Definition of the Porting Layer for the X v11 Sam-
			 ple 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































Porting Layer Definition    - 1 -	       April 8, 1994
























Copyright (C) 1994 X Consortium

Permission is hereby granted, free of charge, to any person
obtaining a copy of this software and associated documenta-
tion files (the ``Software''), to deal in the Software with-
out 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 fol-
lowing conditions:

The above copyright notice and this permission notice shall
be included in all copies or substantial portions of the
Software.

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 PUR-
POSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE X CONSOR-
TIUM 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.


















Porting Layer Definition    - 2 -	       April 8, 1994








The following document explains the structure of the X Win-
dow System display server and the interfaces among the
larger pieces.	It is intended as a reference for program-
mers who are implementing an X Display Server on their work-
station 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:


1)   Read the first section of the "Strategies for Porting"
     document (Overview of Porting Process).


2)   Skim over this document (the Definition document).


3)   Skim over the remainder of the Strategies document.


4)   Start planning and working, referring to the Strategies
     and Definition documents.

     You may also want to look at the following documents:

o    "The X Window System" for an overview of X.

o    "Xlib - C Language X Interface" for a view of what the
     client programmer sees.

o    "X Window System Protocol" for a terse description of
     the byte stream protocol between the client and server.

LK201 and DEC are trademarks of Digital Equipment Corpora-
tion.  Macintosh and Apple are trademarks of Apple Computer,
Inc.  PostScript is a trademark of Adobe Systems, Inc.	Eth-
ernet is a trademark of Xerox Corporation.  X Window System
is a trademark of X Consortium, Inc.  Cray is a trademark of
Cray Research, Inc.


To understand this document and the accompanying source
code, you should know the C language.  You should be famil-
iar with 2D graphics and windowing concepts such as clip-
ping, bitmaps, fonts, etc.  You should have a general knowl-
edge of the X Window System.  To implement the server code
on your hardware, you need to know a lot about your hard-
ware, its graphic display device(s), and (possibly) its net-
working and multitasking facilities.

This document depends a lot on the source code, so you
should have a listing of the code handy.





Porting Layer Definition    - 3 -	       April 8, 1994








Some source on the distribution tape is directly compilable
on your machine.  Some of it will require modification.
Other parts may have to be completely written from scratch.

The tape also includes source for a sample implementation of
a display server which runs on a variety of color and
monochrome displays which you will find useful for imple-
menting any type of X server.



1.  The X Window System

The X Window System, or simply "X," is a windowing system
that provides high-performance, high-level, device-indepen-
dent graphics.

X is a windowing system designed for bitmapped graphic dis-
plays.	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.)

X is designed for a networking environment where users can
run applications on machines other than their own worksta-
tions.	Sometimes, the connection is over an Ethernet net-
work 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 com-
pression techniques.

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 gen-
eral, it can be the foundation for a wide variety of inter-
active software.

For a more detailed overview, see the document "The X Window
System."  For details on the byte stream protocol, see "X
Window System protocol."


2.  OVERVIEW OF THE SERVER

The display server manages windows and simple graphics
requests for the user on behalf of different client applica-
tions.	The client applications can be running on any
machine on the network.  The server mainly does three
things:





Porting Layer Definition    - 4 -	       April 8, 1994








o    Responds to protocol requests from existing clients
     (mostly graphic and text drawing commands)

o    Sends device input (keystrokes and mouse actions) and
     other events to existing clients

o    Maintains client connections


The server code is organized into four major pieces:


o    Device Independent (DIX) layer - code shared among all
     implementations

o    Operating System (OS) layer - code that is different
     for each operating system but is shared among all
     graphic devices for this operating system

o    Device Dependent (DDX) layer - code that is (poten-
     tially) different for each combination of operating
     system and graphic device

o    Extension Interface - a standard way to add features to
     the X server


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 docu-
mented.  Extensions incorporate new functionality into the
server; and require additional functionality over a simple
DDX.

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.


2.1.  Notes On Resources and Large Structs

X resources are C structs inside the server.  Client appli-
cations 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 inte-
gers that are sent over the network.  Within the server, of
course, they are just C structs, and we refer to them by
pointers.




Porting Layer Definition    - 5 -	       April 8, 1994








The DDX layer has several kinds of resources:

o    Window

o    Pixmap

o    Screen

o    Device

o    Colormap

o    Font

o    Cursor

o    Graphics Contexts

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."

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.

Three kinds of fields are in these structs:

o    Attribute fields - struct fields that contain values
     like normal structs

o    Pointers to procedures, or structures of procedures,
     that operate on the object

o    A private field (or two) used by your DDX code to keep
     private data (probably a pointer to another data struc-
     ture), or an array of private fields, which is sized as
     the server initializes.

DIX calls through the struct's procedure pointers to do its
tasks.	These procedures are set either directly or indi-
rectly 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

	(* pScreen->CreatePixmap)(pScreen, width, height, depth).

All procedure pointers must be set to some routine unless



Porting Layer Definition    - 6 -	       April 8, 1994








noted otherwise; a null pointer will have unfortunate conse-
quences.

Procedure routines will be indicated in the documentation by
this convention:

     void pScreen->MyScreenRoutine(arg, arg, ...)

as opposed to a free routine, not in a data structure:

     void MyFreeRoutine(arg, arg, ...)


The attribute fields are mostly set by DIX; DDX should not
modify them unless noted otherwise.


3.  DIX LAYER

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.

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 ver-
sion.

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 pro-
cess the asynchronous requests that it needs to process.
Most of these operations are performed by OS and DDX rou-
tines that you must supply.


4.  OS LAYER

This part of the source consists of a few routines that you
have to rewrite for each operating system.  These OS func-
tions 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.



Porting Layer Definition    - 7 -	       April 8, 1994








     void OsInit()
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.


4.1.  Scheduling and Request Delivery

The main dispatch loop in DIX creates the illusion of multi-
tasking 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.

You must supply some of the pieces for proper scheduling
between clients.

     int WaitForSomething(pClientReady)
	  int *pClientReady;

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:

o    There is an input event from the user or hardware (see
     SetInputCheck())

o    There are requests waiting from known clients, in which
     case you should return a count of clients stored in
     pClientReady

o    A new client tries to connect, in which case you should
     create the client and  then continue waiting

Before WaitForSomething() computes the masks to pass to
select, it needs to see if there is anything to do on the
work queue; if so, it must call a DIX routine called Pro-
cessWorkQueue.
     extern WorkQueuePtr workQueue;

     if (workQueue)
	  ProcessWorkQueue ();

If WaitForSomething() decides it is about to do something
that might block (in the sample server,  before it calls
select()) it must call a DIX routine called BlockHandler().

     void BlockHandler(pTimeout, pReadmask)



Porting Layer Definition    - 8 -	       April 8, 1994








	  pointer pTimeout;
	  pointer pReadmask;
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,  and the pReadmask is a pointer to the information
describing the descriptors that will be waited on.

In the sample server,  pTimeout is a struct timeval **,  and
pReadmask is the address of the select() mask for reading.

The DIX BlockHandler() iterates through the Screens,  for
each one calling its BlockHandler.  A BlockHandler is
declared thus:

     void xxxBlockHandler(nscreen, pbdata, pptv, pReadmask)
	  int nscreen;
	  pointer pbdata;
	  struct timeval ** pptv;
	  pointer pReadmask;
The arguments are the index of the Screen,  the blockData
field of the Screen,  and the arguments to the DIX BlockHan-
dler().

Immediately after WaitForSomething returns from the block,
even if it didn't actually block,  it must call the DIX rou-
tine WakeupHandler().

     void WakeupHandler(result, pReadmask)
	  int result;
	  pointer pReadmask;

Once again,  the types are not specified by DIX.  The result
is the success indicator for the thing that (may have)
blocked, and the pReadmask is a mask of the descriptors that
came active.  In the sample server,  result is the result
from select(),	and pReadmask is the address of the select()
mask for reading.

The DIX WakeupHandler() calls each Screen's WakeupHandler.
A WakeupHandler is declared thus:

     void xxxWakeupHandler(nscreen, pbdata, err, pReadmask)
	  int nscreen;
	  pointer pbdata;
	  unsigned long result;
	  pointer pReadmask;
The arguments are the index of the Screen,  the blockData
field of the Screen,  and the arguments to the DIX Wake-
upHandler().

In addition to the per-screen BlockHandlers, any module may
register block and wakeup handlers (only together) using:




Porting Layer Definition    - 9 -	       April 8, 1994








     Bool RegisterBlockAndWakeupHandlers (blockHandler, wakeupHandler, blockData)
	  BlockHandlerProcPtr	 blockHandler;
	  WakeupHandlerProcPtr	 wakeupHandler;
	  pointer blockData;
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:

     RemoveBlockAndWakeupHandlers (blockHandler, wakeupHandler, blockData)
	  BlockHandlerProcPtr	blockHandler;
	  WakeupHandlerProcPtr	wakeupHandler;
	  pointer blockData;
All three arguments must match the values passed to Regis-
terBlockAndWakeupHandlers.

These registered block handlers are called after the per-
screen handlers:

     void (*BlockHandler) (blockData, pptv, pReadmask)
	  pointer   blockData;
	  OSTimePtr pptv;
	  pointer   pReadmask;

Any wakeup handlers registered with RegisterBlockAndWake-
upHandlers will be called before the Screen handlers:

     void (*WakeupHandler) (blockData, err, pReadmask)
	  pointer   blockData;
	  int err;
	  pointer pReadmask;

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.

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 ser-
vice some of them for a given client before going on to the
next client (see isItTimeToYield, below).  Therefore, Wait-
ForSomething() will have to report that these same clients
still have requests queued up the next time around.

An implementation should return information on as many out-
standing things as it can.  For instance, if your implemen-
tation 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 applica-
tion that constantly barrages the server with graphics



Porting Layer Definition   - 10 -	       April 8, 1994








drawing requests.

A list of indexes (client->index) for clients with data
ready to be read or processed should be returned in pClien-
tReady, 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 ReadRequest-
FromClient(), below.

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:

     ClientPtr NextAvailableClient(ospriv)
	  pointer ospriv;
This routine returns NULL if a new client cannot be allo-
cated (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 con-
tains the number of the socket for this client. See also
"New Client Connections."  NextAvailableClient() will call
InsertFakeRequest(), so you must be prepared for this.

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 ProcessIn-
putEvents() until the SetInputCheck() locations are equal.

The sample server contains an implementation of WaitForSome-
thing().  The following two routines indicate to WaitFor-
Something() what devices should be waited for.	 fd is an OS
dependent type; in the sample server it is an open file
descriptor.

     int AddEnabledDevice(fd)
	  int fd;

     int RemoveEnabledDevice(fd)
	  int fd;
These two routines are usually called by DDX from the ini-
tialize 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.

4.2.  New Client Connections

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.



Porting Layer Definition   - 11 -	       April 8, 1994








The first thing that happens is a client initiates a connec-
tion 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 worksta-
tion 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 termi-
nal.  Fred types in a command to run an X client application
that was linked with Xlib.  Xlib looks at the shell environ-
ment variable DISPLAY, which has the value "fredsbit-
tube:0.0."  The host name of Fred's workstation is "freds-
bittube," 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.)

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.

All X servers on TCP listen for new clients on port 6000 by
default; this is known as a "well-known port" in IP termi-
nology.

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.

Actually, it is a bit more complicated than that.  Each X
server process running on the host machine is called a "dis-
play."	Each display can have more than one screen that it
manages.  "corporatehydra:3.2" represents screen 2 on dis-
play 3 on the multi-screened network node corporatehydra.
The open request would be sent on well-known port number
6003.

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 ver-
sion 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 autho-
rization 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.

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,



Porting Layer Definition   - 12 -	       April 8, 1994








it closes any connection that are too old.  The connection
timeout can be set from the command line.

You must implement whatever authorization schemes you want
to support.  The sample server on the distribution tape sup-
ports a simple authorization scheme.  The only interface
seen by DIX is:

     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;

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.

Accepting new connections happens internally to WaitForSome-
thing().  WaitForSomething() must call the DIX routine Nex-
tAvailableClient() 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 informa-
tion 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 osPri-
vate field in the ClientPtr to store the file descriptor for
the socket, the input and output buffers, and authorization
information.

To initialize the methods you choose to allow clients to
connect to your server, main() calls the routine

     void CreateWellKnownSockets()

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:

     void ResetWellKnownSockets()
Sample implementations of both of these routines are found
in Xserver/os/connection.c.

For more details, see the section called "Connection Setup"
in the X protocol specification.








Porting Layer Definition   - 13 -	       April 8, 1994








4.3.  Reading Data from Clients

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.

     int ReadRequestFromClient(who)
	  ClientPtr who;

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 mea-
sured in 32-bit words, not in bytes.  Therefore, the theo-
retical 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 maxi-
mum 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.

The 'who' argument is the ClientPtr returned from WaitFor-
Something.  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.

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.

The sample server implementation is in Xserver/os/io.c.


DIX can insert data into the client stream, and can cause a
"replay" of the current request.

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

     int ResetCurrentRequest(client)
	 ClientPtr client;




Porting Layer Definition   - 14 -	       April 8, 1994








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 con-
nection 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 exe-
cuting 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 (isItTime-
ToYield).


4.4.  Sending Events, Errors And Replies To Clients


     int WriteToClient(who, n, buf)
	  ClientPtr who;
	  int n;
	  char *buf;
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.

     void SendErrorToClient(client, majorCode, minorCode, resId, errorCode)
	 ClientPtr client;
	 unsigned int majorCode;
	 unsigned int minorCode;
	 XID resId;
	 int errorCode;
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.


     void FlushAllOutput()

     void FlushIfCriticalOutputPending()

     void SetCriticalOutputPending()
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



Porting Layer Definition   - 15 -	       April 8, 1994








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.

4.5.  Font Support

In the sample server, fonts are encoded in disk files or
fetched from the font server.  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."

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.

With the incorporation of font-server based fonts and the
Speedo donation from Bitstream, the font interfaces have
been moved into a separate library, now called the Font
Library (../fonts/lib).  These routines are shared between
the X server and the Font server, so instead of this docu-
ment 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

4.6.  Memory Management

Memory management is based on functions in the C runtime
library.  Xalloc(), Xrealloc(), and Xfree() work just like
malloc(), realloc(), and free(), except that you can pass a
null pointer to Xrealloc() to have it allocate anew or pass
a null pointer to Xfree() and nothing will happen.  The ver-
sions in the sample server also do some checking that is
useful for debugging.  Consult a C runtime library reference
manual for more details.

The macros ALLOCATE_LOCAL and DEALLOCATE_LOCAL are provided
in Xserver/include/os.h.  These are useful if your compiler
supports alloca() (or some method of allocating memory from
the stack); and are defined appropriately on systems which
support it.

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.



Porting Layer Definition   - 16 -	       April 8, 1994








4.7.  Client Scheduling

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 wait-
ing for a long-term network activity (like loading a font)
before continuing with the first client.
     Bool isItTimeToYield;

isItTimeToYield is a global variable you can set if you want
to tell DIX to end the client's "time slice" and start pay-
ing attention to the next client.  After the current request
is finished, DIX will move to the next client.

In the sample server, ReadRequestFromClient() sets isItTime-
ToYield after 10 requests packets in a row are read from the
same client.

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 switchs
between different clients too quickly, performance may suf-
fer due to too much switching between contexts.  For exam-
ple, 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 perfor-
mance is impacted.

See the Strategies document for heuristics on setting isIt-
TimeToYield.

The following functions provide the ability to suspend
request processing on a particular client, resuming it at
some later time:

     int IgnoreClient (who)
	  ClientPtr who;

     int AttendClient (who)
	  ClientPtr who;
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.

Three functions support "process control" for X clients:




Porting Layer Definition   - 17 -	       April 8, 1994








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


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 Client-
Wakeup is called.

     Bool ClientSignal (client)
	  ClientPtr client;


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:

     ClientWakeup (client)
	  ClientPtr client;


This re-enables X request processing for the specified
client.

4.8.  Other OS Functions

     void
     ErrorF(char *f, ...)

     void
     FatalError(char *f, ...)

     void
     Error(str)
	 char *str;

You should write these three routines to provide for diag-
nostic output from the dix and ddx layers, although imple-
menting them to produce no output will not affect the cor-
rectness of your server.  ErrorF() and FatalError() take a
printf() type of format specification in the first argument
and an implementation-dependent number of arguments follow-
ing that.  Normally, the formats passed to ErrorF() and
FatalError() should be terminated with a newline.  Error()
provides an os interface for printing out the string passed
as an argument followed by a meaningful explanation of the
last system error.  Normally the string does not contain a
newline, and it is only called by the ddx layer.  In the
sample implementation, Error() uses the perror() function.





Porting Layer Definition   - 18 -	       April 8, 1994








After printing the message arguments, FatalError() must be
implemented such that the server will call AbortDDX() to
give the ddx layer a chance to reset the hardware, and then
terminate the server; it must not return.

The sample server implementation for these routines is in
Xserver/os/util.c.

4.9.  Idiom Support

The DBE specification introduces the notion of idioms, which
are groups of X requests which can be executed more effi-
ciently when taken as a whole compared to being performed
individually and sequentially.	This following server inter-
nal support to allows DBE implementations, as well as other
parts of the server, to do idiom processing.


     xReqPtr PeekNextRequest(xReqPtr req, ClientPtr client, Bool readmore)

If req is NULL, the return value will be a pointer to the
start of the complete request that follows the one currently
being executed for the client.	If req is not NULL, the
function assumes that req is a pointer to a request in the
client's request buffer, and the return value will be a
pointer to the the start of the complete request that fol-
lows req.  If the complete request is not available, the
function returns NULL; pointers to partial requests will
never be returned.  If (and only if) readmore is TRUE,
PeekNextRequest should try to read an additional request
from the client if one is not already available in the
client's request buffer.  If PeekNextRequest reads more data
into the request buffer, it should not move or change the
existing data.


     void SkipRequests(xReqPtr req, ClientPtr client, int numskipped)

The requests for the client up to and including the one
specified by req will be skipped.  numskipped must be the
number of requests being skipped.  Normal request processing
will resume with the request that follows req.	The caller
must not have modified the contents of the request buffer in
any way (e.g., by doing byte swapping in place).

Additionally, two macros in os.h operate on the xReq pointer
returned by PeekNextRequest:


     int ReqLen(xReqPtr req, ClientPtr client)

The value of ReqLen is the request length in bytes of the
given xReq.




Porting Layer Definition   - 19 -	       April 8, 1994








     otherReqTypePtr CastxReq(xReq *req, otherReqTypePtr)

The value of CastxReq is the conversion of the given request
pointer to an otherReqTypePtr (which should be a pointer to
a protocol structure type).  Only those fields which come
after the length field of otherReqType may be accessed via
the returned pointer.

Thus the first two fields of a request, reqType and data,
can be accessed directly using the xReq * returned by
PeekNextRequest.  The next field, the length, can be
accessed with ReqLen.  Fields beyond that can be accessed
with CastxReq.	This complexity was necessary because of the
reencoding of core protocol that can happen due to the
BigRequests extension.

5.  DDX LAYER

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.

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 deter-
mined 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.


5.1.  INPUT

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 inter-
faces 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 abstrac-
tion is hidden from view for core devices by DIX.

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.







Porting Layer Definition   - 20 -	       April 8, 1994








5.1.1.	Input Device Data Structures

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.

The DeviceRec (Xserver/include/input.h) is a device-indepen-
dent structure that contains the state of an input device.
A DevicePtr is simply a pointer to a DeviceRec.

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 vari-
ants overlap, so that the struct is actually very small in
memory.


5.1.2.	Processing Events

The main DDX input interface is the following routine:

     void ProcessInputEvents()
You must write this routine to deliver input events from the
user.  DIX calls it when input is pending (see next sec-
tion), 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 fol-
lowing routine:

     void DevicePtr->processInputProc(pEvent, device, count)
	      xEventPtr events;
	      DeviceIntPtr device;
	      int count;
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 > 1).

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 pro-
cessInputProc as many times as needed to flush its internal
queue.

event is an xEvent struct you pass to the input proc.  When



Porting Layer Definition   - 21 -	       April 8, 1994








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.

You should deliver the events in the same order that they
were generated.

For keyboard and pointing devices the xEvent variant should
be keyButtonPointer.  Fill in the following fields in the
xEvent record:

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

The rest of the fields are filled in by DIX.

The time stamp is maintained by your code in the DDX layer,
and it is your responsibility to stamp all events correctly.

The x and y coordinates of the pointing device and the time
must be filled in for all event types including keyboard
events.

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.

There are quite a collection of sample implementations of
this routine, one for each supported device.


5.1.3.	Telling DIX When Input is Pending

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().

This check for pending input must be very quick; a procedure
call is too slow.  The code that does the check is a hard-
wired IF statement in DIX code that simply compares the val-
ues pointed to by two pointers.  If the values are differ-
ent, then it assumes that input is pending and ProcessIn-
putEvents() is called by DIX.





Porting Layer Definition   - 22 -	       April 8, 1994








You must pass pointers to DIX to tell it what values to com-
pare.  The following procedure is used to set these point-
ers:

     void SetInputCheck(p1, p2)
	  long *p1, *p2;

You should call it sometime during initialization to indi-
cate 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.

These two pointers are initialized by DIX to point to arbi-
trary 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.

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.

The DDX layer of the sample server calls SetInputCheck()
once when the server's private internal queue is initial-
ized.  It passes pointers to the queue's head and tail.  See
Xserver/mi/mieq.c.

     int TimeSinceLastInputEvent()
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.


5.1.4.	Controlling Input Devices

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.


     int pInternalDevice->valuator->GetMotionProc(pdevice, coords, start, stop, pScreen)
	  DeviceIntPtr pdevice;
	  xTimecoord * coords;
	  unsigned long start;
	  unsigned long stop;
	  ScreenPtr pScreen;
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 num-
ber of motion events returned.	If there is no motion events



Porting Layer Definition   - 23 -	       April 8, 1994








support, this routine should do nothing and return zero.
The maximum number of coords to return is set in InitPoint-
erDeviceStruct(), below.

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.

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

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


You write two versions of this procedure, one for the key-
board and one for the pointing device.	DIX calls it to
inform DDX when a client has requested changes in the cur-
rent 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.


5.1.5.	Input Initialization

Input initialization is a bit complicated.  It all starts
with InitInput(), a routine that you write to call AddInput-
Device() twice (once for pointing device and once for key-
board.)  You also want to call RegisterKeyboardDevice() and
RegisterPointerDevice() on them.

When you Add the devices, a routine you supply for each
device gets called to initialize them.	Your individual ini-
tialize 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.



Porting Layer Definition   - 24 -	       April 8, 1994








     void InitInput(argc, argv)
	 int argc;
	 char **argv;

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.  In addition, you
must register the main keyboard and pointing devices by
calling RegisterPointerDevice() and RegisterKeyboardDe-
vice().

     DevicePtr AddInputDevice(deviceProc, autoStart)
	  DeviceProc deviceProc;
	  Bool autoStart;

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.

Note also that except for the main keyboard and pointing
device, an extension is needed to provide for a client
interface to a device.

     void RegisterPointerDevice(device)
	  DevicePtr device;

RegisterPointerDevice is a DIX routine that your DDX code
calls that makes that device the main pointing device.	This
routine is called once upon initialization and cannot be
called again.

     void RegisterKeyboardDevice(device)
	  DevicePtr device;

RegisterKeyboardDevice makes the given device the main key-
board.	This routine is called once upon initialization and
cannot be called again.

The following DIX procedures return the specified DevicePtr.
They may or may not be useful to DDX implementors.

     DevicePtr LookupKeyboardDevice()

LookupKeyboardDevice returns pointer for current main key-
board device.

     DevicePtr LookupPointerDevice()

LookupPointerDevice returns pointer for current main point-
ing device.



Porting Layer Definition   - 25 -	       April 8, 1994








A DeviceProc (the kind passed to AddInputDevice()) in the
following form:

     Bool pInternalDevice->DeviceProc(device, action);
	  DeviceIntPtr device;
	  int action;

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):

o    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 point-
     ing device cannot be initialized, the DIX code will
     refuse to continue booting up.

o    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.

o    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.

o    DEVICE_CLOSE - At DEVICE_CLOSE (terminate or reset)
     time, the device should be totally closed down.

	  void InitPointerDeviceStruct(device, map, mapLength,
		    GetMotionEvents, ControlProc, numMotionEvents)
	       DevicePtr device;
	       CARD8 *map;
	       int mapLength;
	       ValuatorMotionProcPtr ControlProc;
	       PtrCtrlProcPtr GetMotionEvents;
	       int numMotionEvents;
     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 pro-
     tocol specification.  ControlProc and GetMotionEvents
     are DDX routines, see above.

     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



Porting Layer Definition   - 26 -	       April 8, 1994








     zero.

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

     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.


5.1.6.	Keyboard Mapping and Keycodes

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 inclu-
sive.

The keycode mapping information that you set up consists of
the following:

o    A minimum and maximum keycode number

o    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.

Once the map is set up, DIX keeps and maintains the client's
changes to it.

The X protocol defines standard names to indicate the sym-
bol(s) printed on each keycap. (See X11/keysym.h)

Legal modifier keys must generate both up and down transi-
tions.	When a client tries to change a modifier key (for
instance, to make "A" the "Control" key), DIX calls the fol-
lowing routine, which should retuurn TRUE if the key can be
used as a modifier on the given device:




Porting Layer Definition   - 27 -	       April 8, 1994








     Bool LegalModifier(key, pDev)
	 unsigned int key;
	 DevicePtr pDev;

5.2.  Screens

Different computer graphics displays have different capabil-
ities.	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 them-
selves, including maintaining their own high-level, graphic
data structures.


5.2.1.	Screen Hardware Requirements

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 exten-
sion 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.

X uses the pixel as its fundamental unit of distance on the
screen.  Therefore, most programs will measure everything in
pixels.

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 reso-
lutions 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 applica-
tions).


5.2.2.	Data Structures

X supports multiple screens that are connected to the same
server.  Therefore, all the per-screen information is bun-
dled 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.

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.



Porting Layer Definition   - 28 -	       April 8, 1994








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.

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.

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.).

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.

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 addi-
tion, 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.


5.2.3.	Output Initialization


     InitOutput(pScreenInfo, argc, argv)
	  ScreenInfo *pScreenInfo;
	  int argc;
	  char **argv;
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.

Your InitOutput() routine should initialize each screen you
wish to use by calling AddScreen(), and then it should



Porting Layer Definition   - 29 -	       April 8, 1994








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.

     int AddScreen(scrInitProc, argc, argv)
	  Bool (*scrInitProc)();
	  int argc;
	  char **argv;
You should call AddScreen(), a DIX procedure, in InitOut-
put() 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 insuffi-
cient memory to add the screen, or (*scrInitProc) returned
FALSE.

The scrInitProc should be of the following form:

     Bool scrInitProc(iScreen, pScreen, argc, argv)
	  int iScreen;
	  ScreenPtr pScreen;
	  int argc;
	  char **argv;
iScreen is the index for this screen; 0 for the first one
initialized, 1 for the second, etc.  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).

This procedure must determine what actual device it is sup-
posed 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.

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.


5.2.4.	Region Routines in the ScreenRec

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



Porting Layer Definition   - 30 -	       April 8, 1994








and OR.

A region is frequently implemented as a list of rectangles
or bitmaps that enclose the selected pixels.  Region opera-
tors control the "clipping policy," or the operations that
work on regions.  (The sample server uses YX-banded rectan-
gles.  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 imple-
mentations using other region formats will need to supply
different versions of the region operators.

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 calcu-
late 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
Xrealloc(), whereupon the new pointer is remembered in the
RegionRec.

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.

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

     macro: RegionPtr REGION_CREATE(pScreen, rect, size)

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 rou-
tines, 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 imple-
mentation might ignore size, if applicable.

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



Porting Layer Definition   - 31 -	       April 8, 1994








	  BoxPtr	 rect;
	  int	    size;

     macro: REGION_INIT(pScreen, pRegion, rect, size)

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 RegionCre-
ate.  This avoids the additional dynamic memory allocation
overhead for the region structure itself.

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

     macro: Bool REGION_COPY(pScreen, dstrgn, srcrgn)

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

     void pScreen->RegionDestroy( pRegion)
	  RegionPtr pRegion;

     macro: REGION_DESTROY(pScreen, pRegion)

RegionDestroy destroys a region and frees all allocated mem-
ory.

     void pScreen->RegionUninit (pRegion)
	  RegionPtr pRegion;

     macro: REGION_UNINIT(pScreen, pRegion)

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, pRe-
gion must not be reused until it has been RegionInit'ed
again.

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

     macro: Bool REGION_INTERSECT(pScreen, newReg, reg1, reg2)

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

     macro: Bool REGION_UNION(pScreen, newReg, reg1, reg2)

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

     macro: Bool REGION_UNION(pScreen, newReg, regMinuend, regSubtrahend)




Porting Layer Definition   - 32 -	       April 8, 1994








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

     macro: Bool REGION_INVERSE(pScreen, newReg, pReg,	pBox)

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.

The subtract function removes the Subtrahend from the Minu-
end and puts the result in newReg.

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.

Each routine must return the value TRUE for success.

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

     macro: REGION_RESET(pScreen, pRegion, pBox)

RegionReset sets the region to describe one rectangle and
reallocates it to a size of one rectangle, if applicable.

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

     macro: REGION_TRANSLATE(pScreen, pRegion, x, y)

TranslateRegion simply moves a region +x in the x direction
and +y in the y direction.

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

     macro: int RECT_IN_REGION(pScreen, pRegion, pBox)

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.




Porting Layer Definition   - 33 -	       April 8, 1994








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

     macro: Bool POINT_IN_REGION(pScreen, pRegion, x, y, pBox)

PointInRegion returns true if the point x, y is in the
region.  In addition, it fills the rectangle pBox with coor-
dinates 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 implemen-
tation, this comes cheaply.)

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 PointInRe-
gion() 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.

     Bool pScreen->RegionNotEmpty(pRegion)
	  RegionPtr pRegion;

     macro: Bool REGION_NOTEMPTY(pScreen, pRegion)

RegionNotEmpty is a boolean function that returns true or
false depending upon whether the region encloses any pixels.

     void pScreen->RegionEmpty(pRegion)
	  RegionPtr pRegion;

     macro: REGION_EMPTY(pScreen, pRegion)

RegionEmpty sets the region to be empty.

     BoxPtr pScreen->RegionExtents(pRegion)
	  RegionPtr pRegion;

     macro: REGION_EXTENTS(pScreen, pRegion)

RegionExtents returns a rectangle that is the smallest pos-
sible superset of the entire region.  The caller will not
modify this rectangle, so it can be the one in your region
struct.

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

     macro: Bool REGION_APPEND(pScreen, pDstRgn, pRegion)



Porting Layer Definition   - 34 -	       April 8, 1994








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

     macro: Bool REGION_VALIDATE(pScreen, pRegion, pOverlap)

These functions provide an optimization for clip list gener-
ation 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 RegionVali-
date 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 over-
lap; FALSE otherwise.

     RegionPtr pScreen->BitmapToRegion (pPixmap)
	  PixmapPtr pPixmap;

     macro: RegionPtr BITMAP_TO_REGION(pScreen, pPixmap)

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 rou-
tine returns NULL if out of memory.

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

     macro: RegionPtr RECTS_TO_REGION(pScreen, nrects, pRects, ordering)

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 rect-
angles 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).

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

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.






Porting Layer Definition   - 35 -	       April 8, 1994








5.2.5.	Cursor Routines for a Screen

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.

For each possible shape the cursor assumes, there is a Cur-
sorRec 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 Cur-
sorRec 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.

CursorBits in Xserver/include/cursorstr.h is a device-inde-
pendent structure containing a device-independent represen-
tation of the bits for the source and mask.  (This is possi-
ble because the bitmap representation is the same for all
screens.)

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 real-
izations 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 particu-
lar screen and cursor:

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

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



Porting Layer Definition   - 36 -	       April 8, 1994








screen.

You must implement the following routines for your hardware:

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

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

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 cur-
rently in use; this happens when the system is reset.

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

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.

     void pScreen->RecolorCursor( pScr, pCurs, displayed)
	  ScreenPtr pScr;
	  CursorPtr pCurs;
	  Bool displayed;

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 unreal-
ize, a realize, and possibly a display (in micursor.c); how-
ever this constrains UnrealizeCursor and RealizeCursor to
always return TRUE as no error indication is returned here.

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

ConstrainCursor should cause the cursor to restrict its
motion to the rectangle pBox.  DIX code is capable of



Porting Layer Definition   - 37 -	       April 8, 1994








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.

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

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 win-
dow.  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 PointerNonIn-
terestBox.

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


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 con-
fined within, pHotBox.	This routine is for informing DIX
only; it alters no state within DDX.

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


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 ProcessIn-
putEvents() and then it must call DevicePtr->processInput-
Proc.  Its effects are, of course, limited in value for
absolute pointing devices such as a tablet.

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



Porting Layer Definition   - 38 -	       April 8, 1994








	 int x,y;


If your ddx provides some mechanism for the user to magi-
cally move the pointer between multiple screens, you need to
inform DIX when this occurs.  You should call NewCur-
rentScreen to accomplish this, specifying the new screen and
the new x and y coordinates of the pointer on that screen.


5.2.6.	Visuals, Depths and Pixmap Formats for Screens

The "depth" of a image is the number of bits that are used
per pixel to display it.

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.

A "pixmap image format" or a "pixmap format" is a descrip-
tion 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?)

For each screen, you must decide upon what depth(s) it sup-
ports.	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.

A "display class" tells whether the display is monochrome or
color, whether there is a lookup table, and how the lookup
table works.

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 dic-
tates 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).

Each visual is identified by a 32-bit visual ID which the



Porting Layer Definition   - 39 -	       April 8, 1994








client uses to choose what visual is desired on a given win-
dow.  Clients can be using more than one visual on the same
screen at the same time.

The class of a display describes how this translation takes
place.	There are three ways to do the translation.

o    Pseudo - The pixel value, as a whole, is looked up in a
     table of length map_entries to determine the color to
     display.

o    True - The pixel value is broken up into red, green,
     and blue fields, each of which are looked up in sepa-
     rate red, green, and blue lookup tables, each of length
     map_entries.

o    Gray - The pixel value is looked up in a table of
     length map_entries to determine a gray level to dis-
     play.

In addition, the lookup table can be static (resulting col-
ors 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:


o    Static Gray - The pixel value (of however many bits)
     determines directly the level of gray that the pixel
     assumes.

o    Gray Scale - The pixel value is fed through a lookup
     table to arrive at the level of gray to display for the
     given pixel.

o    Static Color - The pixel value is fed through a fixed
     lookup table that yields the color to display for that
     pixel.

o    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.

o    True Color - Each pixel value consists of one or more
     bits that directly determine each primary color inten-
     sity after being fed through a fixed table.

o    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.



Porting Layer Definition   - 40 -	       April 8, 1994








Here are some examples:

     A simple monochrome 1 bit per pixel display is Static
     Gray.

     A display that has 2 bits per pixel for a choice
     between the colors of black, white, green and violet is
     Static Color.

     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.

     If you take the last example and scramble the corre-
     spondence between pixel values and colors it becomes a
     Static Color display.

     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 col-
     ors on the screen at once.  This is a pseudocolor dis-
     play.  The client application gets to fill the lookup
     table in this class of display.

     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.

     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 Visual-
     Rec data structure along with other display character-
     istics like the depth and other numbers.

     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 Xalloc() appropriate
     blocks or use static variables initialized to the cor-
     rect values.




Porting Layer Definition   - 41 -	       April 8, 1994








5.2.7.	Colormaps for Screens

A colormap is a device-independent mapping between pixel
values and colors displayed on the screen.

Different windows on the same screen can have different col-
ormaps 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.

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.


5.2.7.1.  Colormap Routines

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 both cfbcmap.c and mfbcmap.c; since
mfb does not do very much with color, the cfb versions are
typically more useful prototypes.

     Bool pScreen->CreateColormap(pColormap)
	  ColormapPtr pColormap;


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.

     void pScreen->DestroyColormap(pColormap)
	  ColormapPtr pColormap;


This routine is called by the DIX FreeColormap routine after
it has uninstalled the colormap and notified all interested



Porting Layer Definition   - 42 -	       April 8, 1994








parties, and before it has freed any of the colormap stor-
age.  It is the DDX layer's chance to free any data it added
to the colormap.

     void pScreen->InstallColormap(pColormap)
	  ColormapPtr pColormap;


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 simul-
taneously.

In the more general case of multiple hardware lookup tables,
this may cause some other colormap to be uninstalled, mean-
ing that windows that subscribed to the colormap that was
uninstalled may end up being off-color.  See the note,
below, about uninstalling maps.

     void pScreen->UninstallColormap(pColormap)
	  ColormapPtr pColormap;


UninstallColormap should remove pColormap from screen pCol-
ormap->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.)

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



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.

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


StoreColors changes some of the entries in the colormap
pmap.  The number of entries to change are ndef, and pdefs



Porting Layer Definition   - 43 -	       April 8, 1994








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 xCol-
orItem 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.

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



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.

Each value is an unsigned value ranging from 0 to 65535.
The bits least likely to be used are the lowest ones.

For example, if you had a pseudocolor display with any num-
ber 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 sim-
ply 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).

5.2.7.2.  Initializing a Colormap

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 mem-
ory 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 hard-
wares notion of the color for that pixel.  If the colormap
is the default for the screen, which you can tell by looking
at the IsDefault bit in the flags field, you should allocate



Porting Layer Definition   - 44 -	       April 8, 1994








BlackPixel and WhitePixel to match the values you set in the
pScreen structure.  (Of course, you picked those values to
begin with.)

You can also wait and use AllocColor() to allocate black-
Pixel and whitePixel after the default colormap has been
created.  If the default colormap is static and you initial-
ized it in pScreen->CreateColormap, then use can use Alloc-
Color 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.

There are two ways to fill in the colormap.  The simplest
way is to use the DIX function AllocColor.

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

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 allo-
cated from entry 0, set *pPix to 0.  This will find the
first free pixel and use that.

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.

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.


5.2.8.	Fonts for Screens

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 charac-
ter.  Parallel fonts may be available in a variety of sizes
and variations, including "bold" and "italic."	X supports



Porting Layer Definition   - 45 -	       April 8, 1994








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.

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.).

A short paper in PostScript format is included with the sam-
ple server that defines BDF.  It includes helpful pictures,
which is why it is done in PostScript and is not included in
this document.

Your implementation should include some sort of font com-
piler 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.

It is important the font properties contained in the BDF
files are preserved across any font compilation. In particu-
lar, copyright information cannot be casually tossed aside
without legal ramifications. Other properties will be impor-
tant to some sophisticated applications.

All clients get font information from the server.  There-
fore, 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.


5.2.8.1.  Portable Compiled Format

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.

The fonts included with the tape are stored in fonts/bdf.
The font compiler is found in fonts/tools/bdftopcf.

5.2.8.2.  Font Realization

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



Porting Layer Definition   - 46 -	       April 8, 1994








into memory.

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 stor-
age.

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

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

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 mfbfont.c (which does very little).


5.2.9.	Other Screen Routines

You must supply several other screen-specific routines for
your X server implementation.  Some of these are described
in other sections:

o    GetImage() is described in the Drawing Primitives sec-
     tion.

o    GetSpans() is described in the Pixblit routine section.

o    Several window and pixmap manipulation procedures are
     described in the Window section under Drawables.

o    The CreateGC() routine is described under Graphics Con-
     texts.


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

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 Stipple-
Shape (defined in X.h).  For CursorShape, return the maximum
width and height for cursors that you can handle.  For Tile-
Shape and StippleShape, start with the suggested values in



Porting Layer Definition   - 47 -	       April 8, 1994








pWidth and pHeight and modify them in place to be optimal
values that are greater than or equal to the suggested val-
ues.  The sample server implementation is in
Xserver/mfb/mfbmisc.c.

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

SourceValidate should be called by CopyArea/CopyPlane primi-
tives when the source drawable is not the same as the desti-
nation, and the SourceValidate function pointer in the
screen is non-null.  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.

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

SaveScreen() is used for Screen Saver support (see WaitFor-
Something()).  pScreen is the screen to save.

     Bool pScreen->CloseScreen(pScreen)
	 ScreenPtr pScreen;

When the server is reset, it calls this routine for each
screen.

     Bool pScreen->CreateScreenResources(pScreen)
	 ScreenPtr pScreen;

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 register their devPrivates on all
structures that support them (see the section on devPrivates
below).  If you need to create any resources that have
dynamic devPrivates as part of your screen initialization,
you should do so in this function instead of in the screen
init function passed to AddScreen to guarantee that the
resources have a complete set of devPrivates.  This routine
returns TRUE if successful.

5.3.  Drawables

A drawable is a descriptor of a surface that graphics are
drawn into, either a window on the screen or a pixmap in
memory.

Each drawable has a type, class, ScreenPtr for the screen it



Porting Layer Definition   - 48 -	       April 8, 1994








is associated with, depth, position, size, and serial num-
ber.  The type is one of the defined constants DRAW-
ABLE_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 proce-
dures used to manipulate and draw into the drawable.  Posi-
tion is used (currently) only by windows; pixmaps must set
these fields to 0,0 as this reduces the amount of condi-
tional 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.

Both PixmapRecs and WindowRecs are structs that start with a
drawable and continue on with more fields.  Pixmaps have
devPrivate pointers which usually point to the pixmap data
but could conceivably be used for anything that DDX wants.
Both windows and pixmaps have an array of devPrivates
unions, one entry of which will probably be used for DDX
specific data.	Entries in this array are allocated using
Allocate{Window|Pixmap}PrivateIndex() (see Wrappers and
devPrivates below).  This is done because different graphics
hardware has different requirements for management; if the
graphics is always handled by a processor with an indepen-
dent address space, there is no point having a pointer to
the bit image itself.

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.


5.3.1.	Pixmaps

A pixmap is a three-dimensional array of bits stored some-
where 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.

Note that the pixmap data structure is reference-counted.

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



Porting Layer Definition   - 49 -	       April 8, 1994








sophisticated implementations may put the data in undis-
played framebuffer storage.

In addition to dynamic devPrivates (see the section on
devPrivates 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
probably a pointer to the bits in the pixmap.

A bitmap is a pixmap that is one bit deep.

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

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/mfb/mfbpixmap.c for the sample server implementa-
tion.

     Bool pScreen->DestroyPixmap(pPixmap)
	  PixmapPtr pPixmap;

This ScreenRec procedure must "destroy" a pixmap.  It should
decrement the reference count and, if zero, it must deallo-
cate the PixmapRec and all attached devPrivate blocks.	If
successful, it returns TRUE.  See Xserver/mfb/mfbpixmap.c
for the sample server implementation.

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

This routine takes a pixmap header (the PixmapRec plus all
the dynamic devPrivates) 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



Porting Layer Definition   - 50 -	       April 8, 1994








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 suc-
cessful.  See Xserver/mi/miscrinit.c for the sample server
implementation.

     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;

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->devPri-
vate.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 FreeScratchPixmap-
Header.

5.3.2.	Windows

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 rela-
tionships 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 win-
dows 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.

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.

Window painting is orchestrated through an expose event sys-
tem.  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.



Porting Layer Definition   - 51 -	       April 8, 1994








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 intelli-
gently 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 per-
formance of window moves.

On the other hand, backing store can be quite complex,
because all graphics drawn to hidden areas must be inter-
cepted 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.

In X, the decision to use the backing-store scheme is made
by you, the server implementor.  X provides hooks for imple-
menting backing store, therefore the decision to use this
strategy can be made on the fly.  For example, you may use
backing store only for certain windows that the user
requests or you may use backing store until memory runs out,
at which time you start dropping pieces of backing as needed
to make more room.

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 rectan-
gles 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 responsibil-
ity to paint the window parts that were exposed but not
restored.

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 con-
trols 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



Porting Layer Definition   - 52 -	       April 8, 1994








used for that, also.

The window has several fields used in drawing operations:

o    clipList - This region, in conjunction with the client
     clip region in the gc, is used to clip output.  cli-
     pList 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 Not-
     ClippedByChildren(pWin) returns the unclipped region.

o    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 chil-
     dren removed.

Most of the other fields are for DIX use only.


5.3.2.1.  Window Procedures in the ScreenRec

You should implement all of the following procedures and
store pointers to them in the screen record.

The device-independent portion of the server "owns" the win-
dow tree.  However, clever hardware might want to know the
relationship of mapped windows.  There are pointers to pro-
cedures 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.

     Bool pScreen->CreateWindow(pWin)
	  WindowPtr pWin;

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.

See Xserver/mfb/mfbwindow.c for the sample server implemen-
tation.

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

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.

See Xserver/mfb/mfbwindow.c for the sample server implemen-
tation.




Porting Layer Definition   - 53 -	       April 8, 1994








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

This routine is a hook for when DIX moves or resizes a win-
dow.  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 long-
word 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.

See Xserver/mfb/mfbwindow.c for the sample server implemen-
tation.

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

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

These routines are hooks for when DIX maps (makes visible)
and unmaps (makes invisible) a window.	It should do what-
ever 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 under-
lying windows; this is taken care of by DIX.  DIX does, how-
ever, call PaintWindowBackground() and PaintWindowBorder()
to perform some of these.

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

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 com-
putes accelerators for quickly putting up background and
border tiles.  (See description of the set of routines
stored in the WindowRec.)

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

ValidateTree calculates the clipping region for the parent
window and all of its children.  This routine must be pro-
vided. The sample server has a machine-independent version
in Xserver/mi/mivaltree.c.  This is a very difficult routine



Porting Layer Definition   - 54 -	       April 8, 1994








to replace.

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

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 sam-
ple server sets this to NULL.

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

The WindowExposures() routine paints the border and gener-
ates exposure events for the window.  pRegion is an unoc-
cluded 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 rou-
tine 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 pro-
vided. The sample server has a machine-independent version
in Xserver/mi/miexpose.c.

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

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).

5.3.2.2.  Window Painting Procedures

In addition to the procedures listed above, there are four
routines which manipulate the actual window image directly.
In the sample server, mi implementations will work for most
purposes and mfb/cfb routines speed up situations, such as
solid backgrounds/borders or tiles that are 8, 16 or 32 pix-
els square.

These three routines are used for systems that implement a
backing-store scheme for it to know when to stash away areas
of pixels and to restore or reposition them.

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



Porting Layer Definition   - 55 -	       April 8, 1994








	  Bool generateExposures;

This routine is called on a window in response to a ClearTo-
Background request from the client.  This request has two
different but related functions, depending upon generateEx-
posures.

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 pro-
cedure HandleExposures(), which calls the WindowExposures()
routine, below, for the window and all of its child windows.

If generateExposures is false, the client is trying to sim-
ply 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 PaintWindow-
Background).  If w or h is zero, it clears all the way to
the right or lower edge of the window.

The sample server implementation is in Xserver/mi/miwin-
dow.c.

     void pScreen->PaintWindowBackground(pWin, region, kind)
	  WindowPtr pWin;
	  RegionPtr region;
	  int kind; /* must be PW_BACKGROUND */

     void pScreen->PaintWindowBorder(pWin, region, kind)
	  WindowPtr pWin;
	  RegionPtr region;
	  int kind; /* must be PW_BORDER */

These two routines are for painting pieces of the window
background or border.  They both actually paint the area
designated by region.  The kind parameter is a defined con-
stant that is always PW_BACKGROUND or PW_BORDER, as shown.
Therefore, you can use the same routine for both.  The
defined constant tells the routine whether to use the win-
dow's border fill style or its background fill style to
paint the given region.  Both fill styles consist of a union
which holds a tile pointer and a pixel value, along with a
separate variable which indicates which entry is valid.  For
PW_BORDER, borderIsPixel != 0 indicates that the border Pix-
Union contains a pixel value, else a tile.  For PW_BACK-
GROUND there are four values, contained in backgroundState;
None, ParentRelative, BackgroundPixmap and BackgroundPixel.
None indicates that the region should be left unfilled,
while ParentRelative indicates that the background of the
parent is inherited (see the Protocol document for the exact
semantics).

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



Porting Layer Definition   - 56 -	       April 8, 1994








	  WindowPtr pWin;
	  DDXPointRec oldpt;
	  RegionPtr oldRegion;

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).

oldpt is the old location of the upper-left corner.  oldRe-
gion 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).

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 mul-
tiple windows with multiple depths, and your routine needs
to take care of that.

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.

The sample server implementation is in Xserver/mfb/mfbwin-
dow.c.


5.3.2.3.  Screen Operations for Backing Store

Each ScreenRec has six functions which provide the backing
store interface.  For screens not supporting backing store,
these pointers may be nul.  Servers that implement some
backing store scheme must fill in the procedure pointers for
the procedures below, and must maintain the backStorage
field in each window struct.  The sample implementation is
in mi/mibstore.c.

     void pScreen->SaveDoomedAreas(pWin, pRegion, dx, dy)
	  WindowPtr pWin;
	  RegionPtr pRegion;
	  int dx, dy;

This routine saves the newly obscured region, pRegion, in
backing store.	dx, dy indicate how far the window is being
moved, useful as the obscured region is relative to the win-
dow as it will appear in the new location, rather then rela-
tive to the bits as the are on the screen when the function
is invoked.




Porting Layer Definition   - 57 -	       April 8, 1994








     RegionPtr pScreen->RestoreAreas(pWin, pRegion)
	  WindowPtr pWin;
	  RegionPtr pRegion;

This looks at the exposed region of the window, pRegion, and
tries to restore to the screen the parts that have been
saved.	It removes the restored parts from the backing stor-
age (because they are now on the screen) and subtracts the
areas from the exposed region.	The returned region is the
area of the window which should have expose events generated
for and can be either a new region, pWin->exposed, or NULL.
The region left in pRegion is set to the area of the window
which should be painted with the window background.

     RegionPtr pScreen->TranslateBackingStore(pWin, dx, dy, oldClip, oldx, oldy)
	  WindowPtr pWin;
	  int dx, dy;
	  RegionPtr oldClip;
	  int oldx, oldy;

This is called when the window is moved or resized so that
the backing store can be translated if necessary.  oldClip
is the old cliplist for the window, which is used to save
doomed areas if the window is moved underneath its parent as
a result of bitgravity.  The returned region represents
occluded areas of the window for which the backing store
contents are invalid.

     void pScreen->ExposeCopy(pSrc, pDst, pGC, prgnExposed, srcx, srcy, dstx, dsty, plane)
	  WindowPtr pSrc;
	  DrawablePtr pDst;
	  GCPtr pGC;
	  RegionPtr prgnExposed;
	  int srcx;
	  int srcy;
	  int dstx;
	  int dsty;
	  unsigned long plane;

Copies a region from the backing store of pSrc to pDst.

     RegionPtr pScreen->ClearBackingStore(pWindow, x, y, w, h, generateExposures)
	  WindowPtr pWindow;
	  int x;
	  int y;
	  int w;
	  int h;
	  Bool generateExposures;

Clear the given area of the backing pixmap with the back-
ground of the window.  If generateExposures is TRUE, gener-
ate exposure events for the area. Note that if the area has
any part outside the saved portions of the window, we do not
allow the count in the expose events to be 0, since there



Porting Layer Definition   - 58 -	       April 8, 1994








will be more expose events to come.

     void pScreen->DrawGuarantee(pWindow, pGC, guarantee)
	  WindowPtr pWindow;
	  GCPtr pGC;
	  int guarantee;

This informs the backing store layer that you are about to
validate a gc with a window, and that subsequent output to
the window is (or is not) guaranteed to be already clipped
to the visible regions of the window.


5.3.2.4.  Screen Operations for Multi-Layered Framebuffers

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/miwin-
dow.c.	This can be easily accomplished by calling miS-
creenInit.

    void pScreen->MarkWindow(pWin)
     WindowPtr pWin;

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.

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

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 frame-
buffer case, pLayerWin == pWin.

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

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



Porting Layer Definition   - 59 -	       April 8, 1994








TRUE if PostChangeSaveUnder needs to be called to finish
turning off backing store.

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

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.

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

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.

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

The formerly dix function SlideAndSizeWindow has moved to
ddx and is accessed via this screen function.  The new posi-
tion is given by x,y.  The new size is given by w,h.

    WindowPtr pScreen->GetLayerWindow(pWin)
     WindowPtr pWin

This is a new function which returns a child of the layer
parent of pWin.

    void pScreen->HandleExposures(pWin)
     WindowPtr pWin;

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.

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




Porting Layer Definition   - 60 -	       April 8, 1994








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.

    void pScreen->SetShape(pWin)
     WindowPtr pWin;

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 func-
tion is called.

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

The formerly dix function ChangeBorderWidth has moved to ddx
and is accessed via this screen function.  The new border
width is given by width.

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

This function is called for windows that are being unreal-
ized as part of an UnrealizeTree.  pChild is the window
being unrealized, pWin is an ancestor, and the fromConfigure
value is simply propogated from UnrealizeTree.

5.4.  Graphics Contexts and Validation

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 pat-
tern generation, the current font for text generation, and
other similar attributes.

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 chalk-
board.	A GC is like a piece of chalk.

Unlike many similar systems, there is no "current pen loca-
tion."	Every graphic operation is accompanied by the coor-
dinates where it is to happen.

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



Porting Layer Definition   - 61 -	       April 8, 1994








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 vec-
tors of procedure pointers to take advantage of hard-
ware/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