserver/mi/miarc.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->FillPolygon(dst, pGC, shape, mode, count, pPoint)
                DrawablePtr dst;
                GCPtr pGC;
                int shape;
                int mode;
                int count;
                DDXPointPtr pPoint;

</programlisting></blockquote>
FillPolygon fills a polygon specified by the points in the array
with the appropriate fill style.
If necessary, an extra border line is assumed between the starting and ending lines.
The shape can be used as a hint
to optimize filling; it indicates whether it is convex (all interior angles
less than 180), nonconvex (some interior angles greater than 180 but
border does not cross itself), or complex (border crosses itself).
You can choose appropriate algorithms or hardware based upon mode.
mode is one of the defined constants Previous or Origin, depending upon
whether the points are each relative to the last or are absolute.</para>
<para>
An example implementation is miFillPolygon() in Xserver/mi/mipoly.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolyFillRect(dst, pGC, n, pRect)
                DrawablePtr dst;
                GCPtr pGC;
                int n;
                xRectangle *pRect;

</programlisting></blockquote>
PolyFillRect fills multiple rectangles.</para>
<para>
Example implementations are fbPolyFillRect() in Xserver/fb/fbfillrect.c and
miPolyFillRect() in Xserver/mi/mifillrct.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolyFillArc(dst, pGC, n, pArc)
                DrawablePtr dst;
                GCPtr pGC;
                int n;
                xArc *pArc;

</programlisting></blockquote>
PolyFillArc fills a shape for each arc in the
list that is bounded by the arc and one or two
line segments with the current fill style.</para>
<para>
An example implementation is miPolyFillArc() in Xserver/mi/mifillarc.c.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PutImage(dst, pGC, depth, x, y, w, h, leftPad, format, pBinImage)
                DrawablePtr dst;
                GCPtr pGC;
                int x, y, w, h;
                int format;
                char *pBinImage;

</programlisting></blockquote>
PutImage copies a pixmap image into the drawable.  The pixmap image
must be in X protocol format (either Bitmap, XYPixmap, or ZPixmap),
and format tells the format.  (See the X protocol specification for
details on these formats).  You must be able to accept all three
formats, because the client gets to decide which format to send.
Either the drawable and the pixmap image have the same depth, or the
source pixmap image must be a Bitmap.  If a Bitmap, the foreground and
background colors will be applied to the destination.</para>
<para>
An example implementation is fbPutImage() in Xserver/fb/fbimage.c.</para>
<para>
<blockquote><programlisting>

        void pScreen->GetImage(src, x, y, w, h, format, planeMask, pBinImage)
                 DrawablePtr src;
                 int x, y, w, h;
                 unsigned int format;
                 unsigned long planeMask;
                 char *pBinImage;

</programlisting></blockquote>
GetImage copies the bits from the source drawable into
the destination pointer.  The bits are written into the buffer
according to the server-defined pixmap padding rules.
pBinImage is guaranteed to be big enough to hold all
the bits that must be written.</para>
<para>
This routine does not correspond exactly to the X protocol GetImage
request, since DIX has to break the reply up into buffers of a size
requested by the transport layer.  If format is ZPixmap, the bits are
written in the ZFormat for the depth of the drawable; if there is a 0
bit in the planeMask for a particular plane, all pixels must have the
bit in that plane equal to 0.  If format is XYPixmap, planemask is
guaranteed to have a single bit set; the bits should be written in
Bitmap format, which is the format for a single plane of an XYPixmap.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->ImageText8(pDraw, pGC, x, y, count, chars)
                DrawablePtr pDraw;
                GCPtr pGC;
                int x, y;
                int count;
                char *chars;

</programlisting></blockquote>
ImageText8 draws text.  The text is drawn in the foreground color; the
background color fills the remainder of the character rectangles.  The
coordinates specify the baseline and start of the text.</para>
<para>
An example implementation is miImageText8() in Xserver/mi/mipolytext.c.</para>
<para>
<blockquote><programlisting>

        int pGC->ops->PolyText8(pDraw, pGC, x, y, count, chars)
                DrawablePtr pDraw;
                GCPtr pGC;
                int x, y;
                int count;
                char *chars;

</programlisting></blockquote>
PolyText8 works like ImageText8, except it draws with
the current fill style for special effects such as
shaded text.
See the X protocol specification for more details.</para>
<para>
An example implementation is miPolyText8() in Xserver/mi/mipolytext.c.</para>
<para>
<blockquote><programlisting>

        int pGC->ops->PolyText16(pDraw, pGC, x, y, count, chars)
                DrawablePtr pDraw;
                GCPtr pGC;
                int x, y;
                int count;
                unsigned short *chars;

        void pGC->ops->ImageText16(pDraw, pGC, x, y, count, chars)
                DrawablePtr pDraw;
                GCPtr pGC;
                int x, y;
                int count;
                unsigned short *chars;

</programlisting></blockquote>
These two routines are the same as the "8" versions,
except that they are for 16-bit character codes (useful
for oriental writing systems).</para>
<para>
The primary difference is in the way the character information is
looked up.  The 8-bit and the 16-bit versions obviously have different
kinds of character values to look up; the main goal of the lookup is
to provide a pointer to the CharInfo structs for the characters to
draw and to pass these pointers to the Glyph routines.  Given a
CharInfo struct, lower-level software can draw the glyph desired with
little concern for other characteristics of the font.</para>
<para>
16-bit character fonts have a row-and-column scheme, where the 2bytes
of the character code constitute the row and column in a square matrix
of CharInfo structs.  Each font has row and column minimum and maximum
values; the CharInfo structures form a two-dimensional matrix.</para>
<para>
Example implementations are miPolyText16() and
miImageText16() in Xserver/mi/mipolytext.c.</para>
<para>
See the X protocol specification for more details on these graphic operations.</para>
<para>
There is a hook in the GC ops, called LineHelper, that used to be used in the
sample implementation by the code for wide lines.  It no longer servers any
purpose in the sample servers, but still exists, #ifdef'ed by NEED_LINEHELPER,
in case someone needs it.</para>
</section>
</section>
<section>
  <title>Pixblit Procedures</title>
<para>
The Drawing Primitive functions must be defined for your server.
One possible way to do this is to use the mi routines from the sample server.
If you choose to use the mi routines (even part of them!) you must implement
these Pixblit routines.
These routines read and write pixel values
and deal directly with the image data.</para>
<para>
The Pixblit routines for the sample server are part of the "fb"
routines.  As with the mi routines, the fb routines are
portable but are not as portable as the mi routines.</para>
<para>
The fb subsystem is a depth-independent framebuffer core, capable of
operating at any depth from 1 to 32, based on the depth of the window
or pixmap it is currently operating on.  In particular, this means it
can support pixmaps of multiple depths on the same screen.  It supplies
both Pixblit routines and higher-level optimized implementations of the
Drawing Primitive routines.  It does make the assumption that the pixel
data it touches is available in the server's address space.</para>
<para>
In other words, if you have a "normal" frame buffer type display, you
can probably use the fb code, and the mi code.  If you
have a stranger hardware, you will have to supply your own Pixblit
routines, but you can use the mi routines on top of them.  If you have
better ways of doing some of the Drawing Primitive functions, then you
may want to supply some of your own Drawing Primitive routines.  (Even
people who write their own Drawing Primitives save at least some of
the mi code for certain special cases that their hardware or library
or fancy algorithm does not handle.)</para>
<para>
The client, DIX, and the machine-independent routines do not carry the
final responsibility of clipping.  They all depend upon the Pixblit
routines to do their clipping for them.  The rule is, if you touch the
frame buffer, you clip.</para>
<para>
(The higher level routines may decide to clip at a high level, but
this is only for increased performance and cannot substitute for
bottom-level clipping.  For instance, the mi routines, DIX, or the
client may decide to check all character strings to be drawn and chop
off all characters that would not be displayed.  If so, it must retain
the character on the edge that is partly displayed so that the Pixblit
routines can clip off precisely at the right place.)</para>
<para>
To make this easier, all of the reasons to clip can be combined into
one region in your ValidateGC procedure.  You take this composite clip
region with you into the Pixblit routines.  (The sample server does
this.)</para>
<para>
Also, FillSpans() has to apply tile and stipple patterns.  The
patterns are all aligned to the window origin so that when two people
write patches that are contiguous, they will merge nicely.  (Really,
they are aligned to the patOrg point in the GC.  This defaults to (0,
0) but can be set by the client to anything.)</para>
<para>
However, the mi routines can translate (relocate) the points from
window-relative to screen-relative if desired.  If you set the
miTranslate field in the GC (set it in the CreateGC or ValidateGC
routine), then the mi output routines will translate all coordinates.
If it is false, then the coordinates will be passed window-relative.
Screens with no hardware translation will probably set miTranslate to
TRUE, so that geometry (e.g. polygons, rectangles) can be translated,
rather than having the resulting list of scanlines translated; this is
good because the list vertices in a drawing request will generally be
much smaller than the list of scanlines it produces.  Similarly,
hardware that does translation can set miTranslate to FALSE, and avoid
the extra addition per vertex, which can be (but is not always)
important for getting the highest possible performance.  (Contrast the
behavior of GetSpans, which is not expected to be called as often, and
so has different constraints.)  The miTranslate field is settable in
each GC, if , for example, you are mixing several kinds of
destinations (offscreen pixmaps, main memory pixmaps, backing store,
and windows), all of which have different requirements, on one screen.</para>
<para>
As with other drawing routines, there are fields in the GC to direct
higher code to the correct routine to execute for each function.  In
this way, you can optimize for special cases, for example, drawing
solids versus drawing stipples.</para>
<para>
The Pixblit routines are broken up into three sets.  The Span routines
simply fill in rows of pixels.  The Glyph routines fill in character
glyphs.  The PushPixels routine is a three-input bitblt for more
sophisticated image creation.</para>
<para>
It turns out that the Glyph and PushPixels routines actually have a
machine-independent implementation that depends upon the Span
routines.  If you are really pressed for time, you can use these
versions, although they are quite slow.</para>
<section>
<title>Span Routines</title>
<para>
For these routines, all graphic operations have been reduced to "spans."
A span is a horizontal row of pixels.
If you can design these routines which write into and read from
rows of pixels at a time, you can use the mi routines.</para>
<para>
Each routine takes
a destination drawable to draw into, a GC to use while drawing,
the number of spans to do, and two pointers to arrays that indicate the list
of starting points and the list of widths of spans.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->FillSpans(dst, pGC, nSpans, pPoints, pWidths, sorted)
                DrawablePtr dst;
                GCPtr pGC;
                int nSpans;
                DDXPointPtr pPoints;
                int *pWidths;
                int sorted;

</programlisting></blockquote>
FillSpans should fill horizontal rows of pixels with
the appropriate patterns, stipples, etc.,
based on the values in the GC.
The starting points are in the array at pPoints; the widths are in pWidths.
If sorted is true, the scan lines are in increasing y order, in which case
you may be able to make assumptions and optimizations.</para>
<para>
GC components: alu, clipOrg, clientClip, and fillStyle.</para>
<para>
GC mode-dependent components: fgPixel (for fillStyle Solid); tile, patOrg
(for fillStyle Tile); stipple, patOrg, fgPixel (for fillStyle Stipple);
and stipple, patOrg, fgPixel and bgPixel (for fillStyle OpaqueStipple).</para>
<para>
<blockquote><programlisting>

        void pGC->ops->SetSpans(pDrawable, pGC, pSrc, ppt, pWidths, nSpans, sorted)
                DrawablePtr pDrawable;
                GCPtr pGC;
                char *pSrc;
                DDXPointPtr pPoints;
                int *pWidths;
                int nSpans;
                int sorted;

</programlisting></blockquote>
For each span, this routine should copy pWidths bits from pSrc to
pDrawable at pPoints using the raster-op from the GC.
If sorted is true, the scan lines are in increasing y order.
The pixels in pSrc are
padded according to the screen's padding rules.
These
can be used to support
interesting extension libraries, for example, shaded primitives.   It does not
use the tile and stipple.</para>
<para>
GC components: alu, clipOrg, and clientClip</para>
<para>
The above functions are expected to handle all modifiers in the current
GC.  Therefore, it is expedient to have
different routines to quickly handle common special cases
and reload the procedure pointers
at validate time, as with the other output functions.</para>
<para>
<blockquote><programlisting>

        void pScreen->GetSpans(pDrawable, wMax, pPoints, pWidths, nSpans)
                DrawablePtr pDrawable;
                int wMax;
                DDXPointPtr pPoints;
                int *pWidths;
                int nSpans;
                char *pDst;

</programlisting></blockquote>
For each span, GetSpans gets bits from the drawable starting at pPoints
and continuing for pWidths bits.
Each scanline returned will be server-scanline padded.
The routine can return NULL if memory cannot be allocated to hold the
result.</para>
<para>
GetSpans never translates -- for a window, the coordinates are already
screen-relative.  Consider the case of hardware that doesn't do
translation: the mi code that calls ddX will translate each shape
(rectangle, polygon,. etc.) before scan-converting it, which requires
many fewer additions that having GetSpans translate each span does.
Conversely, consider hardware that does translate: it can set its
translation point to (0, 0) and get each span, and the only penalty is
the small number of additions required to translate each shape being
scan-converted by the calling code.  Contrast the behavior of
FillSpans and SetSpans (discussed above under miTranslate), which are
expected to be used more often.</para>
<para>
Thus, the penalty to hardware that does hardware translation is
negligible, and code that wants to call GetSpans() is greatly
simplified, both for extensions and the machine-independent core
implementation.</para>
<section>
  <title>Glyph Routines</title>
<para>
The Glyph routines draw individual character glyphs for text drawing requests.</para>
<para>
You have a choice in implementing these routines.  You can use the mi
versions; they depend ultimately upon the span routines.  Although
text drawing will work, it will be very slow.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PolyGlyphBlt(pDrawable, pGC, x, y, nglyph, ppci, pglyphBase)
                DrawablePtr pDrawable;
                GCPtr pGC;
                int x , y;
                unsigned int nglyph;
                CharInfoRec **ppci;             /* array of character info */
                pointer unused;                 /* unused since R5 */

</programlisting></blockquote>
GC components: alu, clipOrg, clientClip, font, and fillStyle.</para>
<para>
GC mode-dependent components: fgPixel (for fillStyle Solid); tile, patOrg
(for fillStyle Tile); stipple, patOrg, fgPixel (for fillStyle Stipple);
and stipple, patOrg, fgPixel and bgPixel (for fillStyle OpaqueStipple).</para>
<para>
<blockquote><programlisting>

        void pGC->ops->ImageGlyphBlt(pDrawable, pGC, x, y, nglyph, ppci, pglyphBase)
                DrawablePtr pDrawable;
                GCPtr pGC;
                int x , y;
                unsigned int nglyph;
                CharInfoRec **ppci;     /* array of character info */
                pointer unused;         /* unused since R5 */

</programlisting></blockquote>
GC components: clipOrg, clientClip, font, fgPixel, bgPixel</para>
<para>
These routines must copy the glyphs defined by the bitmaps in
pglyphBase and the font metrics in ppci to the DrawablePtr, pDrawable.
The poly routine follows all fill, stipple, and tile rules.  The image
routine simply blasts the glyph onto the glyph's rectangle, in
foreground and background colors.</para>
<para>
More precisely, the Image routine fills the character rectangle with
the background color, and then the glyph is applied in the foreground
color.  The glyph can extend outside of the character rectangle.
ImageGlyph() is used for terminal emulators and informal text purposes
such as button labels.</para>
<para>
The exact specification for the Poly routine is that the glyph is
painted with the current fill style.  The character rectangle is
irrelevant for this operation.  PolyText, at a higher level, includes
facilities for font changes within strings and such; it is to be used
for WYSIWYG word processing and similar systems.</para>
<para>
Both of these routines must clip themselves to the overall clipping region.</para>
<para>
Example implementations in mi are miPolyGlyphBlt() and
miImageGlyphBlt() in Xserver/mi/miglblt.c.</para>
</section>
<section>
<title>PushPixels routine</title>
<para>
The PushPixels routine writes the current fill style onto the drawable
in a certain shape defined by a bitmap.  PushPixels is equivalent to
using a second stipple.  You can thing of it as pushing the fillStyle
through a stencil.  PushPixels is not used by any of the mi rendering code,
but is used by the mi software cursor code.
<blockquote><para>
        Suppose the stencil is: 00111100
        and the stipple is:     10101010
        PushPixels result:      00101000
</para></blockquote>
</para>
<para>
You have a choice in implementing this routine.
You can use the mi version which depends ultimately upon FillSpans().
Although it will work, it will be slow.</para>
<para>
<blockquote><programlisting>

        void pGC->ops->PushPixels(pGC, pBitMap, pDrawable, dx, dy, xOrg, yOrg)
                GCPtr pGC;
                PixmapPtr pBitMap;
                DrawablePtr pDrawable;
                int dx, dy, xOrg, yOrg;

</programlisting></blockquote>
GC components: alu, clipOrg, clientClip, and fillStyle.</para>
<para>
GC mode-dependent components: fgPixel (for fillStyle Solid); tile, patOrg
(for fillStyle Tile); stipple, patOrg, fgPixel (for fillStyle Stipple);
and stipple, patOrg, fgPixel and bgPixel (for fillStyle OpaqueStipple).</para>
<para>
PushPixels applies the foreground color, tile, or stipple from the pGC
through a stencil onto pDrawable.  pBitMap points to a stencil (of
which we use an area dx wide by dy high), which is oriented over the
drawable at xOrg, yOrg.  Where there is a 1 bit in the bitmap, the
destination is set according to the current fill style.  Where there
is a 0 bit in the bitmap, the destination is left the way it is.</para>
<para>
This routine must clip to the overall clipping region.</para>
<para>
An Example implementation is miPushPixels() in Xserver/mi/mipushpxl.c.</para>
</section>
</section>
</section>
<section>
  <title>Shutdown Procedures</title>
<para>
<blockquote><programlisting>
        void ddxGiveUp(enum ExitCode error)
</programlisting></blockquote>
Some hardware may require special work to be done before the server
exits so that it is not left in an intermediate state.  As explained
in the OS layer, FatalError() will call ddxGiveUp() just before
terminating the server.  In addition, ddxGiveUp() will be called just
before terminating the server on a "clean" death.  What
ddxGiveUp does is left unspecified, only that it must exist in the
ddx layer.  It is up to local implementors as to what they should
accomplish before termination.</para>
<section>
  <title>Command Line Procedures</title>
<para>
<blockquote><programlisting>
        int ddxProcessArgument(argc, argv, i)
            int argc;
            char *argv[];
            int i;

        void
        ddxUseMsg()

</programlisting></blockquote>
You should write these routines to deal with device-dependent command line
arguments.  The routine ddxProcessArgument() is called with the command line,
and the current index into argv; you should return zero if the argument
is not a device-dependent one, and otherwise return a count of the number
of elements of argv that are part of this one argument.  For a typical
option (e.g., "-realtime"), you should return the value one.  This
routine gets called before checks are made against device-independent
arguments, so it is possible to peek at all arguments or to override
device-independent argument processing.  You can document the
device-dependent arguments in ddxUseMsg(), which will be
called from UseMsg() after printing out the device-independent arguments.</para>
</section>
</section>
<section id="wrappers_and_privates">
  <title>Wrappers and Privates</title>
<para>
Two new extensibility concepts have been developed for release 4, Wrappers
and devPrivates.  These replace the R3 GCInterest queues, which were not a
general enough mechanism for many extensions and only provided hooks into a
single data structure.  devPrivates have been revised substantially for
X.Org X server release 1.5, updated again for the 1.9 release and extended
again for the 1.13 relealse.</para>
<section>
  <title>devPrivates</title>
<para>
devPrivates provides a way to attach arbitrary private data to various server structures.
Any structure which contains a <structfield>devPrivates</structfield> field of
type <type>PrivateRec</type> supports this mechanism.  Some structures allow
allocating space for private data after some objects have been created, others
require all space allocations be registered before any objects of that type
are created.  <filename class="headerfile">Xserver/include/privates.h</filename>
lists which of these cases applies to each structure containing
<structfield>devPrivates</structfield>.</para>

<para>
To request private space, use
<blockquote><programlisting>
        Bool dixRegisterPrivateKey(DevPrivateKey key, DevPrivateType type, unsigned size);
</programlisting></blockquote>
The first argument is a pointer to a <type>DevPrivateKeyRec</type> which
will serve as the unique identifier for the private data.  Typically this is
the address of a static <type>DevPrivateKeyRec</type> in your code.
The second argument is the class of objects for which this key will apply.
The third argument is the size of the space being requested, or
<constant>0</constant> to only allocate a pointer that the caller will manage.
If space is requested, this space will be automatically freed when the object
is destroyed.  Note that a call to <function>dixSetPrivate</function>
that changes the pointer value may cause the space to be unreachable by the caller, however it will still be automatically freed.
The function returns <literal>TRUE</literal> unless memory allocation fails.
If the function is called more than once on the same key, all calls must use
the same value for <type>size</type> or the server will abort.</para>

<para>
To request per-screen private space in an object, use
<blockquote><programlisting>
        Bool dixRegisterScreenPrivateKey(DevScreenPrivateKey key, ScreenPtr pScreen, DevPrivateType type, unsigned size);
</programlisting></blockquote>
The <parameter>type</parameter> and <parameter>size</parameter> arguments are
the same as those to <function>dixRegisterPrivateKey</function> but this
function ensures the given <parameter>key</parameter> exists on objects of
the specified type with distinct storage for the given
<parameter>pScreen</parameter>. The key is usable on ScreenPrivate variants
that are otherwise equivalent to the following Private functions.</para>

<para>
  To request private space in objects created for a specific screen, use
  <blockquote><programlisting>
    Bool dixRegisterScreenSpecificPrivateKey(ScreenPtr pScreen, DevPrivateKey key, DevPrivateType type, unsigned size);
  </programlisting></blockquote>
  The <parameter>type</parameter> and <parameter>size</parameter> arguments are
  the same as those to <function>dixRegisterPrivateKey</function> but this
  function ensures only that the given <parameter>key</parameter> exists on objects of
  the specified type that are allocated with reference to the specified
  <parameter>pScreen</parameter>. Using the key on objects allocated for
  other screens will result in incorrect results; there is no check made to
  ensure that the caller's screen matches the private's screen. The key is
  usable in any of the following functions. Screen-specific private storage is available
  only for Windows, GCs, Pixmaps and Pictures. Attempts to allocate screen-specific
  privates on other objects will result in a call to FatalError.
</para>

<para>
To attach a piece of private data to an object, use:
<blockquote><programlisting>
        void dixSetPrivate(PrivateRec **privates, const DevPrivateKey key, pointer val)
</programlisting></blockquote>
The first argument is the address of the <structfield>devPrivates</structfield>
field in the target structure.  This field is managed privately by the DIX
layer and should not be directly modified.  The second argument is a pointer
to the <type>DevPrivateKeyRec</type> which you registered with
<function>dixRegisterPrivateKey</function> or allocated with
<function>dixCreatePrivateKey</function>.  Only one
piece of data with a given key can be attached to an object, and in most cases
each key is specific to the type of object it was registered for.   (An
exception is the PRIVATE_XSELINUX class which applies to multiple object types.)
The third argument is the value to store.</para>
<para>
If private data with the given key is already associated with the object,
<function>dixSetPrivate</function> will overwrite the old value with the
new one.</para>

<para>
To look up a piece of private data, use one of:
<blockquote><programlisting>
        pointer dixLookupPrivate(PrivateRec **privates, const DevPrivateKey key)
        pointer *dixLookupPrivateAddr(PrivateRec **privates, const DevPrivateKey key)
</programlisting></blockquote>
The first argument is the address of the <structfield>devPrivates</structfield> field
in the target structure.  The second argument is the key to look up.
If a non-zero size was given when the key was registered, or if private data
with the given key is already associated with the object, then
<function>dixLookupPrivate</function> will return the pointer value
while <function>dixLookupPrivateAddr</function>
will return the address of the pointer.</para>

<para>
When implementing new server resource objects that support devPrivates, there
are four steps to perform:
Add a type value to the <type>DevPrivateType</type> enum in
<filename class="headerfile">Xserver/include/privates.h</filename>,
declare a field of type <type>PrivateRec *</type> in your structure;
initialize this field to <literal>NULL</literal> when creating any objects; and
when freeing any objects call the <function>dixFreePrivates</function> or
<function>dixFreeObjectWithPrivates</function> function.</para>
</section>
<section>
  <title>Wrappers</title>
<para>
Wrappers are not a body of code, nor an interface spec.  They are, instead,
a technique for hooking a new module into an existing calling sequence.
There are limitations on other portions of the server implementation which
make using wrappers possible; limits on when specific fields of data
structures may be modified.  They are intended as a replacement for
GCInterest queues, which were not general enough to support existing
modules; in particular software cursors needed more
control over the activity.  The general mechanism for using wrappers is:
<blockquote><programlisting>
privateWrapperFunction (object, ...)
        ObjectPtr       object;
{
        pre-wrapped-function-stuff ...

        object->functionVector = dixLookupPrivate(&amp;object->devPrivates, privateKey);
        (*object->functionVector) (object, ...);
        /*
         * this next line is occasionally required by the rules governing
         * wrapper functions.  Always using it will not cause problems.
         * Not using it when necessary can cause severe troubles.
         */
        dixSetPrivate(&amp;object->devPrivates, privateKey, object->functionVector);
        object->functionVector = privateWrapperFunction;

        post-wrapped-function-stuff ...
}

privateInitialize (object)
        ObjectPtr       object;
{
        dixSetPrivate(&amp;object->devPrivates, privateKey, object->functionVector);
        object->functionVector = privateWrapperFunction;
}
</programlisting></blockquote>
</para>
<para>
Thus the privateWrapperFunction provides hooks for performing work both
before and after the wrapped function has been called; the process of
resetting the functionVector is called "unwrapping" while the process of
fetching the wrapped function and replacing it with the wrapping function
is called "wrapping".  It should be clear that GCInterest queues could
be emulated using wrappers.  In general, any function vectors contained in
objects can be wrapped, but only vectors in GCs and Screens have been tested.</para>
<para>
Wrapping screen functions is quite easy; each vector is individually
wrapped.  Screen functions are not supposed to change after initialization,
so rewrapping is technically not necessary, but causes no problems.</para>
<para>
Wrapping GC functions is a bit more complicated.  GC's have two tables of
function vectors, one hanging from gc->ops and the other from gc->funcs, which
should be initially wrapped from a CreateGC wrapper.  Wrappers should modify
only table pointers, not the contents of the tables, as they
may be shared by more than one GC (and, in the case of funcs, are probably
shared by all gcs).  Your func wrappers may change the GC funcs or ops
pointers, and op wrappers may change the GC op pointers but not the funcs.</para>
<para>
Thus, the rule for GC wrappings is: wrap the funcs from CreateGC and, in each
func wrapper, unwrap the ops and funcs, call down, and re-wrap.  In each op
wrapper, unwrap the ops, call down, and rewrap afterwards.  Note that in
re-wrapping you must save out the pointer you're replacing again.  This way the
chain will be maintained when wrappers adjust the funcs/ops tables they use.</para>
</section>
</section>
<section>
    <title>Work Queue</title>
<para>
To queue work for execution when all clients are in a stable state (i.e.
just before calling select() in WaitForSomething), call:
<blockquote><programlisting>
        Bool QueueWorkProc(function,client,closure)
                Bool            (*function)();
                ClientPtr       client;
                pointer         closure;
</programlisting></blockquote>
</para>
<para>
When the server is about to suspend itself, the given function will be
executed:
<blockquote><programlisting>
        (*function) (client, closure)
</programlisting></blockquote>
</para>
<para>
Neither client nor closure are actually used inside the work queue routines.</para>
</section>
</section>
<section>
  <title>Summary of Routines</title>
<para>
This is a summary of the routines discussed in this document.
The procedure names are in alphabetical order.
The Struct is the structure it is attached to; if blank, this
procedure is not attached to a struct and must be named as shown.
The sample server provides implementations in the following
categories.  Notice that many of the graphics routines have both
mi and fb implementations.</para>
<para>
<itemizedlist>
<listitem><para>dix     portable to all systems; do not attempt to rewrite (Xserver/dix)</para></listitem>
<listitem><para>os      routine provided in Xserver/os or Xserver/include/os.h</para></listitem>
<listitem><para>ddx     frame buffer dependent (examples in Xserver/fb)</para></listitem>
<listitem><para>mi      routine provided in Xserver/mi</para></listitem>
<listitem><para>hd      hardware dependent (examples in many Xserver/hw directories)</para></listitem>
<listitem><para>none    not implemented in sample implementation</para></listitem>
</itemizedlist>
</para>
        <table frame="all" id="routines-1">
          <title>Server Routines (Page 1)</title>
          <tgroup cols='3' align='left' colsep='1' rowsep='1'>
            <thead>
              <row>
                <entry>Procedure</entry>
                <entry>Port</entry>
                <entry>Struct</entry>
              </row>
            </thead>
            <tbody>
<row><entry><function>ALLOCATE_LOCAL</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>AddCallback</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>AddEnabledDevice</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>AddInputDevice</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>AddScreen</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>AdjustWaitForDelay</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>Bell</function></entry><entry><literal>hd</literal></entry><entry><para>Device</para></entry></row>
<row><entry><function>ChangeClip</function></entry><entry><literal>mi</literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>ChangeGC</function></entry><entry><literal></literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>ChangeWindowAttributes</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>ClearToBackground</function></entry><entry><literal>ddx</literal></entry><entry><para>Window</para></entry></row>
<row><entry><function>ClientAuthorized</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>ClientSignal</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>ClientSleep</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>ClientWakeup</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>ClipNotify</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>CloseScreen</function></entry><entry><literal>hd</literal></entry><entry><para></para></entry></row>
<row><entry><function>ConstrainCursor</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>CopyGCDest</function></entry><entry><literal>ddx</literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>CopyGCSource</function></entry><entry><literal>none</literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>CopyWindow</function></entry><entry><literal>ddx</literal></entry><entry><para>Window</para></entry></row>
<row><entry><function>CreateGC</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>CreateCallbackList</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>CreatePixmap</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>CreateScreenResources</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>CreateWellKnowSockets</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>CreateWindow</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>CursorLimits</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>DEALLOCATE_LOCAL</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>DeleteCallback</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>DeleteCallbackList</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>DestroyClip</function></entry><entry><literal>ddx</literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>DestroyGC</function></entry><entry><literal>ddx</literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>DestroyPixmap</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>DestroyWindow</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>DisplayCursor</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>Error</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>ErrorF</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>FatalError</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>FillPolygon</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>FillSpans</function></entry><entry><literal>ddx</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>FlushAllOutput</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>FlushIfCriticalOutputPending</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>FreeScratchPixmapHeader</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>GetMotionEvents</function></entry><entry><literal>hd</literal></entry><entry><para>Device</para></entry></row>
<row><entry><function>GetScratchPixmapHeader</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>GetSpans</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>GetStaticColormap</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
            </tbody>
          </tgroup>
        </table>

        <table frame="all" id="routines-2">
          <title>Server Routines (Page 2)</title>
          <tgroup cols='3' align='left' colsep='1' rowsep='1'>
            <thead>
              <row>
                <entry>Procedure</entry>
                <entry>Port</entry>
                <entry>Struct</entry>
              </row>
            </thead>
            <tbody>
<row><entry><function>ImageGlyphBlt</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>ImageText16</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>ImageText8</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>InitInput</function></entry><entry><literal>hd</literal></entry><entry><para></para></entry></row>
<row><entry><function>InitKeyboardDeviceStruct</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>InitOutput</function></entry><entry><literal>hd</literal></entry><entry><para></para></entry></row>
<row><entry><function>InitPointerDeviceStruct</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>InsertFakeRequest</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>InstallColormap</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>Intersect</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>Inverse</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>LineHelper</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>ListInstalledColormaps</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>LookupKeyboardDevice</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>LookupPointerDevice</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>ModifyPixmapHeader</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>NextAvailableClient</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>OsInit</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>PaintWindowBackground</function></entry><entry><literal>mi</literal></entry><entry><para>Window</para></entry></row>
<row><entry><function>PaintWindowBorder</function></entry><entry><literal>mi</literal></entry><entry><para>Window</para></entry></row>
<row><entry><function>PointerNonInterestBox</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>PointInRegion</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>PolyArc</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyFillArc</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyFillRect</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyGlyphBlt</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>Polylines</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyPoint</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyRectangle</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolySegment</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyText16</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PolyText8</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>PositionWindow</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>ProcessInputEvents</function></entry><entry><literal>hd</literal></entry><entry><para></para></entry></row>
<row><entry><function>PushPixels</function></entry><entry><literal>mi</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>QueryBestSize</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>ReadRequestFromClient</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>RealizeCursor</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RealizeFont</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RealizeWindow</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RecolorCursor</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RectIn</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionCopy</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionCreate</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionDestroy</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionEmpty</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionExtents</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionNotEmpty</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>RegionReset</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>ResolveColor</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
            </tbody>
          </tgroup>
        </table>

        <table frame="all" id="routines-3">
          <title>Server Routines (Page 3)</title>
          <tgroup cols='3' align='left' colsep='1' rowsep='1'>
            <thead>
              <row>
                <entry>Procedure</entry>
                <entry>Port</entry>
                <entry>Struct</entry>
              </row>
            </thead>
            <tbody>
<row><entry><function>RemoveEnabledDevice</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>ResetCurrentRequest</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>SaveScreen</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>SetCriticalOutputPending</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>SetCursorPosition</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>SetInputCheck</function></entry><entry><literal>dix</literal></entry><entry><para></para></entry></row>
<row><entry><function>SetSpans</function></entry><entry><literal>ddx</literal></entry><entry><para>GC op</para></entry></row>
<row><entry><function>StoreColors</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>Subtract</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>TimerCancel</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>TimerCheck</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>TimerForce</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>TimerFree</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>TimerInit</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>TimerSet</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>TimeSinceLastInputEvent</function></entry><entry><literal>hd</literal></entry><entry><para></para></entry></row>
<row><entry><function>TranslateRegion</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>UninstallColormap</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>Union</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>UnrealizeCursor</function></entry><entry><literal>hd</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>UnrealizeFont</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>UnrealizeWindow</function></entry><entry><literal>ddx</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>ValidateGC</function></entry><entry><literal>ddx</literal></entry><entry><para>GC func</para></entry></row>
<row><entry><function>ValidateTree</function></entry><entry><literal>mi</literal></entry><entry><para>Screen</para></entry></row>
<row><entry><function>WaitForSomething</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
<row><entry><function>WindowExposures</function></entry><entry><literal>mi</literal></entry><entry><para>Window</para></entry></row>
<row><entry><function>WriteToClient</function></entry><entry><literal>os</literal></entry><entry><para></para></entry></row>
            </tbody>
          </tgroup>
        </table>
</section>
</article>