RPG IV Socket Tutorial

RPG IV Socket Tutorial
RPG IV Socket Tutorial
Scott Klement
RPG IV Socket Tutorial
by Scott Klement
Copyright © 2001, 2002 by Scott Klement
This tutorial strives to teach network programming using sockets to AS/400 or iSeries programmers who use the
RPG IV programming language.
It is assumed that the reader of this tutorial is already familiar with the RPG IV language, including the use of
prototypes, subprocedures, service programs and pointers in an RPG IV environment.
Table of Contents
1. Introduction to TCP and Sockets ............................................................................................................................1
1.1. TCP/IP Concepts and Terminology................................................................................................................1
1.2. Overview of a TCP communications session.................................................................................................3
2. Looking up host names & services ..........................................................................................................................6
2.1. Services and Ports ..........................................................................................................................................6
2.2. Host names and addresses ..............................................................................................................................8
3. Socket API calls to create our first client program..............................................................................................14
3.1. The socket() API call....................................................................................................................................14
3.2. The connect() API call .................................................................................................................................15
3.3. The send() and recv() API calls....................................................................................................................17
3.4. Translating from ASCII to EBCDIC............................................................................................................19
3.5. The close() API call......................................................................................................................................20
3.6. Our first client program ................................................................................................................................21
4. Improving our client program ...............................................................................................................................27
4.1. improvement #1: header files .......................................................................................................................27
4.2. improvement #2: error handling...................................................................................................................30
4.3. improvement #3: Creating a Read Line utility .............................................................................................33
4.4. improvement #4: Creating a Write Line utility ............................................................................................37
4.5. Packaging our utilities into a service program .............................................................................................39
4.6. improvement #5: Stripping out the HTTP response.....................................................................................41
4.7. improvement #6: Sending back escape messages ........................................................................................43
4.8. improvement #7: Displaying the data using DSPF ......................................................................................45
4.9. Our updated client program..........................................................................................................................46
5. Creating server programs ......................................................................................................................................51
5.1. Introduction to server programs ...................................................................................................................51
5.2. The bind() API call.......................................................................................................................................52
5.3. The listen() API call .....................................................................................................................................54
5.4. The accept() API call....................................................................................................................................54
5.5. Our first server program ...............................................................................................................................56
5.6. Testing server programs ...............................................................................................................................60
5.7. Making our program end..............................................................................................................................62
5.8. The setsockopt() API call .............................................................................................................................63
5.9. The revised server program ..........................................................................................................................67
6. Handling many sockets at once using select().......................................................................................................73
6.1. Overview of handling many clients..............................................................................................................73
6.1.1. The "single program" approach: ......................................................................................................73
6.2. The select() API call.....................................................................................................................................74
6.3. Utility routines for working with select().....................................................................................................76
6.4. A combination client/server example...........................................................................................................80
6.5. Blocking vs. non-blocking sockets...............................................................................................................87
6.6. The fcntl() API call ......................................................................................................................................88
iii
6.7. A multi-client example of our server ...........................................................................................................90
6.8. Adding some refinements to our approach.................................................................................................102
6.9. A Chat Application.....................................................................................................................................103
7. Handling many sockets by spawning jobs ..........................................................................................................119
7.1. Overview of job spawning approach ..........................................................................................................119
7.2. The takedescriptor() and givedescriptor() API calls ..................................................................................120
7.3. Retrieving the internal job id......................................................................................................................121
7.4. Communicating the job information ..........................................................................................................124
7.5. Our server as a multi-job server .................................................................................................................126
7.6. Working towards our next example............................................................................................................134
7.7. Validating a user-id and password..............................................................................................................137
7.8. Running with the user’s authority ..............................................................................................................137
7.9. A "generic server" example........................................................................................................................138
7.10. Trying the "generic server" out ................................................................................................................150
8. The User Datagram Protocol ...............................................................................................................................153
8.1. Overview of using UDP .............................................................................................................................153
8.2. Creating a socket for use with UDP ...........................................................................................................154
8.3. The sendto() API call .................................................................................................................................154
8.4. The recvfrom() API call .............................................................................................................................156
8.5. Sample UDP server ....................................................................................................................................157
8.6. Sample UDP client .....................................................................................................................................161
8.7. Trying it out................................................................................................................................................165
Colophon....................................................................................................................................................................167
iv
Chapter 1. Introduction to TCP and Sockets
Written by Scott Klement.
1.1. TCP/IP Concepts and Terminology
This section is an introduction to TCP/IP programming using a sockets API. (Sockets can also be used to work with
other network protocols, such as IPX/SPX and Appletalk, but that is beyond the scope of this document.) The
standard socket API was originally developed in the Unix world, but has been ported to OS/400 as part of the
"Unix-type" APIs, and a modified version was also ported to the Windows platform under the name "Windows
Sockets" (or "Winsock" for short)
Usually when someone refers to "TCP/IP" they are referring to the entire suite of protocols, all based on the Internet
Protocol ("IP"). Unlike a single network, where every computer is directly connected to every other computer, an
"inter-network" (or "internet") is a collection of one or more networks. These networks are all connected together to
form a larger "virtual network". Any host on this virtual network can exchange data with any other host, by referring
to the hosts "address".
The address is a 32-bit number which is unique across the entire internet. Typically, this number is broken into 4
8-bit pieces, separated by periods to make it easier for humans to read. This human readable format is called
"dotted-decimal" format, or just "dot-notation". An address displayed in this fashion looks something like
"192.168.66.21".
Different parts of this "IP Address" are used to identify which network a host is located on, and the rest of the address
is used to identify the host itself. Which part of the address and which part is the host is determined by a "network
mask" (or "netmask" for short.) The netmask is another 32-bit number which acts like like a "guide" to the IP address.
Each bit that is turned on in the netmask means that the corresponding bit in the IP address is part of the network’s
address. Each bit that is turned off means that the corresponding bit in the IP address is part of the host’s address.
Here’s an example of an IP address and netmask:
IP Address:
Netmask:
dotted-decimal
-------------192.168.66.21
255.255.255.0
Network Address is: 192.168.66
Host Address is:
.21
same number in binary format:
----------------------------------11000000 10101000 01000010 00010101
11111111 11111111 11111111 00000000
11000000 10101000 01000010
00010101
A slightly more complicated example:
IP Address:
Netmask:
dotted-decimal
-------------192.168.41.175
255.255.255.248
Network Address is: 192.168.41.168
Host Address is:
7
same number in binary format:
----------------------------------11000000 10101000 00101001 10101111
11111111 11111111 11111111 11111000
11000000 10101000 00101001 10101
111
1
Chapter 1. Introduction to TCP and Sockets
When a system sends data over the network using internet protocol, the data is sent fixed-length data records called
datagrams. (These are sometimes referred to as "packets") The datagram consists of a "header" followed by a "data
section". The header contains addressing information, much like an envelope that you send through your local postal
service. The header contains a "destination" and a "return to" address, as well as other information used by the
internet protocol. Another similarity between IP and your postal service is that each packet that gets sent isn’t
guarenteed to arrive at the destination. Although every effort is made to get it there, sometimes datagrams get lost or
duplicated in transit. Furthermore, if you send 5 datagrams at once, there’s no guarantee that they’ll arrive at their
destination at the same time or in the same order.
What’s really needed is a straight-forward way to ensure that all the packets that get sent arrive at their destination.
When they arrive, make sure they’re in the same sequence, and that all duplicated datagrams get discarded. To solve
this problem, the Transmission Control Protocol (TCP) was created. It runs on top of IP, and takes care of the chore
of making certain that every packet that is sent will arrive at its destination. It also allows many packets to be joined
together into a "continuous stream" of bytes, eliminating the need for you to split your data into packets and re-join
them at the other end.
It’s useful to remember that TCP runs "on top of" IP. That means that any data you send via TCP gets converted into
one or more datagrams, then sent over the networking using IP, then is reassembled into a stream of data on the other
end.
TCP is a "connection oriented protocol", which means that when you want to use it, you must first "establish a
connection." To do this, one program must take the role of a "server", and another program must take the role of a
"client." The server will wait for connections, and the client will make a connection. Once this connection has been
established, data may be sent in both directions reliably until the connection is closed. In order to allow multiple TCP
connections to and from a given host, "port" numbers are established. Each TCP packet contains an "origin port" and
a "destination port", which is used to determine which program running under which of the system’s tasks is to
receive the data.
There are two other protocols that are used over IP. They are the "User Datagram Protocol (UDP)", and the "Internet
Control Message Protocol" (ICMP).
UDP is similar to TCP, except that data is sent one datagram at a time. The major difference between the UDP
datagrams and the "raw" IP datagrams is that UDP adds port numbers to the packets. This way, like TCP, many tasks
on the system can use UDP at the same time. UDP is usually used when you know that you only want to send a tiny
amount of data (one packets worth) at a time, and therefore you don’t need all the extra overhead of TCP.
ICMP is used internally by the internet protocol to exchange diagnostic and error messages. For example, when you
attempt to connect to a port on a remote machine, and that machine chooses to refuse your connection, it needs some
way of telling you that the connection has been refused. This is done by sending ICMP messages. You never need to
write or receive ICMP messages directly, they are always handled by the TCP/IP stack. They are strictly a "control
message protocol."
Another important concept is that of a socket. A socket is an "endpoint for network communications." In other
words, it’s the virtual device that your program uses to communicate with the network. When you want to write bytes
out over the network, you write them to a socket. When you read from the network, you read from that socket. In this
way, it is similar to the way a "file" is used to interact with hard drive storage.
The last thing that I’d like to cover here is "domain names." As you read above, all TCP/IP communications are done
using an "address." Without this address, no data can be sent or received. However, while addresses work very well
for the computer, they’re a little hard for people to remember. Perhaps you wanted to connect to a computer the
computer that keeps track of inventory at Acme, Inc. How do you know it’s address? If you knew it’s IP address
2
Chapter 1. Introduction to TCP and Sockets
already, how would you remember it, along with all of the other addresses that you use? The answer is the "domain
name system" or "DNS".
DNS is a large, distributed, database containing mappings between human readable names (such as
"inventory.acme.com") and IP addresses (such as "199.124.84.12") When you ask the computer for the IP address for
"inventory.acme.com", it follows these steps:
1.
Checks to see if that host name is in the local computer’s "host table". (On the AS/400, you can type ’CFGTCP’
and choose opt#10 to work with the host table) If it finds an entry for "inventory.acme.com", it returns this to
your program. If not, it tries step #2.
2.
It tries to contact a DNS server. (The DNS server may be on the same machine, or on another machine on the
network, it doesn’t matter)
3.
The DNS server may already know the IP address that’s associated with "inventory.acme.com". Each time it
looks up a new name, it "caches" it for a period of time. So, if this particular name is in it’s cache, it can return it
right away. If not, it goes on to the next step.
4.
Since there are so many millions (billons?) of host names in the world, you cannot store them all on one server.
Instead, the names are served by an entire hierarchy of DNS servers. Each level of the hierarchy relates to a
different component of the host’s name. The components are separated by periods.
5.
So, "inventory.acme.com" gets separated into "inventory", "acme" and "com". The DNS server asks the "root
level" DNS servers for the server that handles "com" domains. (The DNS server will cache these requests as
well, so once it knows who handles "com" domains, it won’t ask again). The root server returns the IP address
for "com" domains.
6.
The DNS server then asks the server for "com" domains for the address of the server that handles "acme"
domains. The "com" server will then return the address of acme’s DNS server. The DNS server caches this, as
well.
7.
The DNS server asks the "acme" server for the address of the "inventory" host. This address gets returned, and
cached by your DNS server.
8.
Finally, the DNS server returns this address to your program.
1.2. Overview of a TCP communications session
A good analogy to a TCP connection is a telephone call.
When you wish to place a telephone call, you first look up the telephone number of the person you wish to call. If
there are many people who can be reached at that telephone number, you also look up an extension number. You then
lift up the receiver and dial the number. When the person you wish to reach has answered the phone, you talk to that
person. That person talks to you, as well. When your conversation is complete, each person hangs up the telephone.
A TCP connection is very similar. First you look up the IP address (telephone number) of the computer you wish to
contact. For a TCP connection, there are always (the potential for) many different services at that address. So you
have to look up the port (extension number). You then create a socket (pick up the receiver) and dial the number
(connect to the IP and socket). Then the program you wish to contact has accepted your connection (answered the
phone) you can send and receive data from that computer (hold a conversation), and finally each side closes their
socket (hangs up the phone).
3
Chapter 1. Introduction to TCP and Sockets
When you make a telephone call, you don’t have to worry about how your voice gets converted to an electrical
signal, or how it gets switched to the telephone of the person that you’re calling. The telephones and the telephone
company takes care of all of these details for you.
Likewise, when you connect to another computer using TCP, you don’t have to worry about the details of how your
data gets split up into different datagrams, or how it ensures that the data gets put back together in the correct order,
or even how it gets routed to the computer that you’re connecting to. The sockets API takes care of all of these details
for you.
The main procedures that you call from the sockets API are listed here with a brief explanation:
•
gethostbyname = Look up an IP address from a domain name (look up a person’s telephone number)
•
getservbyname = Look up a port number from a service name (look up a person’s extension number)
•
socket = create a socket (lift up the telephone handset)
•
connect = connect to a given IP address and port number (dial the telephone number)
•
bind = force the socket API to utilize a particular port or address. (doesn’t really work with the telephone analogy,
but this is used to tell the API which port number your socket will use on the local machine)
•
listen = Tell the socket API that you wish to receive connections (switching on the "ringer" on your telephone)
•
accept = Accept an incoming connection (answer a telephone call)
•
send = send data out over a socket (talk into your telephone handset)
•
recv = receive data from a socket (listen to your telephone handset)
•
close = close a socket (hang up the telephone)
Therefore, a typical TCP session looks like this:
server
client
getservbyname()
socket()
bind()
listen()
gethostbyname()
getservbyname()
socket()
accept()
connect()
send() & recv()
send() & recv()
close()
close()
The getservbyname() call asks the operating system which port number a server will accept connections on for a
given service.
For the sake of an example, we’ll use http, which is the protocol used to transport web pages across the internet.
The server will start by calling getservbyname to ask the operating system which port is used for the http requests
over tcp. The getservbyname API will return port number 80, which happens to be the worldwide standard for the
http service. The server will then call the socket() procedure, this socket will then be bound to port 80 by calling the
4
Chapter 1. Introduction to TCP and Sockets
bind() procedure. Now, when we call the listen() procedure, the socket API will cause port 80 to be "open" for
receiving requests. The program then calls accept(), which waits for someone to connect to it. Once a client has
connected, it can send and receive data. Finally, when it’s done, it closes the connection.
The client first asks the user for the host to connect to. Once it has a response from the user, it checks to see if the
user supplied an IP address or if the user actually supplied a host name. If the user supplied a host name, it calls
gethostbyname to find the IP address of the host it needs to connect to. Now, it needs to know which port to connect
to, so it calls getservbyname to ask the operating system which port is used for http. The operating system will, of
course, return 80. Now, it creates a socket which it can use to talk to the server by calling the socket() procedure. It
uses the connect() procedure to connect that socket to the server, who is hopefully waiting for it to connect. Now it
can send & receive data from the server. And finally, it calls close() to disconnect.
5
Chapter 2. Looking up host names & services
Written by Scott Klement.
2.1. Services and Ports
The previous sections described an overview of TCP/IP terminology as well as an overview of calling the sockets
API for a TCP connection. This section describes the details of calling "getservbyname()" and explains how to
understand the IBM UNIX-type API manual.
The Socket API is described in the IBM Information Center under the category "UNIX-type APIs." Unfortunately,
the calling syntax that’s described is for people using the ILE C/400 programming language.
The getservbyname() API is documented here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/gsrvnm.htm
(http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/gsrvnm.htm)
And it says that in order to call getservbyname, we need to specify parms like this:
struct servent *getservbyname(char *service_name,
char *protocol_name)
By now, unless you’re familiar with programming in C, you’re probably saying "huh? what does that mean?" So, I’ll
tell you :) Let’s look at that statement again, but break it into pieces:
➊struct servent *➋getservbyname(➌char *service_name, ➍char *protocol_name)
➊
Each prototype in C starts with the prcoedure’s return value. In this case, the ’*’ means that it returns a pointer,
and the ’struct servent’ means that this pointer points to a structure of type ’servent’ (service entry).
➋
The name of the procedure to be called is ’getservbyname’.
➌
The first parameter is a pointer. (I know this because of the ’*’) and it points to a character variable called
"service_name." Character variables are type ’A’ in RPG and DDS.
➍
The second parameter is another pointer to a character variable. This one is called "protocol name".
Since names in C are case sensitive, we MUST supply the name ’getservbyname’ as all lowercase. (That’s important
since RPG likes to convert everything to uppercase at compile-time).
So, to define this same prototype in RPG, we need to do this:
DName+++++++++++ETDsFrom+++To/L+++IDc.Keywords+++++++++++++++++++++++
D getservbyname
PR
*
ExtProc(’getservbyname’)
D
service_name
*
value options(*string)
D
protocol_name
*
value options(*string)
6
Chapter 2. Looking up host names & services
Note that all pointers passed to C procedures are always passed by "value". The keyword "options(*string)" is an
option added to RPG to help make it compatible with C programs. Strings in C end with a "null" character (x’00’)
that allows it to recognize the end of a variable length string. Specifying options(*string) causes RPG to
automatically add this trailing null when it makes the call.
If you look further down, (in the return value section) they describe this "servent" structure by showing us this:
struct servent {
char
char
int
char
};
*s_name;
➊
**s_aliases;➋
s_port;
➌
*s_proto
➍
This means that the servent structure contains 4 pieces of data:
➊
A pointer to an area of memory containing an alphanumeric string. This piece of data is called "s_name".
➋
This one is quite difficult for your typical RPG programmer to understand. It is a pointer to an array. That array
is an array of pointers. Each element of that array points to an alphanumeric string.
➌
An integer called "s_port".
➍
The last piece of data is a pointer to an alphanumeric string call "s_proto".
So, if we wanted to define this structure in RPG, we’d do it like this:
DName+++++++++++ETDsFrom+++To/L+++IDc.Keywords+++++++++++++++++++++++
D p_servent
S
*
D servent
DS
based(p_servent)
D
s_name
*
D
s_aliases
*
D
s_port
10I 0
D
s_proto
*
As you can probably tell from this reading this page, so far, the hardest part of calling the UNIX-type APIs from
RPG programs is converting things from C syntax to RPG syntax. That’s why it’s so frustrating that IBM doesn’t
provide a manual and header files specifically for RPG programmers.
For this tutorial, I will show you how to make the conversions from the C prototypes to the RPG prototypes. If you
want to learn more about how to do this yourself, you might try this link: http://www.opensource400.org/callc.html
But, anyway... I’ll get off the soap box and get back to the tutorial...
Once we have set up this data structure and procedure prototype, we can call the getservbyname procedure simply by
coding:
CL0N01Factor1+++++++Opcode&ExtExtended-factor2+++++++++++++++++++++++++
C
eval
p_servent = getservbyname(’http’: ’tcp’)
7
Chapter 2. Looking up host names & services
And then check the value of s_port to find out which port http is located on. As you can see, defining the prototype
and data structure is the hard part. Once we’ve got that, making the call is easy.
So, here’s a simple program that accepts a service name as a parameter, and displays the corresponding port number:
Member: QRPGLESRC, SERVPORT
H DFTACTGRP(*NO) ACTGRP(*NEW)
D getservbyname
D service_name
D protocol_name
PR
*
*
*
D p_servent
D servent
D
s_name
D
s_aliases
D
s_port
D
s_proto
S
DS
*
D service
D msg
S
S
c
c
*entry
ExtProc(’getservbyname’)
value options(*string)
value options(*string)
based(p_servent)
*
*
10I 0
*
10A
50A
plist
parm
service
c
c
eval
p_servent = getservbyname(
%trim(service): ’tcp’)
c
c
c
c
c
if
eval
else
eval
endif
p_servent = *NULL
➋
msg = ’No such service found!’
c
dsply
msg
c
eval
➊
msg = ’port = ’ + %editc(s_port:’L’)
*inlr = *on
Compile this by typing: CRTBNDRPG SERVPORT SRCFILE(SOCKTUT/QRPGLESRC)
Run it by typing CALL SERVPORT PARM(’http’)
A few notes:
➊
Note that we must do a %trim() on the service when we call getservbyname. Otherwise, it would search for a
service called ’http ’ instead of ’http’. Since service names can’t legally have spaces in them, this wouldn’t work.
➋
The IBM manual states that getservbyname will return NULL when it can’t find a service. In RPG, the special
value NULL can be represented by the special constant *NULL, so we check for *NULL to see that the service
wasn’t found.
8
Chapter 2. Looking up host names & services
2.2. Host names and addresses
The previous section described how to look up the port number of a service using the getservbyname() API. This
section complements that section by desribing the procedure for looking up an IP address using the gethostbyname()
procedure. This section will also describe the system routines for converting from "dotted decimal" IP address format
to the 32-bit address format, and back again.
To start with, we’ll explain the ’inet_addr’ API. It’s the system API to convert from dotted decimal formatted IP
address to a 32-bit network address. This API is explained in IBM’s manuals at the following link:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/inaddr.htm
(http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/inaddr.htm)
IBM tells us that the prototype for the inet_addr procedure looks like this:
unsigned long inet_addr(char *address_string)
This means that the procedure is named ’inet_addr’, and it accepts one parameter, called ’address_string’, which is a
pointer to a character string. It returns an unsigned long integer, which in RPG is expressed as ’10U 0’.
This is pretty straight forward in RPG. We simply prototype it like this:
D inet_addr
D address_str
PR
10U 0 ExtProc(’inet_addr’)
*
value options(*string)
Reading further down the IBM page on this procedure, you’ll see that it returns the 32-bit network address when
successful. And returns a -1 when it’s not successful.
Now you should be confused! If it returns an ’unsigned’ integer, how can it possibly return a negative number? By
definition, an unsigned integer CANNOT be a negative number. In C, when the return value is compared against -1,
the computer actually generates a 32-bit integer with the value of -1, and then compares, bit for bit, whether they
match, even though one is signed and one isn’t, as long as they have the same bit values, it thinks they’re the same.
So the question is, what ’unsigned’ value has the same binary value as the signed integer value of -1? Well, if you’re
familiar with the twos complement format, you’ll know in order to make a positive number negative, you take it’s
binary complement, and add 1. Therefore, x’00000001’ is a positive 1, and it’s binary complement would be
x’FFFFFFFE’. If we add 1 to that, we get x’FFFFFFFF’. IF we convert that to decimal, we get the value 4294967295.
Do you still think this is straight-forward? :) To simplify things somewhat, we’ll define a named-constant called
’INADDR_NONE’ (which, though they dont tell us that, is the name of the same constant in the C header file) When
we check for errors, we’ll simply check against that constant.
D INADDR_NONE
C
CONST(4294967295)
Once we have all the definitions straightened out, this is quite easy to call in RPG:
c
c
c*
c
eval
if
conversion failed
endif
IP = inet_addr(’192.168.66.21’)
IP = INADDR_NONE
9
Chapter 2. Looking up host names & services
To convert an IP address from 32-bit network format back to dotted decimal uses a similar API called ’inet_ntoa’.
The IBM manual for this is found at this link:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/inntoa.htm
IBM tells us that the prototype for inet_ntoa looks like this:
char *inet_ntoa(struct in_addr internet_address)
The procedure is called inet_ntoa. It accepts one parameter, which is a ’in_addr’ structure, passed by value. It returns
a pointer to a character string. That’s nice. But what the heck is an ’in_addr’ structure?
You’ll note that the return value doesn’t even bother to explain this structure. If you have the "system openness
includes" installed on your system, you’ll be able find the definition for in_addr in the file QSYSINC/NETINET in
member IN. It looks like this:
struct in_addr {
u_long s_addr;
};
What this means is that the structure contains nothing but a single, unsigned, 32-bit integer. That’s a bit confusing to
try to figure out from IBM’s documentation, but it’s easy to code. The prototype for inet_ntoa in RPG will look like
this:
D inet_ntoa
D internet_addr
PR
*
ExtProc(’inet_ntoa’)
10U 0 value
Now, when you call inet_ntoa, it will either return NULL (which in RPG is called *NULL) if there’s an error, or it
will return a pointer to a character string. Unfortunately, we don’t know how long this character string that it returns
is! In C, you’d find the length of this string by searching for the ’null’ character at the end of the string.
Fortunately, the RPG compiler developers created a BIF for just that purpose. This BIF is called %str, and it accepts
a pointer, and returns the string beginning with the pointer, and ending with the trailing null. So we can call inet_ntoa
like this:
c
c
c*
c
c
c
eval
if
conversion failed
else
eval
endif
ptr = inet_ntoa(IP)
ptr = *NULL
dottedIP = %str(ptr)
The final API in this chapter is the ’gethostbyname’ API, which does a DNS lookup on a given host name, and
returns a ’host entry structure’ that contains the details about that host. The main piece of data that we’re interested in
from that structure is the host’s IP address.
IBM’s manual page for this procedure is found at this link:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/ghostnm.htm
The prototype IBM lists for this API looks like this:
struct hostent *gethostbyname(char *host_name)
10
Chapter 2. Looking up host names & services
Which means that the API is ’gethostbyname’, it accepts one parameter which is a pointer to a character string
containing the host name. It returns a pointer to a ’hostent’ structure.
So, the RPG prototype for the gethostbyname API looks like this:
D gethostbyname
D
host_name
PR
*
*
extproc(’gethostbyname’)
value options(*string)
Looking down at the "return value" of the API, it tells us that it either returns NULL (or, in RPG, *NULL) if there’s
an error, or it returns a pointer to a ’hostent’ structure. The hostent structure is described as being in this format:
struct hostent {
char
*h_name;
char
**h_aliases;
int
h_addrtype;
int
h_length;
char
**h_addr_list;
};
#define h_addr
h_addr_list[0]
This means that the hostent data structure contains 5 different items, a pointer to a character string containing the
’correct’ hostname, a pointer to an array of pointers, each element pointing to an ’alias’ of the hostname, an integer
containing the address type, an integer containing the length of each address, and a pointer to an array of pointers,
each one pointing to a character string.
The ’#define’ statement tells us that whenever we refer to ’h_addr’ in this structure, that it will actually return the
first pointer in the h_addr_list array.
If you read on, under the "Return" value, it explains that both of the arrays pointed to here are ’NULL-terminated
lists’. This means that if you loop thru the array of pointers, you’ll know you’ve reached the end when you find a
’NULL’ (or *NULL in RPG).
It further tells you that each element of the h_addr_list array actually points to a structure of type ’in_addr’, rather
than a character string, as the structure definition implies. Why does it do that? Because gethostbyname can be used
with other protocols besides TCP/IP. In the case of those other protocols, it might point to something besides an
’in_addr’ structure. In fact, we could use the ’h_addrtype’ member of the structure to determine what type of address
was returned, if we so desired. However, this document is only intended to work with TCP/IP addresses.
Note, also, that we already know from our experiences with inet_ntoa above that ’in_addr’ structures are merely an
unsigned integer in disguise. Therefore, we can define the ’hostent’ structure as follows:
D p_hostent
D hostent
D
h_name
D
h_aliases
D
h_addrtype
D
h_length
D
h_addr_list
D p_h_addr
S
DS
S
*
Based(p_hostent)
*
*
10I 0
10I 0
*
*
Based(h_addr_list)
11
Chapter 2. Looking up host names & services
D h_addr
S
10U 0 Based(p_h_addr)
So, if we want to refer to the first IP address listed under h_addr_list, we can now just refer to h_addr. If we need to
deal with later addresses in the list (which is very unusual, in my experiences) we can do so like this:
D addr_list
D p_oneaddr
D oneaddr
S
S
S
*
DIM(100) Based(h_addr_list)
*
10U 0 based(p_oneaddr)
CL0N01Factor1+++++++Opcode&ExtFactor2+++++++Result++++++++Len++D+HiLo
C
do
100
x
C
if
addr_list(X) = *NULL
C
leave
C
endif
C
eval
p_oneaddr = addr_list(X)
C*** oneaddr now contains the IP address of this
C***
position of the array, use it now...
C
enddo
As I stated, however, it’s very unusual to need to work with anything but the first address in the list. Therefore,
’h_addr’ is usually what we’ll refer to when we want to get an IP address using gethostbyname.
And we’d normally call gethostbyname like this:
c
c
c
c
eval
if
eval
endif
p_hostent = gethostbyname(’www.ibm.com’)
p_hostent <> *NULL
IP = h_addr
Now that we know the basic syntax to call these APIs, lets put them together and write a program that uses them.
When we write a client program for TCP/IP, usually the first thing we need to do is figure out the IP address to
connect to. Generally, this is supplied by the user, and the user will either supply an IP address directly (in the
dotted-decimal format) or he’ll supply a hostname that needs to be looked up.
The way that we go about doing this is by first calling ’inet_addr’ to see if the host is a valid dotted-decimal format
IP address. If it is not, we’ll actually call ’gethostbyname’ to look the address up using DNS and/or the host table. In
our example program, we’ll then either print a ’host not found’ error message, or we’ll call ’inet_ntoa’ to get a
dotted-decimal format IP address that we can print to the screen.
So here’s our example of looking up an IP address for a hostname:
File: SOCKTUT/QRPGLESRC, Member: DNSLOOKUP
H DFTACTGRP(*NO) ACTGRP(*NEW)
D inet_addr
D address_str
PR
D INADDR_NONE
C
D inet_ntoa
PR
10U 0 ExtProc(’inet_addr’)
*
value options(*string)
CONST(4294967295)
*
ExtProc(’inet_ntoa’)
12
Chapter 2. Looking up host names & services
D
D
D
D
D
D
D
D
D
D
internet_addr
10U 0 value
p_hostent
hostent
h_name
h_aliases
h_addrtype
h_length
h_addr_list
p_h_addr
h_addr
S
DS
D gethostbyname
D
host_name
PR
*
*
D host
D IP
D Msg
S
S
S
32A
10U 0
50A
c
c
*entry
*
Based(p_hostent)
*
*
10I 0
10I 0
*
*
Based(h_addr_list)
10U 0 Based(p_h_addr)
S
S
extproc(’gethostbyname’)
value options(*string)
plist
parm
host
c
eval
IP = inet_addr(%trim(host))
c
c
c
c
c
c
if
eval
if
eval
endif
endif
IP = INADDR_NONE
p_hostent = gethostbyname(%trim(host))
p_hostent <> *NULL
IP = h_addr
c
c
c
c
c
if
eval
else
eval
endif
IP = INADDR_NONE
Msg = ’Host not found!’
c
dsply
c
eval
Msg = ’IP = ’ + %str(inet_ntoa(IP))
Msg
*inlr = *on
This program can be compiled with: CRTBNDRPG DNSLOOKUP SRCFILE(xxx/QRPGLESRC)
And you can call it like this: CALL DNSLOOKUP PARM(’www.yahoo.com’) or CALL DNSLOOKUP
PARM(’www.scottklement.com’) or with a dotted-decimal address like CALL DNSLOOKUP
PARM(’192.168.66.21’)
After all that explanation, the code itself doesn’t seem so complicated, does it? :)
13
Chapter 3. Socket API calls to create our first
client program
Written by Scott Klement.
3.1. The socket() API call
The previous sections explained how to find out the port number for a service name, and how to get the IP address
for a host name. This section will utilize that information to create a simple client program.
The socket() API is used to create a socket. You can think of a socket as being the virtual device that is used to read
& write data from a network connection.
The IBM manual page that documents the socket API is at this link:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/socket.htm
It lists the following as the prototype for the socket() API:
int socket(int address_family,
int type,
int protocol)
This tells us that the name of the procedure to call is ’socket’, and that it accepts 3 parameters. Each one is an integer,
passed by value. It also returns an integer. Therefore, the RPG prototype for the socket() API looks like this:
D socket
D addr_family
D type
D protocol
PR
10I
10I
10I
10I
0
0
0
0
ExtProc(’socket’)
value
value
value
It’s important to realize that the socket APIs can be used for other networking protocols besides TCP/IP. When we
create a socket, we need to explain to the socket API that we wish to communicate using the IP protocol, and that we
wish to use TCP on top of the IP protocol.
For address family, the manual tells us that we need to specify a value of ’AF_INET’ if we wish to do network
programming in the ’Internet domain’. Therefore, when we specify a value of ’AF_INET’, what we’re really telling
the API is to ’use the IP protocol’.
Under the ’type’ parameter, it allows us to give values of ’SOCK_DGRAM’, ’SOCK_SEQPACKET’,
’SOCK_STREAM’ or ’SOCK_RAW’. The TCP protocol is the standard streaming protocol for use over IP. So, if we
say ’SOCK_STREAM’, we’ll use the TCP protocol. As you might imagine, SOCK_DGRAM is used for the UDP
protocol and SOCK_RAW is used for writing raw IP datagrams.
Finally, we specify which protocol we wish to use with our socket. Note that, again, we can specify IPPROTO_TCP
for TCP, IPPROTO_UDP for UDP, etc. However, this isn’t necessary! Because we already specified that we wanted a
’stream socket’ over ’internet domain’, it already knows that it should be using TCP. Therefore, we can specify
’IPPROTO_IP’ if we want, and the API will use the default protocol for the socket type.
14
Chapter 3. Socket API calls to create our first client program
Now, we just have one problem: We don’t know what integer values AF_INET, SOCK_STREAM and IPPPROTO_IP
are! IBM is referencing named constants that they’ve defined in the appropriate header files for C programs, but we
don’t have these defined for us in RPG! But, if you do a bit of snooping into the ’System Openness Includes’ library,
you’ll find that AF_INET is defined to be ’2’, SOCK_STREAM is defined to be ’1’, and IPPROTO_IP is defined as
’0’. To make this easier for us, we’ll make named constants that match these values, like so:
D AF_INET
D SOCK_STREAM
D IPPROTO_IP
C
C
C
CONST(2)
CONST(1)
CONST(0)
Now we can call the socket() API like so:
c
eval
s = socket(AF_INET:SOCK_STREAM:IPPROTO_IP)
3.2. The connect() API call
Once we have a socket to work with, we need to connect it to something. We do that using the connect() API, which
is documented in IBM’s manual at this location:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/connec.htm
It tells us here that the prototype for the connect() API looks like this:
int connect(int socket_descriptor,
struct sockaddr *destination_address,
int address_length)
So, as you can see, the procedure is named ’connect’, and it accepts 3 parameters. An integer, a pointer to a
’sockaddr’ structure, and another integer. It also returns an integer. This means that the RPG prototype will look like
this:
D connect
D sock_desc
D dest_addr
D addr_len
PR
10I 0 ExtProc(’connect’)
10I 0 value
*
value
10I 0 value
Looking further down the manual, we see that the a ’sockaddr’ structure is defined as follows:
struct sockaddr {
u_short sa_family;
char
sa_data[14];
};
Remember, the purpose of this structure is to tell the API which IP address and port number to connect to. Why,
then, doesn’t it contain fields that we can put the address and port numbers into? Again, we have to remember that
the socket APIs can work with many different network protocols. Each protocol has a completely different format for
15
Chapter 3. Socket API calls to create our first client program
how addresses work. This ’sockaddr’ structure is, therefore, a generic structure. It contains a place to put the
identifying address family, along with a generic "data" field that the address can be placed in, regardless of the
format of the address.
Although it’s not documented on IBM’s page for the connect() API, there is actually a different structure called
’sockaddr_in’ which is designed especially for internet addresses. The C definition for sockaddr_in can be found in
the file QSYSINC/NETINET, member IN, if you have the System Openness Includes loaded. It looks like this:
struct sockaddr_in {
short sin_family;
u_short sin_port;
struct in_addr sin_addr;
char sin_zero[8];
};
/*
/*
/*
/*
/*
socket address (internet)
address family (AF_INET)
port number
IP address
reserved - must be 0x00’s
*/
*/
*/
*/
*/
To make it easier to use these structures in RPG, I like to make them based in the same area of memory. This means
that you can look at the data as a ’sockaddr’, or the same data as a ’sockaddr_in’ without moving the data around.
Having said that, here’s the definition that I use for the sockaddr & sockaddr_in structures:
D p_sockaddr
D sockaddr
D
sa_family
D
sa_data
D sockaddr_in
D
sin_family
D
sin_port
D
sin_addr
D
sin_zero
S
DS
*
based(p_sockaddr)
5I 0
14A
DS
based(p_sockaddr)
5I 0
5U 0
10U 0
8A
Before we can call the connect() API, we need to ask the operating system for some memory that we can store our
sockaddr structure into. Then, we can populate the sockaddr_in structure, and actually call the connect() API. Like
so:
DName+++++++++++ETDsFrom+++To/L+++IDc.Keywords+++++++++++++++++++
D p_connto
S
*
D addrlen
S
10I 0
C* Ask the operating system for some memory to store our socket
C* address into:
C
eval
addrlen = %size(sockaddr)
C
alloc
addrlen
p_connto
C* Point the socket address structure at the newly allocated
C*
area of memory:
C
eval
p_sockaddr = p_connto
C* Populate the sockaddr_in structure
C* Note that IP is the ip address we previously looked up
C* using the inet_addr and/or gethostbyname APIs
C* and port is the port number that we looked up using the
C* getservbyname API.
16
Chapter 3. Socket API calls to create our first client program
C
C
C
C
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
IP
PORT
*ALLx’00’
C* Finally, we can connect to a server:
C
if
connect(s: p_connto: addrlen) < 0
C*** Connect failed, report error here
C
endif
3.3. The send() and recv() API calls
Once we’ve made a connection, we’ll want to use that connection to send and receive data across the network. We’ll
do that using the send() and recv() APIs.
IBM’s manual page for the send() API can be found at this link:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/send.htm
It tells us that the prototype for the send() API looks like this:
int send(int socket_descriptor,
char *buffer,
int buffer_length,
int flags)
Yes, the procedure is called ’send’, and it accepts 4 parameters. Those parameters are an integer, a pointer, an integer
and another integer. The send() API also returns an integer. Therefore, the RPG prototype for this API is:
D send
D
sock_desc
D
buffer
D
buffer_len
D
flags
PR
10I
10I
*
10I
10I
0 ExtProc(’send’)
0 value
value
0 value
0 value
You may have noticed that for other ’char *’ definitions, we put the ’options(*string)’ keyword in our D-specs, but
we didn’t this time. Why? Because the send() API doesn’t use a trailing null-character to determine the end of the
data to send. Instead, it uses the buffer_length parameter to determine how much data to send.
That is a useful feature to us, because it means that we are able to transmit the null-character over the network
connection as well as the rest of the string, if we so desire.
The flags parameter is used for ’out of band data’, and for sending ’non-routed’ data. You’ll almost never use these
flags. Why? Because ’out of band data’ has never been widely adopted. Many TCP/IP stacks don’t even implement it
properly. In fact, for a long time, sending ’out-of-band’ data to a Windows machine caused it to crash. The popular
program called ’winnuke’ does nothing more than send some out-of-band data to a Windows machine. The other
flag, ’dont route’ is really only used when writing routing applications. In all other situations, you want your packets
to be routed! Therefore, it’s very rare for us to specify anything but a 0 in the flags parameter.
The return value of the send() API will be the number of bytes sent, or a negative number if an error occurred.
17
Chapter 3. Socket API calls to create our first client program
Consequently, we typically call the send() API like this:
D miscdata
D rc
C
C
c
C*
C
S
S
25A
10I 0
eval
miscdata = ’The data to send goes here’
eval
rc = send(s: %addr(miscdata): 25: 0)
if
rc < 25
for some reason we weren’t able to send all 25 bytes!
endif
The recv() API is used to receive data over the network. IBM has documented this API here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/recv.htm
recv() is very similar to send(). In fact, the prototype for recv is nearly identical to send(), the only difference is the
name of the procedure that you call. The prototype looks like this:
int recv(int socket_descriptor,
char *buffer,
int buffer_length,
int flags)
And, just like send, the RPG prototype looks like this:
D recv
D
sock_desc
D
buffer
D
buffer_len
D
flags
PR
10I
10I
*
10I
10I
0 ExtProc(’recv’)
0 value
value
0 value
0 value
The obvious difference between send() and recv() is what the system does with the memory pointed to by the ’buffer’
parameter. When using send(), the data in the buffer is written out to the network. When using recv(), data is read
from the network and is written to the buffer.
Another, less obvious, difference is how much data gets processed on each call to these APIs. By default, when you
call the send() API, the API call won’t return control to your program until the entire buffer has been written out to
the network. By contrast, the recv() API will receive all of the data that’s currently waiting for your application.
By default, recv() will always wait for at least one byte to be received. But, if there are more bytes, it will return them
all, up to the length of the buffer that you’ve requested.
In the send() example above, 25 bytes are always written to the network unless an error has occurred. In the recv()
example below, we can receive anywhere from 1 to 25 bytes of data. We have to check the return code of the recv()
API to see how much we actually received.
Here’s a quick example of calling recv():
D miscdata
D rc
C
c
S
S
25A
10I 0
eval
if
rc = recv(s: %addr(miscdata): 25: 0)
rc < 1
18
Chapter 3. Socket API calls to create our first client program
C*
C
Something is wrong, we didnt receive anything.
endif
3.4. Translating from ASCII to EBCDIC
Almost all network communications use the ASCII character set, but the AS/400 natively uses the EBCDIC
character set. Clearly, once we’re sending and receiving data over the network, we’ll need to be able to translate
between the two.
There are many different ways to translate between ASCII and EBCDIC. The API that we’ll use to do this is called
QDCXLATE, and you can find it in IBM’s information center at the following link:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/QDCXLATE.htm
There are other APIs that can be used to do these conversions. In particular, the iconv() set of APIs does really a good
job, however, QDCXLATE is the easiest to use, and will work just fine for our purposes.
The QDCXLATE API takes the following parameters:
Parm# Description
1
Length of data to convert
2
Data to convert
3
Conversion table
Usage
Input
Data Type
Packed (5,0)
I/O
Char (*)
Input
Char (10)
And, since QDCXLATE is an OPM API, we actually call it as a program. Traditionally, you’d call an OPM API with
the RPG ’CALL’ statement, like this:
C
C
C
C
CALL
PARM
PARM
PARM
’QDCXLATE’
128
’QTCPEBC’
LENGTH
DATA
TABLE
5 0
128
10
However, I find it easier to code program calls using prototypes, just as I use for procedure calls. So, when I call
QDCXLATE, I will use the following syntax:
D Translate
D
Length
D
Data
D
Table
C
PR
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
callp
Translate(128: Data: ’QTCPEBC’)
There are certain advantages to using the prototyped call. The first, and most obvious, is that each time we want to
call the program, we can do it in one line of code. The next is that the ’const’ keyword allows the compiler to
automatically convert expressions or numeric variables to the data type required by the call. Finally, the prototype
allows the compiler to do more thorough syntax checking when calling the procedure.
19
Chapter 3. Socket API calls to create our first client program
There are two tables that we will use in our examples, QTCPASC and QTCPEBC. These tables are easy to remember
if we just keep in mind that the table name specifies the character set that we want to translate the data into. In other
words ’QTCPEBC’ is the IBM-supplied table for translating TCP to EBCDIC (from ASCII) and QTCPASC is the
IBM supplied table for translating TCP data to ASCII (from EBCDIC).
3.5. The close() API call
This section documents the easiest, by far, socket API to call. The close() API. This API is used to disconnect a
connected socket, and destroy the socket descriptor. (In other words, to use the socket again after calling close() you
have to call socket() again).
Here’s the IBM manual page that describes the close() API:<
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/close.htm
The manual tells us that the prototype for close() looks like this:
int close(int fildes)
So, the procedure’s name is ’close’ and it accepts one parameter, an integer. It also returns an integer. The RPG
prototype looks like this:
D close
D sock_desc
PR
10I 0 ExtProc(’close’)
10I 0 value
To call it, we can simply to:
C
eval
C
if
C*** Socket didn’t close.
c
endif
rc = close(s)
rc < 0
Now what?
Or, more commonly (because there isn’t much we can do if close() fails) we do something like this:
C
callp
close(s)
Too easy to be a UNIX-Type API, right? Well, never fear, there’s one complication. The system uses the same close()
API for closing sockets that it uses for closing files in the integrated file system.
This means that if you use both sockets and IFS reads/writes in your program, that you only need to define one
prototype for close(). Handy, right? Unfortunately, most people put all of the definitions needed for socket APIs into
one source member that they can /COPY into all of the programs that need it. Likewise, the IFS prototypes and other
definitions are put into their own /COPY member.
When you try to use both /COPY members in the same program, you end up with a duplicate definition of the close()
API, causing the compiler to become unhappy.
20
Chapter 3. Socket API calls to create our first client program
The solution is relatively simple... When we make a header file for either sockets or IFS, we use the RPG /define and
/if defined directives to force it to only include the close() prototype once. So, our prototype will usually look like
this:
D/if not defined(CLOSE_PROTOTYPE)
D close
PR
10I 0 ExtProc(’close’)
D sock_desc
10I 0 value
D/define CLOSE_PROTOTYPE
D/endif
3.6. Our first client program
We’ve learned a lot of new API calls over the past few sections. It’s time to put these new APIs to use with an
example program.
This program is a very simple http client. It connects to a web server on the internet, and requests that a web page (or
another file) on the server be sent back to it. It then receives the data that the web server returns and displays it on the
screen.
It’s important to understand that most data that is sent or received over the internet uses the concept of ’lines of text.’
A line of text is a variable-length string of bytes, you can tell the end of a line by looking for the ’carriage-return’ and
’line-feed’ characters. When these characters appear in the text, it means that its time to start a new line.
In ASCII, the ’carriage-return’ character (CR) is x’0D’, and the ’line-feed’ character is x’0A’. When translated to
EBCDIC, these are x’0D’ and x’25’, respectively.
Therefore, the pseudocode for this client program looks like this:
1.
look up the port number for the HTTP service and store it into the variable ’port’.
2.
look up the IP address for the hostname that was passed as a parameter, and store it in the variable ’IP’.
3.
call the socket() API to create a socket that we can use for communicating with the HTTP server.
4.
create a socket address structure (’sockaddr’) that will tell the connect() API which host & service to connect to.
5.
call the connect() API to connect to the HTTP server.
6.
place a two-line ’request’ into a variable. The first line will contain the phrase "GET /pathname/filename
HTTP/1.0" which tells the HTTP server that we wish to get a file from it, and also tells the HTTP server where
that file is. The "HTTP/1.0" means that we’re using version 1.0 of the HTTP specifications (more about that
later) The second line of the request is blank, that’s how we tell the server that we’re done sending requests.
7.
Translate our request to ASCII so the server will understand it.
8.
Call the send() API to send our request to the server.
9.
Call the recv() API to read back 1 byte of the server’s reply.
10. If an error occurred (that is, the server disconnected us) then we’re done receiving, jump ahead to step 13.
11. If the byte that we’ve read is not the ’end-of-line’ character, and our receive buffer isn’t full, then add the byte to
the end of the receive buffer, and go to step 9.
21
Chapter 3. Socket API calls to create our first client program
12. Translate the receive buffer to EBCDIC so we can read it, and display the receive buffer. Then go back to step 9
to get the next line of data.
13. Close the connection.
14. Pause the screen so the user can see what we received before the program ends.
Without further ado, here’s the sample program, utilizing all of the concepts from the past few sections of this
tutorial:
File: SOCKTUT/QRPGLESRC, Member: CLIENTEX1
H DFTACTGRP(*NO) ACTGRP(*NEW)
D getservbyname
D service_name
D protocol_name
PR
*
*
*
D p_servent
D servent
D
s_name
D
s_aliases
D
s_port
D
s_proto
S
DS
*
D inet_addr
D address_str
PR
D INADDR_NONE
C
D inet_ntoa
D internet_addr
PR
D
D
D
D
D
D
D
D
D
p_hostent
hostent
h_name
h_aliases
h_addrtype
h_length
h_addr_list
p_h_addr
h_addr
S
DS
D gethostbyname
D
host_name
PR
*
*
D socket
D addr_family
D type
D protocol
PR
10I
10I
10I
10I
D AF_INET
D SOCK_STREAM
D IPPROTO_IP
C
C
C
ExtProc(’getservbyname’)
value options(*string)
value options(*string)
based(p_servent)
*
*
10I 0
*
S
S
10U 0 ExtProc(’inet_addr’)
*
value options(*string)
CONST(4294967295)
*
ExtProc(’inet_ntoa’)
10U 0 value
*
Based(p_hostent)
*
*
10I 0
10I 0
*
*
Based(h_addr_list)
10U 0 Based(p_h_addr)
extproc(’gethostbyname’)
value options(*string)
0
0
0
0
ExtProc(’socket’)
value
value
value
CONST(2)
CONST(1)
CONST(0)
22
Chapter 3. Socket API calls to create our first client program
D connect
D sock_desc
D dest_addr
D addr_len
PR
D p_sockaddr
D sockaddr
D
sa_family
D
sa_data
D sockaddr_in
D
sin_family
D
sin_port
D
sin_addr
D
sin_zero
S
DS
D send
D
sock_desc
D
buffer
D
buffer_len
D
flags
PR
10I
10I
*
10I
10I
0 ExtProc(’send’)
0 value
value
0 value
0 value
D recv
D
sock_desc
D
buffer
D
buffer_len
D
flags
PR
10I
10I
*
10I
10I
0 ExtProc(’recv’)
0 value
value
0 value
0 value
D close
D sock_desc
PR
10I 0 ExtProc(’close’)
10I 0 value
D translate
D
length
D
data
D
table
PR
D
D
D
D
D
D
D
D
D
D
D
D
D
D
S
S
S
S
S
s
s
s
S
S
S
S
S
S
msg
sock
port
addrlen
ch
host
file
IP
p_Connto
RC
Request
ReqLen
RecBuf
RecLen
10I 0 ExtProc(’connect’)
10I 0 value
*
value
10I 0 value
*
based(p_sockaddr)
5I 0
14A
DS
based(p_sockaddr)
5I 0
5U 0
10U 0
8A
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
50A
10I
5U
10I
1A
32A
32A
10U
*
10I
60A
10I
50A
10I
0
0
0
0
0
0
0
C*************************************************
C* The user will supply a hostname and file
23
Chapter 3. Socket API calls to create our first client program
C* name as parameters to our program...
C*************************************************
c
*entry
plist
c
parm
host
c
parm
file
c
eval
*inlr = *on
C*************************************************
C* what port is the http service located on?
C*************************************************
c
eval
p_servent = getservbyname(’http’:’tcp’)
c
if
p_servent = *NULL
c
eval
msg = ’Can”t find the http service!’
c
dsply
msg
c
return
c
endif
c
eval
port = s_port
C*************************************************
C* Get the 32-bit network IP address for the host
C* that was supplied by the user:
C*************************************************
c
eval
IP = inet_addr(%trim(host))
c
if
IP = INADDR_NONE
c
eval
p_hostent = gethostbyname(%trim(host))
c
if
p_hostent = *NULL
c
eval
msg = ’Unable to find that host!’
c
dsply
msg
c
return
c
endif
c
eval
IP = h_addr
c
endif
C*************************************************
C* Create a socket
C*************************************************
c
eval
sock = socket(AF_INET: SOCK_STREAM:
c
IPPROTO_IP)
c
if
sock < 0
c
eval
msg = ’Error calling socket()!’
c
dsply
msg
c
return
c
endif
C*************************************************
C* Create a socket address structure that
C*
describes the host & port we wanted to
C*
connect to
C*************************************************
c
eval
addrlen = %size(sockaddr)
c
alloc
addrlen
p_connto
24
Chapter 3. Socket API calls to create our first client program
c
c
c
c
c
eval
eval
eval
eval
eval
p_sockaddr
sin_family
sin_addr =
sin_port =
sin_zero =
= p_connto
= AF_INET
IP
port
*ALLx’00’
C*************************************************
C* Connect to the requested host
C*************************************************
C
if
connect(sock: p_connto: addrlen) < 0
c
eval
msg = ’unable to connect to server!’
c
dsply
msg
c
callp
close(sock)
c
return
c
endif
C*************************************************
C* Format a request for the file that we’d like
C* the http server to send us:
C*************************************************
c
eval
request = ’GET ’ + %trim(file) +
c
’ HTTP/1.0’ + x’0D25’ + x’0D25’
c
eval
reqlen = %len(%trim(request))
c
callp
Translate(reqlen: request: ’QTCPASC’)
C*************************************************
c* Send the request to the http server
C*************************************************
c
eval
rc = send(sock: %addr(request): reqlen:0)
c
if
rc < reqlen
c
eval
Msg = ’Unable to send entire request!’
c
dsply
msg
c
callp
close(sock)
c
return
c
endif
C*************************************************
C* Get back the server’s response
C*************************************************
c
dou
rc < 1
C
exsr
DsplyLine
c
enddo
C*************************************************
C* We’re done, so close the socket.
C*
do a dsply with input to pause the display
C*
and then end the program
C*************************************************
c
callp
close(sock)
c
dsply
pause
c
return
1
25
Chapter 3. Socket API calls to create our first client program
C*===============================================================
C* This subroutine receives one line of text from a server and
C* displays it on the screen using the DSPLY op-code
C*===============================================================
CSR
DsplyLine
begsr
C*-----------------------C*************************************************
C* Receive one line of text from the HTTP server.
C* note that "lines of text" vary in length,
C* but always end with the ASCII values for CR
C* and LF. CR = x’0D’ and LF = x’0A’
C*
C* The easiest way for us to work with this data
C* is to receive it one byte at a time until we
C* get the LF character.
Each time we receive
C* a byte, we add it to our receive buffer.
C*************************************************
c
eval
reclen = 0
c
eval
recbuf = *blanks
c
c
c
c
c
c
c
c
c
c
dou
eval
if
leave
endif
if
eval
eval
endif
enddo
reclen = 50 or ch = x’0A’
rc = recv(sock: %addr(ch): 1: 0)
rc < 1
ch<>x’0D’ and ch<>x’0A’
reclen = reclen + 1
%subst(recbuf:reclen:1) = ch
C*************************************************
C* translate the line of text into EBCDIC
C* (to make it readable) and display it
C*************************************************
c
if
reclen > 0
c
callp
Translate(reclen: recbuf: ’QTCPEBC’)
c
endif
c
recbuf
dsply
C*-----------------------Csr
endsr
Compile this program with: CRTBNDRPG PGM(CLIENTEX1) SRCFILE(xxx/QRPGLESRC) DBGVIEW(*LIST)
Run the program by typing: CALL CLIENTEX1 PARM(’ods.ods.net’ ’/index.html’)
(You should be able to use this to retrieve just about any web page)
There are a lot of things that we could improve about this client. We’ll discuss these, and start implementing the
improvements, in the upcoming sections.
26
Chapter 4. Improving our client program
Written by Scott Klement.
4.1. improvement #1: header files
Now that we’ve had a "basic example" of sending and receiving some information from a server program, let’s take a
look at how we can improve our example.
The first improvement that we should make will involve converting all of the prototypes, constants and data
structures that we use for sockets into a "header file", and this will be described in this section:
First of all, our example contains about 75 lines of code that are simply prototypes, constants and data structures
which will be used in just about every program that we write using sockets. One way that we can improve our
program is to place them into their own source member, use the /copy compiler directive to add them into our
programs. This gives us the following benefits:
•
Each time we want to use them, we just /copy them. We don’t have to re-type them, and if we find a mistake in on
of our definitions, we only have to fix it in one place.
•
We don’t have to remember every detail of every API we call, by putting the prototypes into a /copy member, we
can use that copy member as a "cheat-sheet" of how to call a given socket function. In fact, with judicious use of
comments, we can often save ourselves a trip to the manuals...
•
programmers also have "header files" which contain the definition that they use for calling system functions. Using
the same type of header files in RPG makes the API more consistent across languages, and makes the IBM
reference manuals more applicable to our situation.
•
In fact, in each of the IBM manual pages, we’ll see lines that say ’#include <sys/socket.h>’ or ’#include
<errno.h>’ and these are telling the C programmer which header files are necessary for a given API call.
When I create an "API header file", I generally give the member a descriptive name followed by a "_H", which
means "header". So my header member relating to sockets is called "SOCKET_H".
Here is an example of what a prototype generally looks like in my header member:
**---------------------------------------------------------------** struct servent *getservbyname(char *service_name,
**
char *protocol_name);
**
**
Retrieve’s a service entry structure for a given service
**
and protocol name. (Usually used to get a port number for
**
a named service)
**
**
service_name = name of service to get (e.g.: ’http’ or ’telnet’)
** protocol_name = protocol that service runs under (’tcp’, ’udp’)
**
** Returns *NULL upon failure, otherwise a pointer to the service
**
entry structure
**---------------------------------------------------------------D getservbyname
PR
*
ExtProc(’getservbyname’)
D service_name
*
value options(*string)
27
Chapter 4. Improving our client program
D
protocol_name
*
value options(*string)
So you can see that it tells me a lot of information. How the prototype was defined in C, a basic description of the
function that I’m calling, a description of the parameters, and a description of the return values.
When I need constants in my header file, I’ll usually group them according to the API that uses them, and give each a
short description, like this:
* Address Family of the Internet Protocol
D AF_INET
C
CONST(2)
* Socket type for reading/writing a stream of data
D SOCK_STREAM
C
CONST(1)
* Default values of Internet Protocol
D IPPROTO_IP
C
CONST(0)
When I place data structures into a header file, I’ll do it like this:
**
** Service Database Entry (which service = which port, etc)
**
**
struct servent {
**
char
*s_name;
**
char
**s_aliases;
**
int
s_port;
**
char
*s_proto;
**
};
**
D p_servent
S
*
D servent
DS
Based(p_servent)
D
s_name
*
D
s_aliases
*
D
s_port
10I 0
D
s_proto
*
I’m planning (when time permits) to go through each of my data structures and add a description of each field in the
structure, as well. I simply haven’t had a chance yet.
You’ll also notice that I always declare my data structures (or at least, those I use with the UNIX-type APIs) to be
BASED() on a pointer. I do that for three reasons:
1. many of these structures (in particular, servent and hostent) are "returned" from an API. That means that the API
actually allocates the memory for them, rather than my program. So, I need to use a pointer so I can refer to the
memory that was allocated by that API.
2. As alluded to in point #1, when something is BASED() the system doesn’t allocate memory to it. Since the
header will usually have many different structures that may not be used in every program, it saves memory to
only allocate it when needed.
3. In all but the simplest programs, we will need more than one ’instance’ of each data structure. For example, if
we wanted to both call bind() to bind our socket to a given port number, and call connect() to connect to a
28
Chapter 4. Improving our client program
remote port in the same program, we’ll need two copies of the sockaddr_in data structure. One containing the
address & port number to bind to, the other containing the address & port number to connect to.
By using BASED() data structures we can do that, simply by allocating two different buffers, one called
"connto" and one called "bindto". We can set the address of the sockaddr_in data structure to the area of memory
that bindto is in when we want to change or examine the contents of the bindto data, and likewise set the address
of sockaddr_in to the area of memory that connto is in when we want to examine or change the connto area.
Point #3 is often difficult for an RPG programmer to understand. You can think of it as being similar in concept to
using the MOVE op-code to put data from a character string into a data structure to break the data into fields. It’s a
little different, however, in that you’re not copying all of the data from one place to another, instead you’re
referencing it in it’s original place in memory. Consequently, you don’t have to move it back after you’ve changed it.
Here’s a quick example of code snippets that use the functionality that I’m trying to describe in Point #3:
D connto
D bindto
D length
S
S
S
*
*
10I 0
C* How much memory do we need for a sockaddr_in structure?
C*
c
eval
length = %size(sockaddr_in)
C* Allocate space for a ’connect to’ copy of the structure:
C*
c
alloc(e) length
connto
c
if
%error
C**** No memory left?!
c
endif
C* Allocate space for a ’bind to’ copy of the structure:
C*
c
alloc(e) length
bindto
c
if
%error
C**** No memory left?!
c
endif
C* Set the values of the connto structure:
c
eval
p_sockaddr =
c
eval
sin_family =
c
eval
sin_addr
=
c
eval
sin_port
=
c
eval
sin_zero
=
connto
AF_INET
some_ip_address
some_port_number
*ALLx’00’
C* Set the values of the bindto structure:
C* Note that each time we do a ’p_sockaddr = XXX’ we are effectively
C* "switching" which copy of the sockaddr_in structure we’re working
C* with.
c
eval
p_sockaddr = bindto
c
eval
sin_family = AF_INET
c
eval
sin_addr
= INADDR_ANY
c
eval
sin_port
= some_port_number
29
Chapter 4. Improving our client program
c
C* call connect()
c
c* Error!
c
C* call bind()
c
c* Error!
c
eval
sin_zero
= *ALLx’00’
if
connect(sock1: connto: length) < 0
endif
if
bind(sock2: bindto: length) < 0
endif
C* examine the contents of connect() version
c
eval
p_sockaddr = connto
c
if
sin_port = 80
c* Do somethign special
c
endif
Hopefully, by now, you have an idea of how our /copy member (i.e. "API header member") should work. You should
be able to go back to the example programs in previous topics in this tutorial and create your own header file.
If you prefer, I will make available the header file that I use for my own sockets programs. This will save you having
to create your own, if you do not wish to. You can find my header file here:
http://www.scottklement.com/rpg/socktut/qrpglesrc.socket_h
After this point, all of the example programs will use my socket_h header member. Each section that explains a new
API call will describe what entries you need to add to your own header, if you’re writing your own.
4.2. improvement #2: error handling
The next improvement that our client program certainly needs is the ability to give more specific and meaningful
error messages when an API call does not work. This section explains how error handling works in the UNIX-type
APIs and explains how to use this in your RPG programs.
In C, there is a global variable called ’errno’ that is set to an error number whenever a procedure has an error. Almost
all C procedures return either an integer or a pointer, therefore, by convention, when an error occurs a C procedure
will return either ’-1’ or ’NULL’. The programmer will use this as an indication that he needs to check the value of
the ’errno’ global variable to see which error occurred. (Although most C procedures follow this convention, some
don’t. Check the manual page for the particular procedure before handling the error)
The ’errno’ variable is an integer. It can only contain a number, which is usually different on different platforms. To
make things more compatible across platforms, constants are placed into a C ’header file’, and programmers are
encouraged to use constants when checking for errors.
As an example, lets go back to the IBM manual page for the socket() API. For convienience, it can be found here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/socket.htm
Page down to the section entitled ’Error Conditions’, it states: When socket() fails, errno can be set to
one of the following:
30
Chapter 4. Improving our client program
The proceeds to list things like "EACCES", "EAFNOSUPPORT", "EIO", "EMFILE", along with descriptions of
what errors they refer to. Each one of these is actually a constant defined in in a C header file that is translated by the
compiler into a number.
In fact, if you have the System Openness Includes licensed program installed, you can bring up the ERRNO member
of the file called SYS in library QSYSINC and you’ll be able to see how this is done in C programs. EACCESS
corresponds to the number 3401, EMFILE to 3452, etc. Unfortunately, some of these are not specifically
"UNIX-Type API" errors, but are generic C language errors. So, for those values, you also need to look at the
member called ERRNO in the file H in library QCLE. This file is part of the ILE C/400 licensed program, and is only
available if you have that licpgm installed.
If you are fortunate enough to have the file ’H’ in ’QCLE’ installed on your machine, you’ll also see the following
definition the end:
#define errno (*__errno())
This is a macro that runs at compile time that changes the word ’errno’ wherever it’s used in a C program, to actually
be ’(*__errno())’. This is very useful to us because it means that whenever a C program refers to the global variable
’errno’, it’s actually referring to the return value from a procedure called ’__errno()’. That means, if we want to
retrieve errno in an RPG program, all we have to do is call that __errno() function ourselves!
For your convienience (and, for mine!) I’ve defined all of the C errno values from both QCLE and QSYSINC into a
single RPG /copy member. I also used some RPG compiler directives to include an errno procedure that will make it
much more intuitive to use errno in our RPG programs.
You can get a copy of this /copy member (called errno_h) at this link:
http://www.scottklement.com/rpg/socktut/qrpglesrc.errno_h
At first glance, you might be wondering "Why didn’t you simply make this a part of the existing socket_h member?"
The reason is quite simple: These error routines are used for all UNIX-Type APIs, not just sockets. And, in fact,
these routines are also used when calling any function from the ILE C runtime libraries. Consequently, there are
many times when we’ll want to use the ERRNO_H header member when we aren’t working with sockets at all.
To use this, you need to /copy the member twice. Once where you need the definitions of the constants (i.e., in your
D-specs), and once where you need the errno procedure (i.e. at the end of the program).
Here’s a quick code snippet:
H BNDDIR(’QC2LE’)
D/copy qrpglesrc,socket_h
D/copy qrpglesrc,errno_h
.... more code is found here ....
c
c
c
c
c
if
socket(AF_INET:SOCK_STREAM:IPPROTO_IP)<0
select
when
eval
errno = EACCES
Msg = ’Permission Denied’
c
c
when
eval
errno = EAFNOSUPPORT
Msg = ’Address Family not supported!’
31
Chapter 4. Improving our client program
c
c
when
goto
errno = EIO
TryAgainMaybe
... etc, etc, etc ...
c
endsl
.... more code is found here ....
/define ERRNO_LOAD_PROCEDURE
/copy libsor/qrpglesrc,errno_h
Note that because ’__errno()’ is part of the C-runtime library (as opposed to being part of the UNIX-Type APIs) we
need to include the QC2LE binding directory when compiling our program. Otherwise, it won’t find the API.
You see that in this example, the program is trying again (maybe) when it gets an I/O error. For the other errors, it’s
simply assigning a human-readable error message that could later be used to tell the user and/or programmer what
the problem was.
As you can imagine, it can get quite tedious to try to have the program handle every possible error that can occur.
Especially when, in most cases, all we do is assign the human-readble error message to a variable. Fortunately, there
is already a procedure we can call from the C runtime library that will return a human-readable error message for any
errno value.
This procedure is called ’strerror’ ("string error") and like all C procedures that return strings, it returns a pointer to a
an area of memory containing a string. The string is variable length, and we know when we’ve reached the end of the
string because we encounter a ’NULL’ character (x’00’). As we’ve done before, we can use the RPG
built-in-function "%str" to decode this string for us.
So, to call strerror from an RPG program (the prototype is already defined in that /copy member I mentioned earlier)
we can simply do something like this:
H BNDDIR(’QC2LE’)
D/copy qrpglesrc,socket_h
D/copy qrpglesrc,errno_h
.... more code is found here ....
c
c
c
c
c
c
c
c
c
c
c
if
socket(AF_INET:SOCK_STREAM:IPPROTO_IP)<0
if
goto
else
eval
endif
errno = EIO
TryAgainMaybe
Msg = %str(strerror(errno))
dsply
Msg
endif
.... more code is found here ....
32
Chapter 4. Improving our client program
/define ERRNO_LOAD_PROCEDURE
/copy libsor/qrpglesrc,errno_h
This will return the human readable error message for everything except for the ’EIO’ error. For EIO, it’ll go back
and ’Try Again Maybe’.
4.3. improvement #3: Creating a Read Line utility
Now, we’ve got two "API header members", one for socket calls, and one for error handling, this will allow us to
greatly simplify our sockets programs, making them much easier for other programmers to read.
However, looking at the code, there’s still one area that’s a bit difficult to read, and that’s the process of reading one
"line" at a time from a socket.
You may recall that almost all almost all of the communications that we’ll have with internet servers will involve
sending and receiving "lines ASCII of text." In fact, this is SO common that I couldn’t come up with a simple client
program that didn’t use them!
Since most RPG programmers are used to thinking of data in terms of "records" and "fields", it might be useful to
think of a line of ASCII text as being a "record". These records are, however, almost always variable-length. You
determine where the end of the record is by looking for an ’end of line sequence’, which consists of one or two
control characters.
There are 3 popular end of line sequences. The first one is that each line ends with the CR (Carriage Return)
character. This is most commonly used in Apple computers. It is based on the way a typewriter works, when you
press the ’Carriage Return’ key on a typewriter, the paper feeds up by one line, and then returns to the leftmost
position on the page. CR is ASCII value 13 (or Hex x’0D’)
The next end of line sequence is the one used by Unix. The ASCII standard states that ends of lines of text should
end with the LF (’line feed’) character. Many people, especially C programmers, will refer to the LF character as the
’Newline’ character. LF is ASCII value 10 (or Hex x’0A’)
Finally, we have the end of line sequence used by DOS and Windows. As noted above, a typewriter feeds to the next
line when CR is pressed. However, most early printers could "overtype" a line of text, so when you send the CR
character it would go back to the start of the SAME line. The LF sequence would feed the paper out one line, but
would not return the carriage to the leftmost position. So, to start a new line of text required two characters, a CR
followed by an LF. DOS (and eventually Windows) kept this standard by requiring a line of text to end with both the
CR character, followed by the LF character. This was a nice compromise between the up & coming UNIX operating
system that DOS was inspired by, and the Apple II DOS, which was at that time the most popular home computer in
the world.
So enough history, already! The point of all this is, there are many different ways to send and receive lines of text,
what we need is a way to read them in an RPG program. In some cases, we’ll want to be able to specify different
end-of-line sequences, and in some cases we’ll want to have the text automatically translated from ASCII to
EBCDIC for us to work with.
So, lets write a routine to do this. We’ll make it a subprocedure and put it into a service program, since we’ll want to
do this in virtually every sockets program that we write.
33
Chapter 4. Improving our client program
To make this as much like the recv() API as possible, we’ll allow the caller to pass us a socket, a pointer, and a max
length as our first 3 parameters. We’ll also return the length of the data read, just as recv() does, or -1 upon error. So,
the prototype for this new routine will look like this (so far):
D RdLine
D
peSock
D
peLine
D
peLength
PR
10I 0
10I 0 value
*
value
10I 0 value
(Note that in the naming conventions of the company that I work for, ’pe’ at the start of a variable means ’parameter’)
Next, we’ll give optional parameters that can be used to tell our subprocedure to translate the data to EBCDIC,
and/or change what the characters for line feed and newline are. This will make our prototype look more like this:
D RdLine
D
peSock
D
peLine
D
peLength
D
peXLate
D
peLF
D
peCR
PR
10I 0
10I 0 value
*
value
10I 0 value
1A
const options(*nopass)
1A
const options(*nopass)
1A
const options(*nopass)
Now, if the optional parameters are not passed to us, we need to use reasonable default values. That is, we’ll use
x’0D’ for CR, x’0A’ for LF and *OFF (do not translate) for the peXLate parm. This is easy to do by checking how
many parms were passed. Like so:
c
c
c
c
c
if
eval
else
eval
endif
%parms > 3
wwXLate = peXLate
c
c
c
c
c
if
eval
else
eval
endif
%parms > 4
wwLF = peLF
c
c
c
c
c
if
eval
else
eval
endif
%parms > 5
wwCR = peCR
wwXLate = *OFF
wwLF = x’0A’
wwCR = x’0D’
(Note that in the naming conventions of the company that I work for, ’ww’ at the start of a variable means ’work field
that is local to a subprocedure’)
Then, just like in the DsplyLine routine that we used in our example client, we’ll read one character at a time until
we receive a LF character. To make the routine simple, we’ll simply drop any CR characters that we come across.
(That way, if LF=x’0A’ and CR=x’0D’, this routine will read using either UNIX or Windows end of line characters
34
Chapter 4. Improving our client program
without any problems. If you wanted to make it work with Apple or Windows end of lines, you’d simply reverse the
values passed as CR and LF)
Finally, if the XLate parameter is set to ON, we’ll translate the line to EBCDIC. If not, we’ll leave it as it was
originally sent. So, the finished routine will look like this:
(This should go in member SOCKUTILR4 in the file QRPGLESRC)
P RdLine
D RdLine
D
peSock
D
peLine
D
peLength
D
peXLate
D
peLF
D
peCR
B
PI
D
D
D
D
D
D
D
S
S
S
S
S
S
S
wwBuf
wwLen
RC
CH
wwXLate
wwLF
wwCR
Export ➊
10I 0
10I 0 value
*
value
10I 0 value
1A
const options(*nopass)
1A
const options(*nopass)
1A
const options(*nopass)
32766A
based(peLine)
10I 0
10I 0
1A
1A
1A
1A
➋
c
c
c
c
c
** Set default values to unpassed parms:
if
%parms > 3
eval
wwXLate = peXLate
else
eval
wwXLate = *OFF
endif
c
c
c
c
c
if
eval
else
eval
endif
%parms > 4
wwLF = peLF
c
c
c
c
c
if
eval
else
eval
endif
%parms > 5
wwCR = peCR
c
** Clear "line" of data:
eval
c
wwLF = x’0A’
wwCR = x’0D’
➌
%subst(wwBuf:1:peLength) = *blanks
dow
1 = 1
eval
if
rc = recv(peSock: %addr(ch): 1: 0)
rc < 1
** read 1 byte:
c
c
35
Chapter 4. Improving our client program
c
c
c
c
c
c
if
leave
else
return
endif
endif
wwLen > 0
-1
c
c
c
** if LF is found, we’re done reading:
if
ch = wwLF
leave
endif
c
c
c
c
** any other char besides CR gets added to the string:
if
ch <> wwCR
eval
wwLen = wwLen + 1
eval
%subst(wwBuf:wwLen:1) = ch
endif
c
c
c
** if variable is full, exit now -- there’s no space left to read data into
if
wwLen = peLength ➍
leave
endif
c
enddo
c
c
c
** if ASCII->EBCDIC translation is required, do it here
if
wwXLate=*ON and wwLen > 0
callp
Translate(wwLen: wwBuf: ’QTCPEBC’)
endif
c
P
** return the length
return
E
wwLen
Personally, I learn things better by typing them in, rather than reading them. Therefore, I recommend that you type
the code examples in this tutorial in yourself. However, if you’d like, you can download my copy of SOCKUTILR4
here: http://www.scottklement.com/rpg/socktut/qrpglesrc.sockutilr4
A few notes:
➊
the opening P-spec declares this procedure with "Export", so we can make it callable from outside of our service
program.
➋
The ’Translate’ prototype isn’t defined here. That’s because we’ll make it global to the entire service program.
➌
Before the recv() loop, we clear the buffer, but we only clear from the start to the ’length’ that was passed into
our program. This is very important, since if the calling program should send us a variable that’s not the entire
32k, we don’t want to clear data that might not be allocate to the variable...
➍
Although we let the caller pass a line length, as written, this procedure has an effective maximum line length of
32,766 chars per line. We COULD make it do an ’unlimited’ number by using pointer math, but this would
make it more difficult to clear the variable at the start, and at any rate, it’d be very unusual for a program to need
to read a line of text bigger than 32k.
36
Chapter 4. Improving our client program
4.4. improvement #4: Creating a Write Line utility
In the last section, we wrote a routine that would read lines of text into a variable. This would make a nice utility
routine that we can put into a service program.
In this section, we’ll add another utility routine, one that can be used to send lines of text.
The requirements of this routine are a little different, because it’s output-only. That means we don’t need to write any
data to a buffer, just take data that we already have and send it out. Since the routines that are calling this will often
want to pass literals, constants, and expressions, it makes sense for us to accept an ’const’ parameter, rather than a
pointer.
Also, it’d be a bit of a pain, especially when working with literals, for the calling routine to have to calculate the
length of the string each time it calls us. So, instead we’ll figure that out ourselves by looking for the last non-blank
character. We can still give an optional ’length’ parameter, just in case the caller wishes to override this.
This method of searching for the last ’non-blank’ character will be slow in most cases with a 32k buffer. Since 99%
of the time, a line is less than 80 bytes long, this means that it’d have to search over 32,000 blank spaces each time we
send a line. To make this a bit more realistic, the buffer on a call to the ’send line’ procedure will only be 256 bytes.
Like the ’read line’ routine, though, we still want to allow the caller the option of specifying different end-of-line
characters. In RdLine() if the caller only wants one character to signify end-of-line, they can still do so... just make
the extra parm be the same char, it won’t hurt anything. When sending, however, that would cause problems. So,
we’ll determine how many chars to add for end-of-line based on how many parms are passed, instead of trying to use
both like we did in RdLine().
I think that pretty much sums up the requirements. Here’s the routine I actually came up with, check it out:
(This should be added to the member SOCKUTILR4 in the file QRPGLESRC)
P WrLine
D WrLine
D peSock
D peLine
D peLength
D peXLate
D peEOL1
D peEOL2
B
PI
Export
10I 0
10I 0 value
256A
const
10I 0 value options(*nopass)
1A
const options(*nopass)
1A
const options(*nopass)
1A
const options(*nopass)
D
D
D
D
D
D
S
S
S
S
S
S
256A
10I 0
1A
2A
10I 0
10I 0
wwLine
wwLen
wwXlate
wwEOL
wwEOLlen
rc
C*******************************************
C* Allow this procedure to figure out the
C* length automatically if not passed,
C* or if -1 is passed.
C*******************************************
c
if
%parms > 2 and peLength <> -1
c
eval
wwLen = peLength
c
else
c
eval
wwLen = %len(%trim(peLine))
37
Chapter 4. Improving our client program
c
endif
C*******************************************
C* Default ’translate’ to *ON. Usually
C* you want to type the data to send
C* in EBCDIC, so this makes more sense:
C*******************************************
c
if
%parms > 3
c
eval
wwXLate = peXLate
c
else
c
eval
wwXLate = *On
c
endif
C*******************************************
C* End-Of-Line chars:
C*
1) If caller passed only one, set
C*
that one with length = 1
C*
2) If caller passed two, then use
C*
them both with length = 2
C*
3) If caller didn’t pass either,
C*
use both CR & LF with length = 2
C*******************************************
c
eval
wwEOL = *blanks
c
eval
wwEOLlen = 0
c
c
c
c
if
eval
eval
endif
%parms > 4
%subst(wwEOL:1:1) = peEOL1
wwEOLLen = 1
c
c
c
c
if
eval
eval
endif
%parms > 5
%subst(wwEOL:2:1) = peEOL2
wwEOLLen = 2
c
c
c
c
if
eval
eval
endif
wwEOLLen = 0
wwEOL = x’0D0A’
wwEOLLen = 2
C*******************************************
C* Do translation if required:
C*******************************************
c
eval
wwLine = peLine
c
if
wwXLate = *On and wwLen > 0
c
callp
Translate(wwLen: wwLine: ’QTCPASC’)
c
endif
C*******************************************
C* Send the data, followed by the end-of-line:
C* and return the length of data sent:
C*******************************************
c
if
wwLen > 0
38
Chapter 4. Improving our client program
c
c
c
c
c
eval
if
return
endif
endif
rc = send(peSock: %addr(wwLine): wwLen:0)
rc < wwLen
rc
c
c
c
c
eval
if
return
endif
rc = send(peSock:%addr(wwEOL):wwEOLLen:0)
rc < 0
rc
return
(rc + wwLen)
c
P
E
Personally, I learn things better by typing them in, rather than reading them. Therefore, I recommend that you type
the code examples in this tutorial in yours elf. However, if you’d like, you can download my copy of SOCKUTILR4
here: http://www.scottklement.com/rpg/socktut/qrpglesrc.sockutilr4
4.5. Packaging our utilities into a service program
Now that we’ve written our RdLine() and WrLine() procedures, a few other details should be ironed out:
1. We need a prototype for Translate() added to the top of the service program.
2. We need to include our socket_h header file in our service program.
3. We need binding language source to use when creating the service program.
I’ll assume at this point that you’ve already coded the RdLine() and WrLine() procedures from the previous chapters.
At the top of the member that you placed those subprocedures in, you’ll need the following code:
H NOMAIN
D/COPY SOCKTUT/QRPGLESRC,SOCKET_H
D/COPY SOCKTUT/QRPGLESRC,SOCKUTIL_H
D Translate
D
peLength
D
peBuffer
D
peTable
PR
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
You’ll also want to create a member that contains the prototypes for the RdLine and WrLine functions. We’ll call this
SOCKUTIL_H, and put that in QRPGLESRC as well. It will look like this:
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* RdLine(): This reads one "line" of text data from a socket.
*
*
peSock = socket to read from
*
peLine = a pointer to a variable to put the line of text into
*
peLength = max possible length of data to stuff into peLine
39
Chapter 4. Improving our client program
*
peXLate = (default: *OFF) Set to *ON to translate ASCII -> EBCDIC
*
peLF (default: x’0A’) = line feed character.
*
peCR (default: x’0D’) = carriage return character.
*
* returns length of data read, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
D RdLine
PR
10I 0
D
peSock
10I 0 value
D
peLine
*
value
D
peLength
10I 0 value
D
peXLate
1A
const options(*nopass)
D
peLF
1A
const options(*nopass)
D
peCR
1A
const options(*nopass)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* WrLine() -- Write a line of text to a socket:
*
*
peSock = socket descriptor to write to
*
peLine = line of text to write to
*
peLength = length of line to write (before adding CRLF)
*
you can pass -1 to have this routine calculate
*
the length for you (which is the default!)
*
peXlate = Pass ’*ON’ to have the routine translate
*
this data to ASCII (which is the default) or *OFF
*
to send it as-is.
*
peEOL1 = First character to send at end-of-line
*
(default is x’0D’)
*
peEOL2 = Second character to send at end-of-line
*
(default is x’0A’ if neither EOL1 or EOL2 is
*
passed, or to not send a second char is EOL1
*
is passed by itself)
*
* Returns length of data sent (including end of line chars)
*
returns a short count if it couldnt send everything
*
(if you’re using a non-blocking socket) or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
D WrLine
PR
10I 0
D peSock
10I 0 value
D peLine
256A
const
D peLength
10I 0 value options(*nopass)
D peXLate
1A
const options(*nopass)
D peEOL1
1A
const options(*nopass)
D peEOL2
1A
const options(*nopass)
If you’d prefer to download my copy of SOCKUTIL_H instead of writing your own, you can get it here:
http://www.scottklement.com/rpg/socktut/qrpglesrc.sockutil_h
Finally, we need to create binding source to tell the system how to create our service program. If you’re not familiar
with this, I’ll explain it a little.
40
Chapter 4. Improving our client program
When creating a service program, the system calculates a ’signature’ for your service program. This works in a
similar manner to the way ’record format level checks’ work on a database file -- if something in the service program
changes, they prevent you from possibly accessing it incorrectly.
Binding language serves to tell the system which procedures are exported, and also to tell the system which
procedures were exported LAST time you ran CRTSRVPGM, and maybe the time before that, and so on. If you
don’t specify binding language for your service program, each time a procedure changes in it, you’ll have to
recompile ALL of the programs that call that service program. If you don’t, you’ll get "Signature Violation" errors
when trying to call them! But, thanks to the ability to keep track of ’previous exports’ in a service program, you can
make your service program backwards compatible.
I hope that’s clear enough -- if not, you probably want to read the ’ILE Concepts’ manual, as this tutorial really isn’t
intended to teach all of the nuances of service programs and signatures.
At any rate, it’s very easy to create binding source. It looks like this:
Member SOCKUTILR4 of file QSRVSRC:
STRPGMEXP PGMLVL(*CURRENT)
EXPORT SYMBOL(RDLINE)
EXPORT SYMBOL(WRLINE)
ENDPGMEXP
That’s it... it’s that simple... we’ve told that this is the program exports for the current level of our service program,
and which procedures to export. It can hardly be easier!
Now, we compile our service program by typing:
CRTRPGMOD MODULE(SOCKUTILR4) SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
CRTSRVPGM SRVPGM(SOCKUTILR4) EXPORT(*SRCFILE) SRCFILE(SOCKTUT/QSRVSRC)
And, while we’re at it, lets create a binding directory to make our service program available to the CRTBNDRPG
command. We do this by typing:
CRTBNDDIR BNDDIR(SOCKTUT/SOCKUTIL) TEXT(’Socket Utility Binding Directory’)
ADDBNDDIRE BNDDIR(SOCKTUT/SOCKUTIL) OBJ((SOCKTUT/SOCKUTILR4 *SRVPGM))
4.6. improvement #5: Stripping out the HTTP response
One of the things that this tutorial has completely glossed over so far is the HTTP protocol itself. This section
attempts to answer the following questions about the HTTP protocol:
•
How do we know that we need to send ’GET ’ followed by a filename, followed by ’HTTP/1.0’ to request a file?
•
Why do we send an extra x’0D0A’ after our request string?
•
Why is there extra data appearing before the actual page that I tried to retrieve?
First, a bit of background:
If you listen to the hype and the buzzwords circulating through the media, you might get the idea that ’The Internet’
and ’The World Wide Web’ are synonyms. In fact, they really aren’t.
41
Chapter 4. Improving our client program
The Internet is the large, publicly-accessible, network of computers that are connected together using the TCP/IP
protocol suite. The Internet is nothing more or less than a very large network of computers. All it does is allow
programs to talk to other programs.
The World Wide Web is one application that can be run over The Internet. The web involves hypertext documents
being transported from a web server to a PC where they can be viewed in a web browser. So, you have two programs
communicating with each other. A client program ("the web browser") and a server program ("the web server").
The client and server programs use the TCP protocol to connect up with each other. Once they are connected, the
client will need to know what data actually needs to be sent to the server in order to get back the document that it
wants to display. Likewise, the web server has to understand the client’s requests, and has to know what an
acceptable response will be. Clearly, a common set of "command phrases" and "responses" must be understood by
both the client and the server in order for them to get meaningful work done. This set of phrases and responses is
called an ’application-level protocol.’ In other words, it’s the protocol that web applications use to talk to each other.
The web’s protocol is offically called ’HyperText Transfer Protocol’ (or HTTP for short) and it is an official Internet
standardized protocol, developed by the Internet Engineering Task Force. The exact semantics of HTTP are
described in a document referred to as a ’Request For Comments’ document, or RFC for short.
The web is, of course, only one of thousands of applications that can utilize the Internet. All of these have their own
Application Level Protocol. Most of these are also based on an official internet standard described as an RFC. Some
examples of these other protocols include: File Transfer Protocol (FTP), Telnet Protocol, Simple Mail Transport
Protocol (SMTP), Post Office Protocol (POP), etc.
All of the RFCs that describe Internet Protocols are available to the public at this site: http://www.rfc-editor.org
(Whew!) Back to the questions, then:
Q: "How do we know that we need to send ’GET ’ followed by a filename, followed by ’HTTP/1.0’ to request a file?"
A: We know how the HTTP protocol works by looking at RFC 2616. There are many, many exact details that should
be taken into account to write a good HTTP client, so if you’re looking to write a professional quality application,
you should make sure that you’ve read RFC 2616 very carefully.
Q: Why do we send an extra x’0D0A’ after our request string?
A: RFC 2616 refers to a ’request chain’, and a ’response chain’. Which involves the client sending a request (in our
case a GET request) followed by zero or more pieces of useful information for the web server to use when delivering
the document. The request chain ends when a blank line is send to the server. Since lines of ASCII text always end
with x’0D’ (CR) followed by x’0A’ (LF), it looks like a blank line to the web server, and terminates our ’request
chain.’
Q: Why is there extra data appearing before the actual page that I tried to retrieve?
A: The server replies with a ’response chain’. Like the ’request chain’ it contains a list of one or more responses,
terminated with a blank line. However, in our simple client program, we were not trying to interpret these responses,
but merely displaying them to the user.
Finally! Let’s make an improvement!
Completely implementing the HTTP protocol in our sample client would be a bit too much for this tutorial. After all,
this is a tutorial on socket programming, not HTTP! However, a simple routine to strip the HTTP responses from the
data that we display should be simple enough, so lets do that.
Instead of the ’DsplyLine’ subroutine in our original client, we’ll utilize our nifty new ’rdline’ procedure. We’ll read
back responses from the server until one of them is a blank line. Like this:
42
Chapter 4. Improving our client program
c
C
c
c
C
c
c
c
c
c
dou
eval
if
eval
dsply
callp
return
endif
enddo
recbuf = *blanks
rc = rdline(sock: %addr(recbuf):
%size(recbuf): *On)
rc < 0
msg = %str(strerror(errno))
msg
close(sock)
And then, we will receive the actual data, which ends when the server disconnects us. We’ll display each line like
this:
c
c
c
c
c
c
c
dou
eval
if
dsply
endif
enddo
recbuf
rc < 0
rc = rdline(sock: %addr(recbuf):
%size(recbuf): *On)
rc >= 0
The result should be the web page, without the extra HTTP responses at the start.
4.7. improvement #6: Sending back escape messages
This time, we’re going to improve the way error messages are returned to the calling program or user.
Currently, we’re sending back responses using the DSPLY op-code. Yuck. Blech. Ewww.
Instead, error messages should be sent back to the calling program using a typical AS/400 *ESCAPE message! This
way, the program that called us will know that we ended abnormally, and can even monitor for a message to handle.
The most convienient way to implement this is to create a subprocedure that gives us a simple interface to the
QMHSNDPGM (Send Program Message) API. The subprocedure will fill in all of the details that the API needs,
except the actual message data -- we’ll pass the message data as a parm.
The result is a subprocedure that looks like this:
P die
D die
D
peMsg
B
PI
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
PR
256A
const
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
43
Chapter 4. Improving our client program
D
D
MessageKey
ErrorCode
4A
32766A
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
1
5
9
16
17
options(*varsize)
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
Now, instead of code that looks like this:
c
c
c
c
c
c
c
eval
if
eval
dsply
return
endif
sock = socket(AF_INET: SOCK_STREAM:
IPPROTO_IP)
sock < 0
msg = %str(strerror(errno))
msg
We’ll make our error handling look more like this:
c
c
c
c
c
c
eval
if
callp
return
endif
sock = socket(AF_INET: SOCK_STREAM:
IPPROTO_IP)
sock < 0
die(’socket(): ’ + %str(strerror(errno)))
Cute, huh? The die() procedure will cause it to send back an escape message when the socket() API fails. The format
of the parms in the die() subprocedure makes it very easy to insert into our code.
Having said that, however, here’s a slightly more complicated usage:
C
c
if
eval
connect(sock: p_connto: addrlen) < 0
err = errno
44
Chapter 4. Improving our client program
c
c
c
c
callp
callp
return
endif
close(sock)
die(’connect(): ’+%str(strerror(err)))
The close() API is a UNIX-type API and can return an error via errno just like the connect() API can! Therefore, we
save the value of errno before calling close() just to make sure that we don’t lose the value.
You should also note that the ’die’ procedure will actually end the program. Unless something goes wrong with the
’die’ procedure, the ’return’ statement will never be executed. This means that if you have something to close, such
as a socket, you need to make sure you do it before calling ’die’.
I’ll now go through and replace DSPLY with die() throughout my program.
4.8. improvement #7: Displaying the data using DSPF
I am writing a tutorial on how to use stream files from RPG, but as of the time that I’m writing this page, I haven’t yet
finished it. Even so, though, I don’t want to take away from the socket tutorial to try to teach stream files, so... I’ll just
assume that you either already know how stream files work, or that you’ve been able to find a tutorial somewhere.
Since data is being returned from the socket in a stream format, and we want to write it to our stream file in a stream
format, it’s very easy to write the web page data to a stream file! In fact, we won’t even bother breaking the data that
we receive into ’lines of text’, but rather just dump it all (as-is) into the stream file.
When the DSPF command is run, it’ll do the job of breaking things up into lines of text.
So, first we will open a stream file to contain the data we receive:
c
c
c
c
c
c
c
c
c
eval
if
eval
callp
callp
return
endif
fd = open(’/http_tempfile.txt’:
O_WRONLY+O_TRUNC+O_CREAT+O_CODEPAGE:
511: 437)
fd < 0
err = errno
close(sock)
Die(’open(): ’+%str(strerror(err)))
Next, we’ll write whatever we receive from the socket into the stream file, without changing a thing:
c
c
c
c
c
c
c
dou
eval
if
callp
endif
enddo
rc < 1
rc = recv(sock: %addr(recbuf):
%size(recbuf): 0)
rc > 0
write(fd: %addr(recbuf): rc)
After we’ve closed the file and the socket, we’ll display the file using IBM’s DSPF command:
45
Chapter 4. Improving our client program
c
c
c
c
callp
callp
callp
close(fd)
close(sock)
Cmd(’DSPF STMF(”/http_tempfile.txt”)’:
200)
Finally, we’ll delete our temporary stream file using the ’unlink’ API call:
c
callp
unlink(’/http_tempfile.txt’)
WooHoo! We’ve gotten rid of all of the nasty DSPLY opcodes, now!
4.9. Our updated client program
The last several topics have added a lot of improvements to our client program. The result will be something that you
could even adapt to your own needs if you wanted an AS/400 program that would fetch web documents.
Here’s the new improved sample http client:
File: SOCKTUT/QRPGLESRC, Member: CLIENTEX2
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’QC2LE’) BNDDIR(’SOCKTUT/SOCKUTIL’)
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,errno_h
D/copy socktut/qrpglesrc,sockutil_h
*********************************************************
* Definitions needed to make IFS API calls. Note that
* these should really be in a separate /copy file!
*********************************************************
D O_WRONLY
C
2
D O_CREAT
C
8
D O_TRUNC
C
64
D O_CODEPAGE
C
8388608
D open
PR
10I 0 ExtProc(’open’)
D filename
*
value options(*string)
D openflags
10I 0 value
D mode
10U 0 value options(*nopass)
D codepage
10U 0 value options(*nopass)
D unlink
PR
10I 0 ExtProc(’unlink’)
D
path
*
Value options(*string)
D write
PR
10I 0 ExtProc(’write’)
D handle
10I 0 value
D buffer
*
value
D bytes
10U 0 value
*********************************************************
* end of IFS API call definitions
*********************************************************
46
Chapter 4. Improving our client program
D die
D
peMsg
PR
D cmd
D
command
D
length
PR
D
D
D
D
D
D
D
D
D
D
D
D
D
D
S
S
S
S
S
s
s
s
S
S
S
S
S
S
msg
sock
port
addrlen
ch
host
file
addr
p_Connto
RC
RecBuf
RecLen
err
fd
256A
const
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
50A
10I
5U
10I
1A
32A
32A
10U
*
10I
50A
10I
10I
10I
0
0
0
0
0
0
0
0
C*************************************************
C* The user will supply a hostname and file
C* name as parameters to our program...
C*************************************************
c
*entry
plist
c
parm
host
c
parm
file
c
eval
*inlr = *on
C*************************************************
C* what port is the http service located on?
C*************************************************
c
eval
p_servent = getservbyname(’http’:’tcp’)
c
if
p_servent = *NULL
c
callp
die(’Can”t find the HTTP service!’)
c
return
c
endif
c
eval
port = s_port
C*************************************************
C* Get the 32-bit network IP address for the host
C* that was supplied by the user:
C*************************************************
c
eval
addr = inet_addr(%trim(host))
c
if
addr = INADDR_NONE
c
eval
p_hostent = gethostbyname(%trim(host))
c
if
p_hostent = *NULL
c
callp
die(’Unable to find that host!’)
c
return
47
Chapter 4. Improving our client program
c
c
c
endif
eval
endif
addr = h_addr
C*************************************************
C* Create a socket
C*************************************************
c
eval
sock = socket(AF_INET: SOCK_STREAM:
c
IPPROTO_IP)
c
if
sock < 0
c
callp
die(’socket(): ’ + %str(strerror(errno)))
c
return
c
endif
C*************************************************
C* Create a socket address structure that
C*
describes the host & port we wanted to
C*
connect to
C*************************************************
c
eval
addrlen = %size(sockaddr)
c
alloc
addrlen
p_connto
c
c
c
c
c
eval
eval
eval
eval
eval
p_sockaddr
sin_family
sin_addr =
sin_port =
sin_zero =
= p_connto
= AF_INET
addr
port
*ALLx’00’
C*************************************************
C* Connect to the requested host
C*************************************************
C
if
connect(sock: p_connto: addrlen) < 0
c
eval
err = errno
c
callp
close(sock)
c
callp
die(’connect(): ’+%str(strerror(err)))
c
return
c
endif
C*************************************************
C* Send a request for the file that we’d like
C* the http server to send us.
C*
C* Then we send a blank line to tell it we’re
C* done sending requests, it can process them...
C*************************************************
c
callp
WrLine(sock: ’GET http://’ +
c
%trim(host) + %trim(file) +
c
’ HTTP/1.0’)
c
callp
WrLine(sock: ’ ’)
C*************************************************
C* Get back the server’s response codes
C*
48
Chapter 4. Improving our client program
C* The HTTP server will send it’s responses one
C* by one, then send a blank line to separate
C* the server responses from the actual data.
C*************************************************
c
dou
recbuf = *blanks
C
eval
rc = rdline(sock: %addr(recbuf):
c
%size(recbuf): *On)
c
if
rc < 0
c
eval
err = errno
c
callp
close(sock)
c
callp
die(’rdline(): ’+%str(strerror(err)))
c
return
c
endif
c
enddo
C*************************************************
C* Open a temporary stream file to put our
C*
web page data into:
C*************************************************
c
eval
fd = open(’/http_tempfile.txt’:
c
O_WRONLY+O_TRUNC+O_CREAT+O_CODEPAGE:
c
511: 437)
c
if
fd < 0
c
eval
err = errno
c
callp
close(sock)
c
callp
Die(’open(): ’+%str(strerror(err)))
c
return
c
endif
C*************************************************
C* Write returned data to the stream file:
C*************************************************
c
dou
rc < 1
c
eval
rc = recv(sock: %addr(recbuf):
c
%size(recbuf): 0)
c
if
rc > 0
c
callp
write(fd: %addr(recbuf): rc)
c
endif
c
enddo
C*************************************************
C* We’re done receiving, do the following:
C*
1) close the stream file & socket.
C*
2) display the stream file
C*
3) unlink (delete) the stream file
C*
4) end program
C*************************************************
c
callp
close(fd)
c
callp
close(sock)
c
callp
Cmd(’DSPF STMF(”/http_tempfile.txt”)’:
c
200)
c
callp
unlink(’/http_tempfile.txt’)
49
Chapter 4. Improving our client program
c
return
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
50
Chapter 5. Creating server programs
Written by Scott Klement.
5.1. Introduction to server programs
The last few chapters have explained how to write a program that acts as a client to an existing server program. Now,
we will begin exploring the other side of the connection, the server-side.
Usually the role of a client program is to initiate the network connection, to request services from the server, and then
to terminate the connection. Most of the time, a client program becomes active when it is run by a user, and is
deactivated when it is done processing the data it received from the server, or when the user tells it to deactivate.
In contrast, a server program usually stays active at all times. It does not initiate any activity on its own, but rather
waits for connections from client programs. When a client program has connected, it waits for the client program to
make requests. When the server program realizes that the client program is done, it waits for the next connection.
There are minor differences in which API calls you need to make when writing a server program versus writing a
client program. You still need to call socket() to create a socket to use for communications, but instead of issuing a
connect() call, you will listen() for connections, and then accept() the connections as they come in.
It’s also important to understand that a client program has to know where the server program is listening for
connections. (pause to let that sink in) In other words, when a client issues a connect(), it has to tell the API which
address & port to connect to -- consequently, we have to make sure that the server is, in fact, listening on the port &
address that the client expects it to be listening on. This is called ’binding’ to a port.
In addition to the differences in which API is called, there are a few other considerations that make server programs
different from client programs:
•
Server programs usually have to be capable of handling many client connections at once. Generally speaking, this
isn’t true of clients. This creates some significant hurdles to overcome when trying to write code.
•
Because a server is available for connections at all times, the security of a server-side program is usually a much
bigger concern. You have to be security conscious when writing a server program.
•
For the same reasons, server programs frequently need to validate a user-id and password of the person connecting,
and be careful to only give them access to what the security officer has deemed that a user should be able to access.
•
Server programs usually aren’t interactive applications. They have no screen associated with them, besides the one
that might be found on the client side. Therefore, they tend to be a little trickier to debug.
The basic model for a server program looks something like this:
1.
Call the getservbyname() API to find out the port number for the service you want to be a server for.
2.
Call the socket() API to create a new socket.
3.
Call the bind() API to bind the socket to the port that we found in step #1.
4.
Call the listen() API to tell the system that you want to listen for connections with this socket. (This "opens up"
the port so that people can connect to it)
5.
Call the accept() API. The accept() API will wait until a client connects to the port, and then will create a new
socket. The new socket will already be connected to the client.
51
Chapter 5. Creating server programs
6.
Here we process the newly connected client. Depending on whether we’re only handling one connection at a
time, or what model we’re using to handle many connections, we’ll do things very differently at this point.
7.
Close the socket you got from accept()
8.
Go back to step 5.
5.2. The bind() API call
Now that we’ve discussed the basic operation of a server program, let’s start getting into specifics. You already know
how to call getservbyname(), and socket(), so the first thing we need to know is bind().
The IBM manual page for the bind() API call is located here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/bind.htm
It tells us that the C language prototype for bind looks like this:
int bind(int socket_descriptor,
struct sockaddr *local_address,
int address_length);
The procedure is called ’bind’. It takes 3 parameters, an integer, a pointer to a sockaddr structure, and another
integer. The API also returns an integer.
So, we’ll add the following prototype to our SOCKET_H /copy member:
D bind
D
socket
D
local_addr
D
addresslen
PR
10I 0 ExtProc(’bind’)
10I 0 Value
*
Value
10I 0 Value
In fact, bind() and connect() have the same exact prototype, (other than the procedure name.) So what’s the
difference between them?
Short answer: bind() specifies the address & port on the local side of the connection. connect() specifies the address
& port of the remote side of the connection.
Long answer:
Each time an IP datagram containing TCP data is sent over the network, the datagram contains a ’local address’,
’remote address’, ’local port’, and ’remote port’. This is the only information that IP has to figure out who ends up
getting the packet.
So, both the client and the server port numbers need to be filled in before the connection can work. Data that is
directed to the server needs a ’destination’ port, so that the data can get sent to the appropriate program running on
the server. Likewise, it needs a ’source’ so that the server knows who to send data back to, and also so that if there are
many connections from the same computer, the server can keep them separate by looking at the source port number.
Since the connection is initiated by the client program, the client program needs to know the server’s port number
before it can make a connection. For this reason, servers are placed on ’well-known’ port numbers. For example, a
telnet server is always on port 23. A http server is always on port 80.
52
Chapter 5. Creating server programs
The bind() API call assigns the ’local’ port number. That is, the port number that is used as the ’source port’ on
outgoing datagrams, and the ’destination port’ on incoming datagrams. The connect() API call assigns the ’remote’
port number, which is the one that is the ’destination port’ on outgoing packets and the ’source port’ on incoming
packets.
Okay, back to bind():
If you don’t call bind(), the operating system will automatically assign you an available port number. In the case of
our client programs, we didn’t call the bind() API, and the operating system assigned us whatever port was available.
It didn’t matter which one, since we were the client and we were initiating the connection. The server would know
our port number because we were sending it to the server in every IP datagram that we sent.
For a server, however, we must use a ’well-known’ port number. Without it, no clients will ever find us! Therefore, a
server program will invariably call the bind() API.
There are some named constants that we will want to use with bind(), so we’ll also want to define those in our
SOCKET_H /copy member. They look like this:
D INADDR_ANY
C
D INADDR_BROADCAST...
D
C
D INADDR_LOOPBACK...
D
C
CONST(0)
CONST(4294967295)
CONST(2130706433)
These constants are special values that can be assigned to the sin_addr parameter of the sockaddr_in structure. (You’ll
recall that sockaddr_in is to be used in place of the sockaddr structure when doing TCP/IP related socket calls.)
The value "INADDR_ANY" means that we will bind to any/all IP addresses that the local computer currently has.
Lets say that your AS/400 is dialed into the Internet using a modem and the PPP protocol. The PPP interface is given
an IP address by your ISP. If you also have a network card, and a TCP/IP connection to your LAN, you’ll have an
address for that interface as well. Also, the special TCP/IP interface called ’loopback’ has it’s own special address
(127.0.0.1). If you use the value ’INADDR_ANY’ when calling bind(), you will be able to accept connections from
people connecting to any of these addresses.
On the other hand, if you only want to allow people on your LAN to connect, you could bind specifically to the
network card’s IP address. Likewise, if you only wanted to allow connections from programs on the same computer,
you could specify ’INADDR_LOOPBACK’.
Whew... after all that explanation, calling bind() should be easy. :)
D/copy socktut/qrpglesrc,socket_h
D bindto
D addrlen
c
c
S
S
*
10I 0
** reserve memory for a sockaddr_in structure:
eval
addrlen = %size(sockaddr_in)
alloc
addrlen
bindto
** set sockaddr_in structure so that we can receive connections
**
on port number "port_number" on any address.
c
eval
p_sockaddr = bindto
c
eval
sin_family = AF_INET
c
eval
sin_addr = INADDR_ANY
53
Chapter 5. Creating server programs
c
c
eval
eval
sin_port = port_number
sin_zero = *ALLx’00’
** bind the socket!
c
if
bind(socket: bindto: addrlen) < 0
C* bind() failed. check errno
c
endif
5.3. The listen() API call
Once we have a socket, and we’ve called bind() so that it’s bound to a given port and address, we have to tell the
system that we want to listen for connections.
The listen() API call is documented in IBM’s manual at this location:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/listen.htm
The manual tells us that the C language prototype for the listen() API looks like this:
int listen(int socket_descriptor,
int back_log);
This is an easy one. The RPG data type for an ’int’ is 10I 0. This prototype accepts two integers, and returns an
integer. Easy. Looks like this:
D listen
D
socket_desc
D
back_log
PR
10I 0 ExtProc(’listen’)
10I 0 Value
10I 0 Value
All the listen() API does is tell the system that we’re willing to accept incoming connections. In other words, it turns
this socket into a ’server socket’.
The ’back_log’ parameter tells the system how many clients will be queued up to wait for our server program. You
can think of this as being similar to a print queue... In a print queue, each time a new report is created, it is placed
into the queue. When the printer is ready to print, it takes the first report from the queue and prints it.
The back_log works the same way. When a client calls connect(), it gets put into our queue. If there are more than
’back_log’ connections in our queue, the system sends back a ’Connection Refused’ message to the client.
When we’re ready to talk to the client, we call the accept() API, which takes the first connection off of the queue.
The listen API is quite simple to call:
c
if
listen(socket: 100) < 0
c** listen() failed, check errno!
c
endif
54
Chapter 5. Creating server programs
5.4. The accept() API call
I like the listen() API, don’t you? However, it doesn’t do us much good to queue up clients that have connected to us,
if we don’t ever take them off of the queue!
That’s exactly what the accept() API does. It takes a new connection off of the queue and creates a new socket which
will be used for talking to the client on the other end.
The IBM manual page for the accept() API is found here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/accept.htm
And it tells us that the C language prototype for accept() looks like this:
int accept(int socket_descriptor,
struct sockaddr *address,
int *address_length);
Is this format starting to look familiar yet? The accept() API has the same parameters as the connect() and bind()
APIs do, with one exception. For the ’address_length’ parameter, we pass a pointer to an integer, instead of passing
the value of the integer.
This means that it accepts 3 parameters. An integer, a pointer and another pointer. And, it returns an integer.
You might recall that when you pass something by reference, you’re not actually passing the data itself, but rather
you’re passing the address of that data. Since pointers are the variables that are used to store addresses, passing the
address of a variable is exactly the same thing as passing the value of a pointer. So, instead of passing the third
parameter as a pointer, we can safely pass an integer by reference.
The accept() prototype will, therefore, look like this:
D accept
D
sock_desc
D
address
D
address_len
PR
10I 0 ExtProc(’accept’)
10I 0 Value
*
Value
10I 0
Passing an integer by reference has two advantages:
•
It’s less typing to code "accept(socket: connfrom: len)" than it is to code "accept(socket: connfrom: %addr(len))"
•
The compiler can do better syntax checking, because it knows that only integers are allowed as the 2nd parameter.
So, as was mentioned earlier... the listen() API tells the system that we’re willing to accept connections. When
someone connects, the system puts them in a queue. The accept() API takes the connections off of the queue in
First-In-First-Out (FIFO) order.
Each time you accept a new connection, the IP address and port number of the connecting client is placed in a
sockaddr_in structure that is pointed to by the ’address’ parameter.
The ’address_len’ parameter has two purposes. On input to the accept() API, it contains the amount of memory that
you’ve allocated to the ’address’ parameter. Accept() uses this to ensure that it doesn’t write data beyond what
you’ve allocated. On output from the accept() API, it contains the actual number of bytes that the accept() API wrote
into the area of memory that you supplied as ’address’.
55
Chapter 5. Creating server programs
It’s important to understand that accept() creates a new socket. The original socket which you used with bind() and
listen() will continue to listen for new connections and put them into the backlog queue.
The new socket created by accept() contains your TCP connection to the client program. Once you’ve accept()-ed it,
you can use the send() and recv() APIs to hold a conversation with the client program.
When you’re done talking to the client, you’ll want to call close() for the descriptor returned by accept(). Again,
remember that this is a separate socket from the one that’s listen()-ing. So, when you close() the socket that was
returned from accept(), your program will still be listening for more connections.
To stop listening for connections, you need to call close() for the original socket.
Here’s how you call accept:
D connfrom
D len
c
c
S
S
*
10I 0
eval
alloc
len = %size(sockaddr_in)
len
connfrom
c
eval
newsock = accept(sock: connfrom: len)
c
if
newsock < 0
C* accept() failed. Check errno
c
endif
c
if
len <> 16
C* illegal length for a TCP connection!
c
endif
c
c
c
eval
eval
p_sockaddr = connfrom
msg = ’Received a connection from ’ +
%str(inet_ntoa(sin_addr)) + ’!’
5.5. Our first server program
Now that we’ve been familiarized with the API calls that we need to do a simple server program, lets give one a try.
To keep our first example from getting too complicated, we’re only going to handle one simultanous client. This
program is intended to be very simple, rather than being practical.
Here’s some pseudocode that explains the basic idea behind this example:
1.
Call socket() to make a socket which will listen for connections.
2.
Call bind() to bind our socket to port number 4000.
3.
Call listen() to indicate that we want to accept incoming connections.
4.
Call accept() to create a new socket containing the next connection.
5.
Reject anyone who doesn’t send a valid address.
6.
Send back the IP address of the connected socket, to demonstrate how to use the ’address’ parameter of the
accept() API.
56
Chapter 5. Creating server programs
7.
Ask the client to send a name
8.
Say ’hello <NAME>’
9.
Say ’goodbye <NAME>’
10. After a second, disconnect the client
11. Go back to step 4 to get the next client.
Note: We will continue to use the socket utilities, header files and techniques that we used in the chapters on
client socket programs.
Here’s the code:
File: SOCKTUT/QRPGLESRC, Member: SERVEREX1
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,errno_h
D/copy socktut/qrpglesrc,sockutil_h
D Cmd
D
command
D
length
PR
D die
D
peMsg
PR
D
D
D
D
D
D
D
D
D
S
S
S
S
S
S
S
S
S
len
bindto
connfrom
port
lsock
csock
line
err
clientip
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
256A
10I
*
*
5U
10I
10I
80A
10I
17A
const
0
0
0
0
0
c
eval
*inlr = *on
c
exsr
MakeListener
c
c
c
c
c
dow
exsr
exsr
callp
enddo
1 = 1
AcceptConn
TalkToClient
close(csock)
57
Chapter 5. Creating server programs
C*===============================================================
C* This subroutine sets up a socket to listen for connections
C*===============================================================
CSR
MakeListener begsr
C*-----------------------C** Normally, you’d look the port up in the service table with
C** the getservbyname command.
However, since this is a ’test’
C** protocol -- not an internet standard -- we’ll just pick a
C** port number.
Port 4000 is often used for MUDs... should be
C** free...
c
eval
port = 4000
C* Allocate some space for some socket addresses
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
alloc
len
connfrom
C* make a new socket
c
eval
c
c
if
c
callp
c
return
c
endif
lsock = socket(AF_INET: SOCK_STREAM:
IPPROTO_IP)
lsock < 0
die(’socket(): ’ + %str(strerror(errno)))
C* bind the socket to port 4000, of any IP address
c
eval
p_sockaddr = bindto
c
eval
sin_family = AF_INET
c
eval
sin_addr = INADDR_ANY
c
eval
sin_port = port
c
eval
sin_zero = *ALLx’00’
c
c
c
c
c
c
if
eval
callp
callp
return
endif
bind(lsock: bindto: len) < 0
err = errno
close(lsock)
die(’bind(): ’ + %str(strerror(err)))
C* Indicate that we want to listen for connections
c
if
listen(lsock: 5) < 0
c
eval
err = errno
c
callp
close(lsock)
c
callp
die(’listen(): ’ + %str(strerror(err)))
c
return
c
endif
C*-----------------------CSR
endsr
C*===============================================================
C* This subroutine accepts a new socket connection
C*===============================================================
58
Chapter 5. Creating server programs
CSR
AcceptConn
begsr
C*-----------------------c
dou
len = %size(sockaddr_in)
C* Accept the next connection.
c
eval
len = %size(sockaddr_in)
c
eval
csock = accept(lsock: connfrom: len)
c
if
csock < 0
c
eval
err = errno
c
callp
close(lsock)
c
callp
die(’accept(): ’ + %str(strerror(err)))
c
return
c
endif
C* If socket length is not 16, then the client isn’t sending the
C* same address family as we are using... that scares me, so
C* we’ll kick that guy off.
c
if
len <> %size(sockaddr_in)
c
callp
close(csock)
c
endif
c
enddo
c
eval
c
eval
C*-----------------------CSR
endsr
p_sockaddr = connfrom
clientip = %str(inet_ntoa(sin_addr))
C*===============================================================
C* This does a quick little conversation with the connecting
c* client.
That oughta teach em.
C*===============================================================
CSR
TalkToClient begsr
C*-----------------------c
callp
WrLine(csock: ’Connection from ’ +
c
%trim(clientip))
c
c
callp
WrLine(csock: ’Please enter your name’ +
’ now!’)
c
c
c
if
leavesr
endif
RdLine(csock: %addr(line): 80: *On) < 0
c
c
callp
callp
WrLine(csock: ’Hello ’ + %trim(line))
WrLine(csock: ’Goodbye ’ + %trim(line))
c
callp
C*-----------------------CSR
endsr
Cmd(’DLYJOB DLY(1)’: 200)
59
Chapter 5. Creating server programs
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
60
Chapter 5. Creating server programs
5.6. Testing server programs
Usually server programs are not interactive programs. There is no screen to output details to, and no user to read
those details. So how do we test them, how do we know if they work?
Since we’re usually working with plain ASCII, line-oriented data, we can use a telnet client to debug our server code.
With a telnet client, we’ll type in the code that we expect the client program to send, and we’ll see the exact
responses that the server gives us. This is an invaluable tool for debugging sockets programs :)
Unfortunately (at least as of V4R5) the AS/400’s TELNET command doesn’t work nicely for debugging sockets
programs. However, the Microsoft Windows and Unix (Linux, FreeBSD, etc) telnet clients work just fine.
So, let’s test the server program:
1.
If you haven’t done so already, compile the server program now.
2.
Run the server by typing: CALL SERVEREX1 (or whatever name you compiled it as)
3.
From Microsoft Windows, Click "Start", then "Run" then type: telnet as400 4000
or, from a Linux/BSD/Unix shell, type: telnet as400 4000
4.
When it says ’Please enter your name now!’, type your name.
Note: Depending on your telnet client, you may not be able to see what you type -- this is normal
5.
It should respond with "hello" and "goodbye" and then disconnect.
this same technique can be used to debug just about any server program. For example, if you wanted to test our my
web server, you might type:
1.
telnet www.scottklement.com 80
2.
Once connected, type: GET
http://www.scottklement.com/rpg/socktut/ip_servertesting.txt HTTP/1.0
3.
press enter twice.
It’s always interesting just how much of the Internet can be experienced with a simple telnet client
This first example server program doesn’t know how to end. It’ll just keep running forever, locking out port 4000.
Sure, you could use system request to end the program, or you could do a WRKACTJOB and then ’End Job’, but if
you do that, OS/400 won’t ever take port 4000 out of listen mode!
Therefore, the easiest way to end this program is by using the NETSTAT command.
1.
Sign on to your AS/400 from a different session and type: NETSTAT *CNN
2.
It will show you a list of TCP/IP ports being listened on, as well as showing anyone who is connected, and
which remote and local ports they are connected with.
3.
If you have sufficient authority, you can place a ’4’ next to a given connection to end that connection.
4.
To end our server program, find the line that says that it is in ’Listen’ state on port 4000. Put a ’4’ next to that
line, and end that connection.
61
Chapter 5. Creating server programs
5.
The server program will end, it will say that it got an error: ’accept(): The protocol required to support the
specified address family is not available at this time.’ This is because OS/400 ends the connection by the
protocol unavailable to the program for that particular socket.
5.7. Making our program end
The last topic showed us how to try out our server program. It also showed us that it’s a bit of a pain to make our
server program end.
So, why don’t we make a quick improvement to our server program? We’ll set it up so that if the person’s name is
’quit’, the program shuts itself down. This should be easy to do. It requires some code to be added in two places in
the program:
In the "TalkToClient" subroutine, after the ’RdLine’ group of code, we’ll add a simple if, like this:
c
c
c
if
leavesr
endif
line = ’quit’
And in the mainline of the program, if the user’s name was ’quit’, we need to close the socket that listens for
connections, and end the program.
Right after the line that says ’callp close(csock)’, we add this:
c
c
c
c
if
callp
return
endif
line = ’quit’
close(lsock)
Now, re-compile and run this program again. Test it with telnet. When the server program gets a user whos name is
NOT quit, it should work as it did before. When the user’s name *is* quit, it should end.
There is a problem with this, however. After entering a user-id of ’quit’, the server program ends as it should. But if
you immediately run it again, you get this error:
bind(): Address already in use.
But... it’s not in use... Take a look at NETSTAT! In fact, if you wait two or three minutes before running your
program again, it works just fine.
This is one of the most common "gotchas" in socket programming. The same port cannot be opened by two different
programs, without taking special actions to make it work. Even though the socket from the first run of our sample
program has already been closed, the system still waits on that socket for a period of time, before allowing you to
re-use it.
The reason it waits has to do with how the TCP protocol works. TCP is a ’reliable’ protocol, it ensures that anything
sent by one side of the connection gets received by the other side of the connection. It sends the data in chunks called
’segments’. Each time a segment is received, the receiving side sends back an ’ACK’ (acknowledgement) datagram,
so that the sending side knows that the data was received. If it never gets an ACK, it re-sends the data.
62
Chapter 5. Creating server programs
When a connection is closed, each side sends a ’FIN’ (finished) datagram to the other. Once both sides have received
a ’FIN’, they know that the connection is closed.
So, like everything else in TCP, after a FIN is received, the side that received it sends back an ’ACK’. But how does
it know that the ACK was received? You can’t acknowledge an ’ACK’, that would create an endless chain of ACKs
going back and forth.
So how does the side sending back the ’final ACK’ know that the ACK has been received?
The answer is... it doesn’t. But it needs to wait a period of time, just in case the ACK got lost, so that it can re-send
the ACK in the event that it never got received. (It will know that the ACK never got received when the FIN
datagram is re-sent) RFC 793, which is the standard for how TCP is to work, tells us that the socket which sends the
’final ACK’ is to wait for twice the Maximum Segment Lifetime (MSL) before closing completely.
Getting back to the original question, each time you close a socket, it must wait until the (MSL x 2) has passed
before it can completely close the socket. During the time that it’s waiting, the AS/400 will show that socket as being
in ’TIME-WAIT’ state. Since the socket that’s in TIME-WAIT state is still using our port! And, by default, two
sockets can’t have the same port open! And THAT is why you receive the "Address already in use" error.
Fortunately, there is an easy fix. Each socket has many different options that can be set using the setsockopt() API.
One of those options is called "SO_REUSEADDR", and it allows a socket to share the same address and port as an
existing socket. If we turn this option on, we will be able to re-bind to our port immediately.
Many people have asked me, "isn’t it dangerous to allow more than one socket to bind to the same port? What if two
different programs tried to use the same port at the same time?"
It’s not as dangerous as you might think. Although it’s possible for two programs to bind to the same port this way,
it’s still NOT possible for two programs to LISTEN on the same port. Therefore, if you had two server programs
trying to bind to the same port, the second one would still get an error, because the first one was already listening.
5.8. The setsockopt() API call
As I explained in the previous topic, there is an API that can be used to set or change certain attributes about a
socket, called ’setsockopt’. Here we will detail using that API call in RPG.
The IBM manual page for the setsockopt() API is found here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/ssocko.htm
The ever-helpful IBM manual refers to setsockopt() as "Set Socket Options" and proceeds to tell us "The
setsockopt() function is used to set socket options". Gee, thanks.
The C language prototype for setsockopt() looks like this:
int setsockopt(int socket_descriptor,
int level,
int option_name,
char *option_value,
int option_length);
By now, you’re probably an old-pro at converting these prototypes to RPG. The C prototype tells us that the
setsockopt() procedure receives 5 parameters. These parameters are an integer, an integer, another integer, a pointer,
and yet another integer. It also returns an integer.
63
Chapter 5. Creating server programs
The ’char *option_value’ parameter does require a bit more research. We know that it’s a pointer to a character
variable. But is it supposed to be a null-terminated string of characters? or is it supposed to be a single byte? or what?
So, we page down to where it describes the option_value, we find that it says "A pointer to the option value. Integer
flags/values are required by setsockopt() for all the socket options except SO_LONGER, IP_OPTIONS, . . ."
Oh great. So, it tells us it’s a pointer to a character... but it really wants a pointer to an integer most of the time, and a
pointer to other things at other times.
Fortunately, this doesn’t present a big problem for us. In RPG, pointers can point to any data type. So really all we
needed to learn from this is that we do not want to make it a pointer to a null terminated string.
Here’s our RPG prototype. Hope it provides hours of enjoyment:
D setsockopt
PR
D
socket_desc
D
level
D
option_name
D
option_value
D
option_length
10I
10I
10I
10I
*
10I
0
0
0
0
ExtProc(’setsockopt’)
Value
Value
Value
Value
0 Value
Please add that prototype to your SOCKET_H member.
Now, we’ll need to define the constants for the ’level’ and ’option_name’ parameters, but first, a bit of explanation.
TCP/IP is a "layered protocol". At the start of this tutorial, I explained a bit how this works, there’s the Internet
Protocol (IP) that handles the transportation of data across an inter-network. Running "on top of" that is (in this case)
the Transmission Control Protocol (TCP), which uses the IP protocol to send and receive data, but adds the extra
functionality of operating like a reliable-stream of data. On top of that is the Application-Level Protocol that varies
from program to program. We send and receive this application data by calling the send() and recv() options on a
socket.
So, when setting the options that control how a socket works, we may be setting them at different levels. Some
options are set at the IP level, others at the TCP level, and others may pertain to the socket itself, so we have a special
"socket level" to deal with.
For example, there is a "Time To Live" (TTL) value associated with every packet sent over the internet. This TTL
defines the maximum number of routers or gateways the packet can travel through before being dropped.
This is useful because sometimes people make mistakes and misconfigure a router. If packets didnt have a
maxmimum time to live, it would be possible for packets in error to bounce around the internet forever. Since this
’TTL’ value relates to the routing of datagrams, it must be a part of the IP layer, therefore to change it with
setsockopt, you set the ’level’ parameter to the constant that means "IP layer" and set the option_name parameter to
the constant that means "TTL".
Here are the values for the different "levels". Note that some of these are already defined in your SOCKET_H
member, because they’re used by the socket() API as well. You should add the other ones, though:
D*
D IPPROTO_IP
D*
D*
D IPPROTO_TCP
D*
D*
D IPPROTO_UDP
Internet Protocol
C
CONST(0)
Transmission Control
Protocol
C
CONST(6)
Unordered Datagram
Protocol
C
CONST(17)
64
Chapter 5. Creating server programs
D*
D IPPROTO_RAW
D*
D*
D IPPROTO_ICMP
D*
D SOL_SOCKET
Raw Packets
C
CONST(255)
Internet Control
Msg Protocol
C
CONST(1)
C
CONST(-1)
socket layer
Here are the values for the "option_name" parameter that are used at the "IP level":
D*
D IP_OPTIONS
C
D*
D IP_TOS
C
D*
D IP_TTL
C
D*
D IP_RECVLCLIFADDR...
D
C
ip options
CONST(5)
type of service
CONST(10)
time to live
CONST(15)
recv lcl ifc addr
CONST(99)
Here are the values for the "option_name" parameter that are used at the "TCP level":
D*
D TCP_MAXSEG
D*
D TCP_NODELAY
max segment size (MSS)
C
5
C
10
dont delay small packets
Here are the values for the "option_name" parameter that are used at the "Socket Level":
D*
D SO_BROADCAST
D*
D SO_DEBUG
D*
D*
D SO_DONTROUTE
D*
D SO_ERROR
D*
D SO_KEEPALIVE
D*
D SO_LINGER
D*
D SO_OOBINLINE
D*
D SO_RCVBUF
D*
D SO_RCVLOWAT
D*
D SO_RCVTIMEO
D*
allow broadcast msgs
C
5
C
10
record debug informatio
just use interfaces,
bypass routing
C
15
C
20
C
25
C
30
C
35
C
40
C
45
C
50
error status
keep connections alive
linger upon close
out-of-band data inline
receive buffer size
receive low water mark
receive timeout value
re-use local address
65
Chapter 5. Creating server programs
D SO_REUSEADDR
D*
D SO_SNDBUF
D*
D SO_SNDLOWAT
D*
D SO_SNDTIMEO
D*
D SO_TYPE
D*
D SO_USELOOPBACK
C
55
C
60
C
65
C
70
C
75
C
80
send buffer size
send low water mark
send timeout value
socket type
send loopback
In addition to all of those options that you can set, there is one option that accepts a data structure as a parameter.
That’s the ’SO_LINGER’ option. So, to be complete, we’ll also add the linger structure to our SOCKET_H header
member:
D p_linger
D linger
D
l_onoff
D
l_linger
S
DS
*
BASED(p_linger)
10I 0
10I 0
If you’re interested in seeing the C definitions for all of these items that we just added, you can find them in the
following members in your QSYSINC ("System Openness Includes") library:
File: QSYSINC/SYS
Member: SOCKET
File: QSYSINC/NETINET Member: IN
File: QSYSINC/NETINET Member: TCP
Obviously, due to the large number of options that can be set using the setsockopt() API, the method of calling it will
vary quite a bit. However, here are 3 different examples:
D value
D len
D ling
S
S
S
10I 0
10I 0
*
*** Change datagram TTL to 16 hops:
***
c
eval
value = 16
c
if
setsockopt(s: IPPROTO_IP: IP_TTL:
c
%addr(value): %size(value)) < 0
C* setsockopt() failed, check errno
c
endif
*** Allow re-use of port in bind():
***
(note that 1 = on, so we’re turning the ’re-use address’
**
capability on)
c
eval
value = 1
c
if
setsockopt(s: SOL_SOCKET: SO_REUSEADDR:
66
Chapter 5. Creating server programs
c
%addr(value): %size(value)) < 0
C* setsockopt() failed, check errno
c
endif
c
c
c
** Make space for a linger structure:
eval
len = %size(linger)
alloc
len
ling
eval
p_linger = ling
*** tell the system to discard buffered data 1 minute
*** after the socket is closed:
c
eval
l_onoff = 1
c
eval
l_linger = 60
c
if
setsockopt(s: SOL_SOCKET: SO_LINGER:
c
ling: %size(linger)) < 0
C* setsockopt() failed, check errno
c
endif
5.9. The revised server program
The vast majority of options in the setsockopt() API are only used rarely. However, when running a server, it’s a very
good idea to set two of them:
•
SO_REUSEADDR -- As we already discussed, if we want to use a specific port, but don’t want to wait for the
MSL timeout, we need to turn this option on in our server program.
•
SO_LINGER -- Whenever the send() API is used to send data, it gets placed into a buffer, and is actually sent
when the program on the other end of the connection is ready to recv() it. If we’ve sent some data, then called
close() to drop the connection, what should the system do with data that’s still in its send buffer? By default, it
tries to send it forever. However, network errors could cause this to waste some memory indefinitely! So, for long
running programs like a server, we should tell it how long to linger on that data...
Therefore, we’ll insert these new options into our example server program. They’ll go into the "MakeListner"
subroutine, right after we’ve called the socket() API.
C* Tell socket that we want to be able to re-use port 4000
C* without waiting for the MSL timeout:
c
callp
setsockopt(lsock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C* create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C* Data shouldnt need to linger on the "listener" socket, but
67
Chapter 5. Creating server programs
C*
c
c
c
c
give it one second, just
eval
eval
callp
in case:
l_onoff = 1
l_linger = 120
setsockopt(lsock: SOL_SOCKET: SO_LINGER:
ling: linglen)
Now, remember, we have two sockets, the one we create in MakeListener, and the one that’s returned by the accept()
API in the AcceptConn subroutine. The SO_REUSEADDR option wouldn’t make sense on the accept() socket
(since we don’t bind it) but we should set the SO_LINGER value!
So, insert this code, right after we call accept():
C* tell socket to only linger
c
eval
c
eval
c
callp
c
for 2 minutes, then discard:
l_onoff = 1
l_linger = 120
setsockopt(csock: SOL_SOCKET: SO_LINGER:
ling: linglen)
After adding these options, we can go ahead and compile & test our program again. Notice that now, we can re-run
the program immediately, after it ends, we don’t have to wait 2 minutes...
Just to dispel any confusion, here’s a full listing of our server program up to this point:
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,errno_h
D/copy socktut/qrpglesrc,sockutil_h
D Cmd
D
command
D
length
PR
D die
D
peMsg
PR
D
D
D
D
D
D
D
D
D
D
D
D
S
S
S
S
S
S
S
S
S
S
S
S
c
len
bindto
connfrom
port
lsock
csock
line
err
clientip
ling
linglen
on
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
256A
10I
*
*
5U
10I
10I
80A
10I
17A
*
10I
10I
eval
const
0
0
0
0
0
0
0 inz(1)
*inlr = *on
68
Chapter 5. Creating server programs
c
exsr
MakeListener
c
c
c
c
dow
exsr
exsr
callp
1 = 1
AcceptConn
TalkToClient
close(csock)
c
c
c
c
if
callp
return
endif
line = ’quit’
close(lsock)
c
enddo
C*===============================================================
C* This subroutine sets up a socket to listen for connections
C*===============================================================
CSR
MakeListener begsr
C*-----------------------C** Normally, you’d look the port up in the service table with
C** the getservbyname command.
However, since this is a ’test’
C** protocol -- not an internet standard -- we’ll just pick a
C** port number.
Port 4000 is often used for MUDs... should be
C** free...
c
eval
port = 4000
C* Allocate some space for some socket addresses
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
alloc
len
connfrom
C* make a new socket
c
eval
c
c
if
c
callp
c
return
c
endif
lsock = socket(AF_INET: SOCK_STREAM:
IPPROTO_IP)
lsock < 0
die(’socket(): ’ + %str(strerror(errno)))
C* Tell socket that we want to be able to re-use port 4000
C* without waiting for the MSL timeout:
c
callp
setsockopt(lsock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C* create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C* tell socket to only linger for 2 minutes, then discard:
c
eval
l_onoff = 1
c
eval
l_linger = 1
69
Chapter 5. Creating server programs
c
c
callp
setsockopt(lsock: SOL_SOCKET: SO_LINGER:
ling: linglen)
C* bind the socket to port 4000, of any IP address
c
eval
p_sockaddr = bindto
c
eval
sin_family = AF_INET
c
eval
sin_addr = INADDR_ANY
c
eval
sin_port = port
c
eval
sin_zero = *ALLx’00’
c
c
c
c
c
c
if
eval
callp
callp
return
endif
bind(lsock: bindto: len) < 0
err = errno
close(lsock)
die(’bind(): ’ + %str(strerror(err)))
C* Indicate that we want to listen for connections
c
if
listen(lsock: 5) < 0
c
eval
err = errno
c
callp
close(lsock)
c
callp
die(’listen(): ’ + %str(strerror(err)))
c
return
c
endif
C*-----------------------CSR
endsr
C*===============================================================
C* This subroutine accepts a new socket connection
C*===============================================================
CSR
AcceptConn
begsr
C*-----------------------c
dou
len = %size(sockaddr_in)
C* Accept the next connection.
c
eval
len = %size(sockaddr_in)
c
eval
csock = accept(lsock: connfrom: len)
c
if
csock < 0
c
eval
err = errno
c
callp
close(lsock)
c
callp
die(’accept(): ’ + %str(strerror(err)))
c
return
c
endif
C* tell socket to only linger
c
eval
c
eval
c
callp
c
for 2 minutes, then discard:
l_onoff = 1
l_linger = 120
setsockopt(csock: SOL_SOCKET: SO_LINGER:
ling: linglen)
C* If socket length is not 16, then the client isn’t sending the
C* same address family as we are using... that scares me, so
70
Chapter 5. Creating server programs
C*
c
c
c
we’ll kick that guy off.
if
callp
endif
c
len <> %size(sockaddr_in)
close(csock)
enddo
c
eval
c
eval
C*-----------------------CSR
endsr
p_sockaddr = connfrom
clientip = %str(inet_ntoa(sin_addr))
C*===============================================================
C* This does a quick little conversation with the connecting
c* client.
That oughta teach em.
C*===============================================================
CSR
TalkToClient begsr
C*-----------------------c
callp
WrLine(csock: ’Connection from ’ +
c
%trim(clientip))
c
c
callp
WrLine(csock: ’Please enter your name’ +
’ now!’)
c
c
c
if
leavesr
endif
RdLine(csock: %addr(line): 80: *On) < 0
c
c
c
if
leavesr
endif
c
c
callp
callp
c
callp
C*-----------------------CSR
endsr
line = ’quit’
WrLine(csock: ’Hello ’ + %trim(line))
WrLine(csock: ’Goodbye ’ + %trim(line))
Cmd(’DLYJOB DLY(1)’: 200)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
PR
7A
20A
256A
ExtPgm(’QMHSNDPM’)
Const
Const
Const
71
Chapter 5. Creating server programs
D
D
D
D
D
D
MsgDtaLen
MsgType
CallStkEnt
CallStkCnt
MessageKey
ErrorCode
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
72
Chapter 6. Handling many sockets at once
using select()
Written by Scott Klement.
6.1. Overview of handling many clients
There are three different approaches to making a server program be capable of handing many simultaneous clients.
These approaches are:
•
Have a single program, running as a single process, that switches between all of the connected clients.
•
Have a "listener" program that listens for new connections, and then hands off each new client to a separate job.
•
Use threads to allow a single, multi-threaded program to handle many clients. (However, threads are not available
in RPG, at least not in V4R5)
There are pros and cons to each of these approaches, and special issues that need to be dealt with. This chapter will
deal with the first approach, and the next chapter will deal with the second approach.
6.1.1. The "single program" approach:
This approach involves a single program that reads from many clients, and writes to many clients in a loop. The
program needs to be careful not to "stop" on any one client’s work, but to keep switching between all of the involved
clients so they all appear to be working.
The pros of this approach:
•
It is efficient. There is no "startup overhead" such as waiting for a new job to be submitted, and it only needs one
set of resources.
•
It allows for easy communication between all of the connected clients. For example, if you wanted to write a chat
room where everything being said had to echoed to all of the connected clients, you’d want to use this approach,
because all of the sockets are together in one program where data can be copied from one to the other.
The cons of this approach:
•
It’s very hard to limit each user’s access according to the userid they signed in with, since all actions for all users
are being taken by a single program.
•
You can’t really call other programs to implement functionality, since your program would have to give up control
to the program that you call, and when that happened, all of the socket processing would stop.
•
It’s very easy for your code to become very complicated and diffiult to maintain.
Up to this point in the tutorial, all of our socket operations have used "blocking" operations. When I say "blocking", I
mean that each operation doesn’t return control to our program until it has completed. For example, if we call recv(),
the computer will wait until there is data to recv() from the other side of the connection. It doesn’t stop waiting until
the other side has disconnected, or until some data actually appears.
73
Chapter 6. Handling many sockets at once using select()
When working with many different clients, however, this can be a problem. We need to be able to read from
whichever client actually sends data to us, without being stuck waiting for one that’s not sending data to us.
To solve all these problems, the select() API was created. The select API allows us to specify 3 different "groups" or
"sets" of sockets. A "read" set, a "write" set and an "exceptional" set. It can tell us when there is data to be read on
each of the sockets, as well as telling us when it’s safe to write to each of the sockets. Select() also has a "timeout"
value which lets us regain control even when there has been no activity.
Extending our server program into one based on this model will be difficult, so we’ll start with something a bit
simpler than that.
You’ll recall that we tested our server programs by using a TELNET client to simply connect to the port we were
listening on. But, what if we wanted to test a client program instead of a server program? We couldn’t do that
because a TELNET client only connects to servers, not to clients...
What we need is a special server program that would accept a connection from a client program as well as a telnet
client. It could then copy the data sent from the client to the telnet socket, and vice-versa.
This program would be relatively easy to write, and would be a good starting example of how to deal with multiple
connected clients -- as well as a useful tool -- so that’s what we’ll start with in this chapter.
6.2. The select() API call
The most important API to use when trying to work with multiple clients in the same instance of the same program is
very definitely the select() API.
The IBM manual page for the select() API is found here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/sselect.htm
The manual tells us that the C language prototype for select() looks like this:
int select(int max_descriptor,
fd_set *read_set,
fd_set *write_set,
fd_set *exception_set,
struct timeval *wait_time);
This tells us that the select() API accepts 5 parameters, and they are an integer, a pointer to a ’fd_set’, another pointer
to an ’fd_set’, another pointer to an ’fd_set’ and a pointer to a ’timeval’ structure.
Therefore, the RPG prototype for the select() API looks like this:
D select
D
max_desc
D
read_set
D
write_set
D
except_set
D
wait_time
PR
10I 0 extproc(’select’)
10I 0 VALUE
*
VALUE
*
VALUE
*
VALUE
*
VALUE
Read the IBM page over, the information here should give you a basic idea of what exactly the select() API does.
74
Chapter 6. Handling many sockets at once using select()
The ’fd_set’ data type is a string of 224 bits, each representing a descriptor to check for data in. In C, the fd_set data
type is represented as an array of integers. The reason for this is that it’s easy to bits on & off in an integer variable in
C. In RPG, however, I prefer to work with a simple string of character variables, since that’s what the bitwise
operations in RPG work with.
In C, there are 4 ’preprocessor macros’ that are included to make it easy to work with the data in a descriptor set. In
order to do the same thing in RPG, we’ll have to write subprocedures of our own that mimic these macros.
I’ll describe the ’fd_set’ data type, and the subprocedures that work with it in the next topic. Suffice it to say, for now,
that I use a descriptor set that’s defined as ’28A’ in RPG.
The last parameter to select() is of type ’struct timeval’. You could call it a ’time value structure’ if you like.
The ’timeval’ structure in C looks like this:
struct timeval {
long tv_sec;
long tv_usec;
};
/* seconds
/* microseconds
*/
*/
It’s nothing more than 2 32-bit integers in a structure. The first integer is the number of seconds to wait before the
select() operation times out. The second integer specifies the number of microseconds (1,000,000th of a second) to
wait before timing out.
So, the timeval structure in RPG would look like this:
D p_timeval
D timeval
D
tv_sec
D
tv_usec
S
DS
*
based(p_timeval)
10I 0
10I 0
Make sure you add the prototype for the select() API as well as the definition of the ’timeval’ structure to your
SOCKET_H header file.
You call the select() API in RPG like this:
D
D
D
D
D
readset
writeset
excpset
tv_len
tv
S
S
S
S
S
28A
28A
28A
10I 0
*
* create a timeval struct, set it to 10.5 seconds:
C
C
C
eval
alloc
eval
tv_len = %size(timeval)
tv_len
tv
p_timeval = tv
C
C
eval
eval
tv_sec = 10
tv_usec = 500000
* call select.
75
Chapter 6. Handling many sockets at once using select()
C
c
c
eval
rc = select(maxsock+1: %addr(readset):
%addr(writeset): %addr(excpset):
p_timeval)
6.3. Utility routines for working with select()
Working with descriptor sets (the ’fd_set’ data type in C) is a little tricky in RPG. We will want to add utility routines
to our ’sockutil’ service program to help us work with them.
As you know, the socket API returns an integer. This integer that it returns is called a socket descriptor. The first
descriptor opened in a job will always return a 0. The next one will be a 1, then a 2, etc. Of course, the system itself,
along with every program in the job, can open some of these -- so you really never know what number you’ll get.
Remember that the select() API is able to tell you which sockets are ’readable’, ’writable’ or ’have a pending
exception’. That means that you need a way to give select() a whole list of sockets, and it needs a way to return a
whole list of results.
You do this by using a bit string. For example, if you called socket() twice, and the first time it returned a 1, and the
second time it returned a 5, you’ll set on bit #1, and bit #5 in the bit string. Now select will know that it needs to
check to see if there’s data to read on descriptors 1 and 5.
This string of bits is called an ’fd_set’ (which stands for ’file descriptor set’, since select() can also be used with
stream files)
In C, the fd_set data type is an array of unsigned integers. Each integer is 32-bits long (that is, 4 bytes) and the array
always has at least 7 integers in it. Therefore, the minimum size for a fd_set array is 28 bytes long.
The bits of each integer of the array are numbered from 0 to 31, starting with the rightmost (least significant) bit.
Therefore, if we were to number the descriptors, it’d look something like this:
First integer in array represents the first 32 descriptors:
3
2
2
1
1
1.....5....0....5....0....5....0
Second integer in array represents the next 32-descriptors:
6 6
5
5
4
4
3 3
3..0....5....0....5....0....5..2
And so on... until we got to descriptor #223.
Since bitwise operations in RPG are actually easier to do on character fields than they are on integer fields, I
implemented the ’fd_set’ data type in RPG as a simple 28-byte long alphanumeric field.
Each bit in that field must be numbered the same way that the array of integers was for the C programs, however,
since we’re calling the same select() API that C programs will call.
76
Chapter 6. Handling many sockets at once using select()
Therefore, I wrote the following subprocedure that will calculate which will take a descriptor number, and calculate
which byte in the ’28A’ field needs to be checked, and a bitmask which can be used to set the specific bit within that
byte on, or off, of whatever you want to do with it.
This subprocedure looks like this:
P CalcBitPos
D CalcBitPos
D
peDescr
D
peByteNo
D
peBitMask
D dsMakeMask
D
dsZeroByte
D
dsMask
D
dsBitMult
C
peDescr
C
C
C
C
c
c
c
c
P
B
PI
10I 0
5I 0
1A
DS
1
2
1
div
mvr
div
mvr
eval
eval
eval
eval
eval
1A
2A
2U 0 INZ(0)
32
wkGroup
5 0
wkByteNo
2 0
8
wkByteNo
2 0
wkBitNo
2 0
wkByteNo = 4 - wkByteNo
peByteNo = (wkGroup * 4) + wkByteNo
dsBitMult = 2 ** wkBitNo
dsZeroByte = x’00’
peBitMask = dsMask
E
This routine will be called by many of the subprocedures that will write to replace the C-language ’pre-processor
macros’.
The first pre-processor macro that we need to duplicate will be the ’FD_ZERO’ macro. All this does is zero out all of
the bits in a ’fd_set’. To do this in RPG, using a 28A field is child’s play. We just do this:
P*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P* Clear All descriptors in a set. (also initializes at start)
P*
P*
peFDSet = descriptor set
P*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P FD_ZERO
B
EXPORT
D FD_ZERO
PI
D
peFDSet
28A
C
eval
peFDSet = *ALLx’00’
P
E
next, we’ll need to do ’FD_SET’. First of all, remember that C is case-sensitive, so ’FD_SET’ is a different name
from ’fd_set’. The FD_SET macro in C is used to turn on one bit (one descriptor) in a descriptor set.
To do this in RPG, we’ll have to call the CalcBitPos routine that we described above to find out which byte &
bitmask to use, then we’ll use that bitmask to turn on the appropriate bit. Therefore, our FD_SET subprocedure will
look like this:
P FD_SET
D FD_SET
B
PI
EXPORT
77
Chapter 6. Handling many sockets at once using select()
D
peFD
D
peFDSet
D wkByteNo
D wkMask
D wkByte
C
c
c
c
P
S
S
S
callp
eval
biton
eval
10I 0
28A
5I 0
1A
1A
CalcBitPos(peFD:wkByteNo:wkMask)
wkByte = %subst(peFDSet:wkByteNo:1)
wkMask
wkByte
%subst(peFDSet:wkByteNo:1) = wkByte
E
The opposite of FD_SET is FD_CLR. FD_CLR turns a single bit off in a descriptor set. The code to do this will be
nearly the same thing as FD_SET, except that we turn the bit off instead of on. Like so:
P
D
D
D
D
D
D
C
c
c
c
P
FD_CLR
FD_CLR
peFD
peFDSet
wkByteNo
wkMask
wkByte
B
PI
EXPORT
S
S
S
callp
eval
bitoff
eval
10I 0
28A
5I 0
1A
1A
CalcBitPos(peFD:wkByteNo:wkMask)
wkByte = %subst(peFDSet:wkByteNo:1)
wkMask
wkByte
%subst(peFDSet:wkByteNo:1) = wkByte
E
The last utility routine that we need is one that can be used to check to see if a bit is still on after calling select(). To
make this work nicely in RPG, it’ll help to return an indicator field -- so that we can simply check for *ON or *OFF
when we call the routine.
This is done with the FD_ISSET macro in C. We simulate it in RPG like this:
P
D
D
D
D
D
D
C
c
c
c
P
FD_ISSET
FD_ISSET
peFD
peFDSet
wkByteNo
wkMask
wkByte
B
PI
S
S
S
callp
eval
testb
return
EXPORT
1N
10I 0
28A
5I 0
1A
1A
CalcBitPos(peFD:wkByteNo:wkMask)
wkByte = %subst(peFDSet:wkByteNo:1)
wkMask
wkByte
*IN88
88
E
Each of these subprocedures listed here should be placed in your SOCKUTILR4 service program. In addition, the
prototypes for the FD_xxx procedures should be placed in the SOCKUTIL_H source member so that they are
available when you call them from other programs.
78
Chapter 6. Handling many sockets at once using select()
The CalcBitPos prototype should be placed at the top of the SOCKUTILR4 member so that routines within that
service program can call it. Nobody should need to call CalcBitPos from outside the service program.
You might also find it helpful to define a 28A field in your SOCKUTIL_H member called ’fdset’. (can’t call it fd_set,
since the name is already taken by the subprocedure) Then you can define your descriptors as being ’like(fdset)’.
For example, add this to SOCKUTIL_H:
D fdset
S
28A
Then use it like this:
D readset
D writeset
D excpset
S
S
S
like(fdset)
like(fdset)
like(fdset)
If you prefer, all of the code for SOCKUTIL_H, SOCKUTILR4, SOCKET_H and ERRNO_H are on my website,
and you can simply download them and use them directly.
http://www.scottklement.com/rpg/socktut/qrpglesrc.sockutil_h
http://www.scottklement.com/rpg/socktut/qrpglesrc.sockutilr4
http://www.scottklement.com/rpg/socktut/qrpglesrc.socket_h
http://www.scottklement.com/rpg/socktut/qrpglesrc.errno_h
Note: Since I originally posted my FD_xxx code, and advice for using the select() API, I’ve received a little bit of
criticism for using a character string for ’fd_set’ instead of an array of integers. However, I stand by my decision to
use a 28A field. I think it’s nicer, especially because of the ability to use "like" to set the data type.
Now that you’ve added these routines to the SOCKUTILR4 service program, you’ll need to recompile it. You want
the new FD_xx routines to be exported so that they can be called from other programs.
To do that, edit the binding language source that you created in chapter 4, so that they now look like this:
STRPGMEXP PGMLVL(*CURRENT)
EXPORT SYMBOL(RDLINE)
EXPORT SYMBOL(WRLINE)
EXPORT SYMBOL(FD_SET)
EXPORT SYMBOL(FD_CLR)
EXPORT SYMBOL(FD_ISSET)
EXPORT SYMBOL(FD_ZERO)
ENDPGMEXP
STRPGMEXP PGMLVL(*PRV)
EXPORT SYMBOL(RDLINE)
EXPORT SYMBOL(WRLINE)
ENDPGMEXP
As you can see... we’ve changed the ’*CURRENT’ section to list all of the procedures that we’re now exporting.
We’ve also added a ’*PRV’ section that will create a ’previous version signature’ for the binder. This previous
version only contains the procedures that we had before our changes. Therefore, the service program will remain
backward compatible, like we discussed back in Chapter 4.
79
Chapter 6. Handling many sockets at once using select()
Type the following commands to re-compile the SOCKUTILR4 service program:
CRTRPGMOD SOCKUTILR4 SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
CRTSRVPGM SOCKUTILR4 EXPORT(*SRCFILE) SRCFILE(SOCKTUT/QSRVSRC) TEXT(’Socket Utility
Service Pgm’) ACTGRP(*CALLER)
6.4. A combination client/server example
Now that we’ve had a fairly thorough discussion of how the select API works, and we’ve put the appropriate utilities
into our SOCKUTILR4 service program, we’re ready to do an example program using select.
This example program is the "proxy" that was described in the introduction to this chapter. It’s main purpose will be
to allow us to use a telnet client to debug other client programs.
In fact, you’ll be able to use it to interact with just about any client program around -- including your web browser... :)
Here’s how this program will work:
1.
We accept two port numbers as parms from the command line.
2.
As we’ve done in the previous chapter, set up a socket to bind() and listen() to each of the two ports.
3.
accept() a connection on the first port. Then close the listener for that port (we’ll only handle one connection per
port)
4.
accept() a connection on the second port. Then close the listener for THAT port.
5.
Set things up so that select() will detect when there is either data to read or an exceptional condition on either of
these sockets.
6.
If there was data to read on the first socket, simply send() that data to the other socket. and vice-versa.
7.
If we receive any errors or exceptional conditions, exit the program. (It probably means that one side or the other
has disconnected)
8.
If we didn’t receive any errors, go back to step 5.
Here’s the source to this new example program. At the bottom of this page, I’ll show you how to play with it. :)
File: SOCKTUT/QRPGLESRC Member: PROXY1
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D/copy socktut/qrpglesrc,errno_h
D die
D
peMsg
PR
D NewListener
D
pePort
D
peError
PR
256A
const
10I 0
5U 0 value
256A
80
Chapter 6. Handling many sockets at once using select()
D copysock
D fromsock
D tosock
PR
10I 0
10I 0 value
10I 0 value
D
D
D
D
D
D
D
D
D
D
D
D
S
S
S
S
S
S
S
S
S
S
S
S
10I
10I
10I
10I
15P
15P
10I
*
s1
s2
s1c
s2c
port1
port2
len
connfrom
read_set
excp_set
errmsg
max
c
C
c
c
c
c
c
c
c
like(fdset)
like(fdset)
256A
10I 0
eval
*entry
0
0
0
0
5
5
0
plist
parm
parm
if
callp
*inlr = *on
port1
port2
%parms < 2
die(’This program requires two port ’ +
’numbers as parameters!’)
return
endif
C*************************************************
C* Listen on both ports given
C*************************************************
c
eval
s1 = NewListener(port1: errmsg)
c
if
s1 < 0
c
callp
die(errmsg)
c
return
c
endif
c
c
c
c
c
eval
if
callp
return
endif
s2 = NewListener(port2: errmsg)
s2 < 0
die(errmsg)
C*************************************************
C* Get a client on first port, then stop
C*
listening for more connections
C*************************************************
c
eval
len = %size(sockaddr_in)
c
eval
s1c = accept(s1: connfrom: len)
c
if
s1c < 0
c
eval
errmsg = %str(strerror(errno))
c
callp
close(s1)
81
Chapter 6. Handling many sockets at once using select()
c
c
c
c
c
callp
callp
return
endif
callp
close(s2)
die(errmsg)
close(s1)
C*************************************************
C* Get a client on second port, then stop
C*
listening for more connections
C*************************************************
c
eval
len = %size(sockaddr_in)
c
eval
s2c = accept(s2: connfrom: len)
c
if
s2c < 0
c
eval
errmsg = %str(strerror(errno))
c
callp
close(s1c)
c
callp
close(s2)
c
callp
die(errmsg)
c
return
c
endif
c
callp
close(s2)
c
c
c
c
eval
if
eval
endif
max = s1c
s2c > max
max = s2c
C*************************************************
C* Main loop:
C*
1) create a descriptor set containing the
C*
"socket 1 client" descriptor and the
C*
"socket 2 client" descriptor
C*
C*
2) Wait until data appears on either the
C*
"s1c" socket or the "s2c" socket.
C*
C*
3) If data is found on s1c, copy it to
C*
s2c.
C*
C*
4) If data is found on s2c, copy it to
C*
s1c.
C*
C*
5) If any errors occur, close the sockets
C*
and end the proxy.
C*************************************************
c
dow
1 = 1
c
c
c
c
callp
callp
callp
eval
FD_ZERO(read_set)
FD_SET(s1c: read_set)
FD_SET(s2c: read_set)
excp_set = read_set
c
c
if
select(max+1: %addr(read_set): *NULL:
%addr(excp_set): *NULL) < 0
82
Chapter 6. Handling many sockets at once using select()
c
c
leave
endif
c
c
c
c
if
c
c
c
c
c
if
if
leave
endif
endif
FD_ISSET(s1c: read_set)
copysock(s1c: s2c) < 0
c
c
c
c
c
if
if
leave
endif
endif
FD_ISSET(s2c: read_set)
copysock(s2c: s1c) < 0
c
enddo
c
c
c
callp
callp
return
FD_ISSET(s1c: excp_set)
or FD_ISSET(s2c: excp_set)
leave
endif
close(s1c)
close(s2c)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Create a new TCP socket that’s listening to a port
*
*
parms:
*
pePort = port to listen to
*
peError = Error message (returned)
*
*
returns: socket descriptor upon success, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewListener
B
D NewListener
PI
10I 0
D
pePort
5U 0 value
D
peError
256A
D
D
D
D
D
D
sock
len
bindto
on
linglen
ling
S
S
S
S
S
S
C*** Create a socket
c
eval
c
10I
10I
*
10I
10I
*
0
0
0 inz(1)
0
sock = socket(AF_INET:SOCK_STREAM:
IPPROTO_IP)
83
Chapter 6. Handling many sockets at once using select()
c
c
c
c
if
eval
return
endif
sock < 0
peError = %str(strerror(errno))
-1
C*** Tell socket that we want to be able to re-use the server
C*** port without waiting for the MSL timeout:
c
callp
setsockopt(sock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C*** create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C*** tell socket to only linger for 2 minutes, then discard:
c
eval
l_onoff = 1
c
eval
l_linger = 120
c
callp
setsockopt(sock: SOL_SOCKET: SO_LINGER:
c
ling: linglen)
C*** free up resources used by linger structure
c
dealloc(E)
ling
C*** Create a sockaddr_in structure
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
eval
p_sockaddr = bindto
c
c
c
c
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
INADDR_ANY
pePort
*ALLx’00’
C*** Bind socket to port
c
if
bind(sock: bindto: len) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Listen for a connection
c
if
listen(sock: 1) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Return newly set-up socket:
c
dealloc(E)
c
return
sock
bindto
84
Chapter 6. Handling many sockets at once using select()
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This copies data from one socket to another.
*
parms:
*
fromsock = socket to copy data from
*
tosock = socket to copy data to
*
*
returns: length of data copied, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P copysock
B
D copysock
PI
10I 0
D fromsock
10I 0 value
D tosock
10I 0 value
D buf
S
1024A
D rc
S
10I 0
c
eval
rc = recv(fromsock: %addr(buf): 1024: 0)
c
if
rc < 0
c
return
-1
c
endif
c
if
send(tosock: %addr(buf): rc: 0) < rc
c
return
-1
c
endif
c
return
rc
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
85
Chapter 6. Handling many sockets at once using select()
D wwMsgLen
D wwTheKey
S
S
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
Compile it by typing: CRTBNDRPG PROXY1 SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
To run it:
1.
On the AS/400, type: CALL PROXY1 PARM(4000 4001)
2.
On a PC, type: telnet as400 4000 Then leave that window open.
3.
Start a 2nd telnet client by typing: telnet as400 4001
4.
Whatever you type in the first telnet client should appear in the other one’s window, and vice-versa. If the client
that you’re using is a ’line-at-a-time’ client, you may have to press enter before the line shows up.
5.
When you tell one of the telnet clients to disconnect, the AS/400 program will end, and the other telnet client
should be disconnected.
(To disconnect using the Windows telnet client, click on the "Connect" menu, and choose "Disconnect". For
FreeBSD or Linux clients, press Ctrl-] and then type quit)
Lets try it with a web browser:
1.
On the AS/400, type: CALL PROXY1 PARM(4000 4001)
2.
On a PC, type: telnet as400 4000
Tip: If you’re using the Windows Telnet client, and you can’t read what you’re typing, you might try clicking
"Terminal", then "Preferences" and enabling "Local Echo".
3.
In a web browser, set the Location: to the URL: http://as400:4001/
4.
Back in the telnet client, you’ll see the web browser send a request for a web page. It’ll also send a bunch of
extra information about itself, such as the type of browser you’re using, what types of documents it accepts, etc.
5.
In the telnet client, type something like:
86
Chapter 6. Handling many sockets at once using select()
<H1>Hello Web Browser!</H1>
<H3>Nice subtitle, eh?</H3>
6.
Then tell the telnet client to disconnect. Take a look at what the web browser has displayed!
I’m sure you can see how this could be a very valuable tool for debugging programs and testing what your clients
and servers are doing!
6.5. Blocking vs. non-blocking sockets
So far in this chapter, you’ve seen that select() can be used to detect when data is available to read from a socket.
However, there are times when its useful to be able to call send(), recv(), connect(), accept(), etc without having to
wait for the result.
For example, let’s say that you’re writing a web browser. You try to connect to a web server, but the server isn’t
responding. When a user presses (or clicks) a stop button, you want the connect() API to stop trying to connect.
With what you’ve learned so far, that can’t be done. When you issue a call to connect(), your program doesn’t regain
control until either the connection is made, or an error occurs.
The solution to this problem is called "non-blocking sockets".
By default, TCP sockets are in "blocking" mode. For example, when you call recv() to read from a stream, control
isn’t returned to your program until at least one byte of data is read from the remote site. This process of waiting for
data to appear is referred to as "blocking". The same is true for the write() API, the connect() API, etc. When you run
them, the connection "blocks" until the operation is complete.
Its possible to set a descriptor so that it is placed in "non-blocking" mode. When placed in non-blocking mode, you
never wait for an operation to complete. This is an invaluable tool if you need to switch between many different
connected sockets, and want to ensure that none of them cause the program to "lock up."
If you call "recv()" in non-blocking mode, it will return any data that the system has in it’s read buffer for that socket.
But, it won’t wait for that data. If the read buffer is empty, the system will return from recv() immediately saying
"Operation Would Block!".
The same is true of the send() API. When you call send(), it puts the data into a buffer, and as it’s read by the remote
site, it’s removed from the buffer. If the buffer ever gets "full", the system will return the error ’Operation Would
Block" the next time you try to write to it.
Non-blocking sockets have a similar effect on the accept() API. When you call accept(), and there isn’t already a
client connecting to you, it will return ’Operation Would Block’, to tell you that it can’t complete the accept()
without waiting...
The connect() API is a little different. If you try to call connect() in non-blocking mode, and the API can’t connect
instantly, it will return the error code for ’Operation In Progress’. When you call connect() again, later, it may tell
you ’Operation Already In Progress’ to let you know that it’s still trying to connect, or it may give you a successful
return code, telling you that the connect has been made.
Going back to the "web browser" example, if you put the socket that was connecting to the web server into
non-blocking mode, you could then call connect(), print a message saying "connecting to host www.floofy.com..."
then maybe do something else, and them come back to connect() again. If connect() works the second time, you
might print "Host contacted, waiting for reply..." and then start calling send() and recv(). If the connect() is still
87
Chapter 6. Handling many sockets at once using select()
pending, you might check to see if the user has pressed a "abort" button, and if so, call close() to stop trying to
connect.
Non-blocking sockets can also be used in conjunction with the select() API. In fact, if you reach a point where you
actually WANT to wait for data on a socket that was previously marked as "non-blocking", you could simulate a
blocking recv() just by calling select() first, followed by recv().
The "non-blocking" mode is set by changing one of the socket’s "flags". The flags are a series of bits, each one
representing a different capability of the socket. So, to turn on non-blocking mode requires three steps:
1.
Call the fcntl() API to retrieve the socket descriptor’s current flag settings into a local variable.
2.
In our local variable, set the O_NONBLOCK (non-blocking) flag on. (being careful, of course, not to tamper
with the other flags)
3.
Call the fcntl() API to set the flags for the descriptor to the value in our local variable.
6.6. The fcntl() API call
The previous topic explained how non-blocking mode works. This topic describes the API call that can be used to
turn non-blocking mode on for a given socket. This API is capable of setting many other options as well, though they
are mostly options that relate to stream files, not sockets, and are not covered by this tutorial.
The fcntl (file control) API is used to retrieve or set the flags that are associrated with a socket or stream file. The
IBM manual page for the fcntl() API (when used on a socket) can be found here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/sfcntl.htm
The C-language prototype for the fcntl API looks like this:
int fcntl(int descriptor, int command, ...);
It starts out pretty simple. We know that the API is called fcntl, and we know that the return value is an integer. We
can tell that the first two parameters are both integers, as well. But what the heck is ’...’?
The ’...’ parameter is a special construct in C to denote that the number of parameters remaining is variable. It
furthermore means that the types of each of the remaining parameters can also vary. In short, in addition to the two
integers, we can pass any number of additional parameters to the fcntl() API, and they can be of any data type. Not
very helpful, is it?
Reading more of the manual page, however, we see that some of the values for the ’command’ parameter of fcntl()
accept an integer as the 3rd parameter. Other values for the ’command’ parameter don’t require a 3rd parameter at
all. Since we now know that the ’...’ can only symbolize either an integer or nothing, we can write the prototype in
RPG like this:
D fcntl
D
SocketDesc
D
Command
D
Arg
PR
10I
10I
10I
10I
0
0
0
0
ExtProc(’fcntl’)
Value
Value
Value Options(*NOPASS)
88
Chapter 6. Handling many sockets at once using select()
But, since fcntl can also be used for stream files (much like close() can) we will use some compiler directives to
prevent it being defined twice, should we want to use the socket APIs in the same program as the IFS (stream file)
APIs. So, we’ll do this:
D/if not defined(FCNTL_PROTOTYPE)
D fcntl
PR
10I
D
SocketDesc
10I
D
Command
10I
D
Arg
10I
D/define FCNTL_PROTOTYPE
D/endif
0
0
0
0
ExtProc(’fcntl’)
Value
Value
Value Options(*NOPASS)
And, of course, add that to the SOCKET_H header file that we’ve been working on.
In addition, we’ll need constants to define the possible values for the ’command’ parameter of fcntl(). For the sake of
being able to turn on the "non-blocking" flag, we’re only concerned with two constants. ’F_GETFL’ and ’F_SETFL’
which are used to get and set the status flags, respectively.
These can be added to our SOCKET_H header file by typing this:
** fcntl() commands
D F_GETFL
C
D F_SETFL
C
CONST(6)
CONST(7)
The O_NONBLOCK flag also needs a constant so that we don’t have to try to remember which flag is used for
blocking/non-blocking :) The page in the manual tells us that ’O_NONBLOCK’, ’O_NDELAY’ and ’FNDELAY’ all
do the same thing, so we’ll just define them all, like this:
** fcntl() flags
D O_NONBLOCK
D O_NDELAY
D FNDELAY
C
C
C
CONST(128)
CONST(128)
CONST(128)
So... all of that should now be in your SOCKET_H member. Of course, if you simply downloaded my copy of
SOCKET_H, it was already there :)
To call fcntl() to put a socket in non-blocking mode, you’ll do something like this:
***** retrieve flags
c
eval
flags = fcntl(sock: F_GETFL)
c
if
flags < 0
C* handle error "unable to retrieve descriptor status flags"
c
endif
***** turn non non-blocking flag:
c
eval
flags = flags + O_NONBLOCK
***** set flags
c
if
fcntl(sock: F_SETFL: flags) < 0
C* handle error "unable to set descriptor status flags"
c
endif
89
Chapter 6. Handling many sockets at once using select()
6.7. A multi-client example of our server
Well, all that discussion about blocking and non-blocking sockets has made me hungry. Wanna stop and get a bite to
eat? No?
This chapter will put everything we’ve learned to the test. We’re going to take the server program that we wrote in
Chapter 5, and re-write it so that it can handle many simultaneous clients.
From the client’s perspective, our server won’t have changed much. It’ll still ask for a name, and still say hello and
goodbye. However the internal functioning of our program will be very different.
This program will use non-blocking sockets and the select() API to switch between a whole array of connected
clients. By spending only a few milliseconds on each client, then switching to the next client and spending time
there, then switching again, etc, we can handle many clients at once.
However, this creates a number of challenges for us:
•
We have to have some way of keeping track of where our client has "left off", so that when we switch to another
client and come back, we’ll be able to pick things up and continue.
•
We won’t be able to sit and wait until a client has sent us a x’0A’ to terminate a line of text, because all the other
clients would have to wait as well. Instead, we’ll have to take whatever the client sends us and throw it into a
buffer. Scan that buffer for a x’0A’, and only act upon it after the x’0A’ has been received.
•
To prevent this program from taking up all of the available CPU time on the system, we’ll need to keep track of
when we’re "waiting" for data, and when we’re "sending" data so that if we don’t have anything to send, we can
just sit and wait and not hog the CPU.
•
In the midst of all of these things, we’ll need to watch for new connections from new clients as well as disconnects
from existing clients. We’ll have to make sure that we only check for input on connected descriptors, and ignore
the disconnected ones.
•
One nice thing about the server we wrote in Chapter 5 is that it’s very simple. We don’t send a lot of data, so
running out of space in the system’s write buffer isn’t really a concern. Consequently, we can just use the
WrLine() routine that we wrote in Chapter 4, and not worry about the fact that we’re using non-blocking sockets
with it... However, when we do a more complex server, we’ll have to take that into consideration as well...
So here it is... the fruit of all of our labors, thus far! Hope I’ve described everything well enough. I strongly suggest
that you "play with it", i.e. try to add features or write new things that it does, because you’ll get a much better idea
of what you’re doing that way.
File: SOCKTUT/QRPGLESRC, Member: SERVEREX3
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D/copy socktut/qrpglesrc,errno_h
90
Chapter 6. Handling many sockets at once using select()
*** Prototypes for externally called programs:
D Translate
D
peLength
D
peBuffer
D
peTable
PR
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
*** Prototypes for local subprocedures:
D die
D
peMsg
PR
D NewListener
D
pePort
D
peError
PR
10I 0
5U 0 value
256A
D NewClient
D
peServ
PR
10I 0
10I 0 value
D ReadClient
D
peClient
PR
10I 0
10I 0 value
D HandleClient
D
peClient
PR
10I 0
10I 0 value
D EndClient
D
peClient
PR
10I 0
10I 0 value
D GetLine
D
peClient
D
peLine
PR
10I 0
10I 0 value
256A
256A
const
*** Configuration
D MAXCLIENTS
C
CONST(100)
*** Global Variables:
D
D
D
D
D
D
D
D
D
D
Msg
to
tolen
serv
max
rc
C
readset
excpset
endpgm
S
S
S
S
S
S
S
S
S
S
256A
*
10I
10I
10I
10I
10I
1N
0
0
0
0
0
like(fdset)
like(fdset)
inz(*off)
*** Variables in the "client" data structure are kept
*** seperate for each connected client socket.
91
Chapter 6. Handling many sockets at once using select()
D Client
D
sock
D
wait
D
rdbuf
D
rdbuflen
D
state
D
line
DS
Occurs(MAXCLIENTS)
10I 0
1N
256A
10I 0
10I 0
256A
c
eval
*inlr = *on
c
exsr
Initialize
C*********************************************************
C* Main execution loop:
C*
C*
1) Make 2 descriptor sets. One to check which
C*
sockets have data to read, one to check which
C*
sockets have exceptional conditions pending.
C*
Each set will contain the listner socket, plus
C*
each connected client socket
C*
C*
2) Call select() to find out which descriptors need
C*
to be read from or have exceptions pending.
C*
We have a timeout value set here as well. It’s
C*
set to 1 minute if all sockets are waiting for
C*
user input, or 1/10 second if the sockets need
C*
us to write data to them. (the 1/10 second is
C*
just to keep this program from gobbling too
C*
much CPU time)
C*
C*
3) Check to see if a user told us to shut down, or
C*
if the job/subsystem/system has requested us to
C*
end the program.
C*
C*
4) If the listener socket ("server socket") has data
C*
to read, it means someone is trying to connect
C*
to us, so call the NewClient procedure.
C*
C*
5) Check each socket for incoming data and load into
C*
the appropriate read buffer.
C*
C*
6) Do the next "task" that each socket needs.
C*
(could be sending a line of text, or waiting
C*
for input, or disconnecting, etc)
C*********************************************************
c
dow
1 = 1
c
exsr
MakeDescSets
c
c
eval
rc = select(max+1: %addr(readset):
*NULL: %addr(excpset): to)
92
Chapter 6. Handling many sockets at once using select()
c
exsr
ChkShutDown
c
c
c
c
c
c
if
if
callp
endif
exsr
endif
rc > 0
FD_ISSET(serv: readset)
NewClient(serv)
c
exsr
c
enddo
CheckSockets
DoClients
C*===============================================================
C* Initialize some program vars & set up a server socket:
C*===============================================================
CSR
Initialize
begsr
C*-----------------------c
do
MAXCLIENTS
C
c
C
occur
client
c
eval
sock = -1
c
callp
EndClient(C)
c
enddo
c
c
c
eval
alloc
eval
tolen = %size(timeval)
tolen
to
p_timeval = to
C*********************************************************
C* Start listening to port 4000
C*********************************************************
c
eval
serv = NewListener(4000: Msg)
c
if
serv < 0
c
callp
die(Msg)
c
endif
C*-----------------------CSR
endsr
C*===============================================================
C* This sets up the "readset" and "excpset" descriptor sets
C* for use with select().
It also calculates the appropriate
C* timeout value, and the maximum descriptor number to check
C*===============================================================
CSR
MakeDescSets begsr
C*-----------------------c
callp
FD_ZERO(readset)
c
callp
FD_SET(serv: readset)
C* By default, set a 60 second timeout... but if one or more
C* of the client’s is not in ’wait’ mode, change that timeout
C* to only 100,000 microseconds (1/10 second)
93
Chapter 6. Handling many sockets at once using select()
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
c
C
eval
eval
eval
tv_sec = 60
tv_usec = 0
max = serv
do
occur
if
callp
if
eval
eval
endif
if
eval
endif
endif
enddo
MAXCLIENTS
C
client
sock <> -1
FD_SET(sock: readset)
not wait
tv_sec = 0
tv_usec = 100000
sock > max
max = sock
C* We can just copy excpset to readset... no point in going thru
C* all of that again :)
c
eval
excpset = readset
C*-----------------------CSR
endsr
C*===============================================================
C* Check for a ’shutdown’ condition. If shutdown was requested
C* tell all connected sockets, and then close them.
C*===============================================================
CSR
ChkShutDown
begsr
C*-----------------------c
shtdn
99
c
if
*in99 = *on
c
eval
endpgm = *On
c
endif
* Note that the ’endpgm’ flag can also be set by the
*
’HandleClient’ subprocedure, not just the code above...
c
if
c
do
c
C
occur
c
if
c
callp
c
c
callp
c
endif
c
enddo
c
callp
c
callp
c
return
c
endif
C*------------------------
endpgm = *on
MAXCLIENTS
C
client
sock <> -1
WrLine(sock: ’Sorry!
’down now!’)
EndClient(C)
We”re shutting ’ +
close(serv)
Die(’shut down requested...’)
94
Chapter 6. Handling many sockets at once using select()
CSR
endsr
C*===============================================================
C* This loads any data that has been sent by the various client
C* sockets into their respective read buffers, and also checks
C* for clients that may have disconnected:
C*===============================================================
CSR
CheckSockets begsr
C*-----------------------c
do
MAXCLIENTS
C
c
C
occur
client
c
if
sock <> -1
c
c
c
c
c
c
if
if
callp
callp
endif
endif
FD_ISSET(sock: readset)
ReadClient(C) < 0
EndClient(C)
FD_CLR(sock: excpset)
c
c
c
if
callp
endif
FD_ISSET(sock: excpset)
EndClient(C)
c
endif
c
enddo
C*-----------------------CSR
endsr
C*===============================================================
C* This finally gets down to "talking" to the client programs.
C* It switches between each connected client, and then sends
C* data or receives data as appropriate...
C*===============================================================
CSR
DoClients
begsr
C*-----------------------c
do
MAXCLIENTS
C
c
C
occur
client
c
if
sock <> -1
c
callp
HandleClient(C)
c
endif
c
enddo
C*-----------------------CSR
endsr
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Create a new TCP socket that’s listening to a port
*
95
Chapter 6. Handling many sockets at once using select()
*
parms:
*
pePort = port to listen to
*
peError = Error message (returned)
*
*
returns: socket descriptor upon success, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewListener
B
D NewListener
PI
10I 0
D
pePort
5U 0 value
D
peError
256A
D
D
D
D
D
D
D
sock
len
bindto
on
linglen
ling
flags
S
S
S
S
S
S
S
C*** Create a socket
c
eval
c
c
if
c
eval
c
return
c
endif
10I
10I
*
10I
10I
*
10I
0
0
0 inz(1)
0
0
sock = socket(AF_INET:SOCK_STREAM:
IPPROTO_IP)
sock < 0
peError = %str(strerror(errno))
-1
C*** Tell socket that we want to be able to re-use the server
C*** port without waiting for the MSL timeout:
c
callp
setsockopt(sock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C*** create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C*** tell socket to only linger for 2 minutes, then discard:
c
eval
l_onoff = 1
c
eval
l_linger = 120
c
callp
setsockopt(sock: SOL_SOCKET: SO_LINGER:
c
ling: linglen)
C*** free up resources used by linger structure
c
dealloc(E)
ling
C*** tell socket we don’t want blocking...
c
eval
flags = fcntl(sock: F_GETFL)
c
eval
flags = flags + O_NONBLOCK
c
if
fcntl(sock: F_SETFL: flags) < 0
c
eval
peError = %str(strerror(errno))
c
return
-1
c
endif
96
Chapter 6. Handling many sockets at once using select()
C*** Create a sockaddr_in structure
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
eval
p_sockaddr = bindto
c
c
c
c
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
INADDR_ANY
pePort
*ALLx’00’
C*** Bind socket to port
c
if
bind(sock: bindto: len) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Listen for a connection
c
if
listen(sock: MAXCLIENTS) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Return newly set-up socket:
c
dealloc(E)
c
return
sock
P
E
bindto
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This accepts a new client connection, and adds him to
*
the ’client’ data structure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewClient
B
D NewClient
PI
10I 0
D
peServ
10I 0 value
D
D
D
D
D
D
D
D
X
S
cl
flags
ling
connfrom
len
Msg
S
S
S
S
S
S
S
S
10I
10I
10I
10I
*
*
10I
52A
0
0
0
0
0
C*************************************************
C* See if there is an empty spot in the data
C* structure.
97
Chapter 6. Handling many sockets at once using select()
C*************************************************
c
eval
cl = 0
c
do
MAXCLIENTS
X
c
X
occur
Client
c
if
sock = -1
c
eval
cl = X
c
leave
c
endif
c
enddo
C*************************************************
C* Accept new connection
C*************************************************
c
eval
len = %size(sockaddr_in)
c
alloc
len
connfrom
c
c
c
c
eval
if
return
endif
c
dealloc(E)
S = accept(peServ: connfrom: len)
S < 0
-1
connfrom
C*************************************************
C* Turn off blocking & limit lingering
C*************************************************
c
eval
flags = fcntl(S: F_GETFL: 0)
c
eval
flags = flags + O_NONBLOCK
c
if
fcntl(S: F_SETFL: flags) < 0
c
eval
Msg = %str(strerror(errno))
c
dsply
Msg
c
return
-1
c
endif
c
c
c
c
c
c
c
c
eval
alloc
eval
eval
eval
callp
len = %size(linger)
len
ling
p_linger = ling
l_onoff = 1
l_linger = 120
setsockopt(S: SOL_SOCKET: SO_LINGER:
ling: len)
dealloc(E)
ling
C*************************************************
C* If we’ve already reached the maximum number
C* of connections, let client know and then
C* get rid of him
C*************************************************
c
if
cl = 0
c
callp
wrline(S: ’Maximum number of connect’ +
c
’tions has been reached!’)
c
callp
close(s)
c
return
-1
98
Chapter 6. Handling many sockets at once using select()
c
endif
C*************************************************
C* Add client into the structure
C*************************************************
c
cl
occur
client
c
eval
sock = S
c
return
0
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* If there is data to be read from a Client’s socket, add it
*
to the client’s buffer, here...
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P ReadClient
B
D ReadClient
PI
10I 0
D
peClient
10I 0 value
D
D
D
D
c
left
p_read
err
len
S
S
S
S
peClient
10I 0
*
10I 0
10I 0
occur
client
c
c
eval
eval
left = %size(rdbuf) - rdbuflen
p_read = %addr(rdbuf) + rdbuflen
c
c
c
c
c
c
c
c
c
eval
if
eval
if
return
else
return
endif
endif
len
len
err
err
0
c
c
P
eval
return
rdbuflen = rdbuflen + len
len
=
<
=
=
recv(sock: p_read: left: 0)
0
errno
EWOULDBLOCK
-1
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This disconnects a client and cleans up his spot in the
*
client data structure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P EndClient
B
D EndClient
PI
10I 0
D
peClient
10I 0 value
c
peClient
occur
client
c
if
sock >= 0
99
Chapter 6. Handling many sockets at once using select()
c
c
c
c
c
c
c
c
c
P
callp
endif
eval
eval
eval
eval
eval
eval
return
close(sock)
sock = -1
wait = *off
rdbuf = *Blanks
rdbuflen = 0
state = 0
line = *blanks
0
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* As we’re switching between each different client, this
* routine is called to handle whatever the next ’step’ is
* for a given client.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P HandleClient
B
D HandleClient
PI
10I 0
D
peClient
10I 0 value
c
peClient
occur
client
c
c
c
c
c
select
when
callp
c
c
c
c
c
c
when
eval
if
eval
eval
endif
state = 1
wait = *on
GetLine(peClient: line) > 0
wait = *off
state = 2
c
c
c
c
c
c
c
c
when
if
eval
eval
else
callp
eval
endif
state = 2
%trim(line) = ’quit’
endpgm = *on
state = 4
c
c
c
when
callp
eval
state = 3
WrLine(sock: ’Goodbye ’ + %trim(line))
state = 4
c
c
c
when
callp
endsl
state = 4
EndClient(peClient)
eval
state = 0
WrLine(sock: ’Please enter your name’ +
’ now!’)
state = 1
WrLine(sock: ’Hello ’ + %trim(line))
state = 3
100
Chapter 6. Handling many sockets at once using select()
c
P
return
0
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This removes one line of data from a client’s read buffer
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P GetLine
B
D GetLine
PI
10I 0
D
peClient
10I 0 value
D
peLine
256A
D pos
S
10I 0
C*** Load correct client:
c
peClient
occur
c
if
c
return
c
endif
client
rdbuflen < 1
0
C*** Look for an end-of-line character:
c
eval
pos = %scan(x’0A’: rdbuf)
c
if
pos < 1 or pos > rdbuflen
c
return
0
c
endif
C*** Add line to peLine variable, and remove from rdbuf:
c
eval
peLine = %subst(rdbuf:1:pos-1)
c
c
c
c
c
if
eval
else
eval
endif
pos < %size(rdbuf)
rdbuf = %subst(rdBuf:pos+1)
c
eval
rdbuflen = rdbuflen - pos
rdbuf = *blanks
C*** If CR character found, remove that too...
c
eval
pos = pos - 1
c
if
%subst(peLine:pos:1) = x’0D’
c
eval
peLine = %subst(peLine:1:pos-1)
c
eval
pos = pos - 1
c
endif
C*** Convert to EBCDIC:
c
if
c
callp
c
endif
pos > 0
Translate(pos: peLine: ’QTCPEBC’)
C*** return length of line:
c
return
P
E
pos
101
Chapter 6. Handling many sockets at once using select()
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
102
Chapter 6. Handling many sockets at once using select()
6.8. Adding some refinements to our approach
Not too bad for our first multi-client server program, is it? There’s some room for improvement however. This topic
will discuss what we could’ve done better, and what to do about it.
First of all, we’re sending any outgoing data with the WrLine procedure. WrLine() is designed to be used with
blocking sockets. As I mentioned before, this doesn’t really cause problems in our sample program because there’s
only a small amount of data to send. The chances of us somehow filling up the system’s TCP buffer is very remote,
so we probably won’t ever notice the problem.
But in a real application? One where files are being sent, or data is being typed by another user? We’d be sure to have
problems! As soon as the TCP buffer is full, data would simply be cut off.
Speaking of buffers, the read buffer has a problem too. Right now, we add to the read buffer. That’s not a problem,
however, we only take data out of the read buffer when a x’0A’ (end-of-line) character is found. That means that if a
client should fill up his read buffer without having an end of line character, it’ll simply STOP receiving data from
that client!
The ’wait’ indicator is a neat idea, but it wouldn’t really be necessary if we handled writes correctly. If we handled
them correctly, data would go into a write buffer, and be written as soon as writing to the socket was feasible. If we
did that, there’d be little point to having a ’wait’ indicator, we’d just always do the full 60 second timeout.
Finally, why are we setting the ’readset’ for a client every single time we do a select()? We should only set descriptor
on if we have space in our read buffer. That way, we don’t have to go through the motions of trying to do a recv()
when we know there isn’t anyplace to put the data.
I suggest that you try to solve all of these problems yourself. It’ll be a good challenge for you -- trying to implement
a write buffer, and handle a full read buffer, etc.
If you get stuck, take a look at my solution to the problem, which is online at this address:
http://www.scottklement.com/rpg/socktut/qrpglesrc.serverex4
6.9. A Chat Application
In the last topic, you improved upon the ’example server’ that we’ve been using since the start of the server portion
of this tutorial. Yes, it’s a nice, simple, example of writing a server.
But how practical is all of this knowledge you’ve picked up? Can you do anything truly useful with it?
Okay. Lets modify that last example, the ’simple example server’ program and turn it into a simple ’chat room’.
Here’s what we want to do:
1.
Each message going from the server to the client should be prefixed by a unique ’message number’. This makes
it much easier for a client program to understand the purpose for each message we send.
2.
When a client program connects, the first message (message number 100) will ask the client for his name.
3.
If the name is a good one, the server will respond with message number 101. "Login Accepted"
4.
From this point on, each line that the client types will be sent to all of the other clients that are logged in. Each of
these messages will be prefixed by the name of the person using the client which sent the message.
This will involve surprisingly few changes to the program that we created in the previous topic.
103
Chapter 6. Handling many sockets at once using select()
Here’s our ’chat server’:
File: SOCKTUT/QRPGLESRC, Member: SERVEREX5
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D/copy socktut/qrpglesrc,errno_h
*** Prototypes for externally called programs:
D Translate
D
peLength
D
peBuffer
D
peTable
PR
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
*** Prototypes for local subprocedures:
D die
D
peMsg
PR
D NewListener
D
pePort
D
peError
PR
10I 0
5U 0 value
256A
D NewClient
D
peServ
PR
10I 0
10I 0 value
D ReadClient
D
peClient
PR
10I 0
10I 0 value
D WriteClient
D
peClient
PR
10I 0
10I 0 value
D HandleClient
D
peClient
PR
10I 0
10I 0 value
D EndClient
D
peClient
PR
10I 0
10I 0 value
D GetLine
D
peClient
D
peLine
PR
10I 0
10I 0 value
256A
D PutLine
D
peClient
D
peLine
PR
10I 0
10I 0 value
256A
const
256A
const
*** Configuration
104
Chapter 6. Handling many sockets at once using select()
D MAXCLIENTS
C
CONST(100)
*** Global Variables:
D
D
D
D
D
D
D
D
D
D
D
Msg
to
tolen
serv
max
rc
C
readset
excpset
writeset
endpgm
S
S
S
S
S
S
S
S
S
S
S
256A
*
10I
10I
10I
10I
10I
0
0
0
0
0
like(fdset)
like(fdset)
like(fdset)
inz(*off)
1N
*** Variables in the "client" data structure are kept
*** seperate for each connected client socket.
D Client
D
sock
D
rdbuf
D
rdbuflen
D
state
D
wrbuf
D
wrbuflen
D
name
DS
Occurs(MAXCLIENTS)
10I
256A
10I
10I
2048A
10I
20A
0
0
0
0
c
eval
*inlr = *on
c
exsr
Initialize
C*********************************************************
C* Main execution loop:
C*
C*
1) Make read/write/exception descriptor sets.
C*
and figure out the timeout value for select()
C*
C*
2) Call select() to find out which descriptors need
C*
data to be written or read, and also to find
C*
any exceptions to handle.
C*
C*
3) Check to see if a user told us to shut down, or
C*
if the job/subsystem/system has requested us to
C*
end the program.
C*
C*
4) If the listener socket ("server socket") has data
C*
to read, it means someone is trying to connect
C*
to us, so call the NewClient procedure.
C*
C*
5) Check each socket for incoming data and load into
105
Chapter 6. Handling many sockets at once using select()
C*
the appropriate read buffer.
C*
C*
6) Check each client for outgoing data and write into
C*
the appropriate socket.
C*
C*
7) Do the next "task" that each socket needs.
C*
(could be sending a line of text, or waiting
C*
for input, or disconnecting, etc)
C*********************************************************
c
dow
1 = 1
c
exsr
MakeDescSets
c
c
eval
rc = select(max+1: %addr(readset):
%addr(writeset): %addr(excpset): to)
c
exsr
ChkShutDown
c
c
c
c
c
c
if
if
callp
endif
exsr
endif
rc > 0
FD_ISSET(serv: readset)
NewClient(serv)
c
exsr
c
enddo
CheckSockets
DoClients
C*===============================================================
C* Initialize some program vars & set up a server socket:
C*===============================================================
CSR
Initialize
begsr
C*-----------------------c
do
MAXCLIENTS
C
c
C
occur
client
c
eval
sock = -1
c
callp
EndClient(C)
c
enddo
c
c
c
eval
alloc
eval
tolen = %size(timeval)
tolen
to
p_timeval = to
C*********************************************************
C* Start listening to port 4000
C*********************************************************
c
eval
serv = NewListener(4000: Msg)
c
if
serv < 0
c
callp
die(Msg)
c
endif
C*------------------------
106
Chapter 6. Handling many sockets at once using select()
CSR
endsr
C*===============================================================
C* This makes the descriptor sets:
C*
readset -- includes the ’server’ (listener) socket plus
C*
any clients that still have space in their read buffer
C*
writeset -- includes any clients that have data in their
C*
write buffer.
C*
excpset -- includes all socket descriptors, since all of
C*
them should be checked for exceptional conditions
C*===============================================================
CSR
MakeDescSets begsr
C*-----------------------c
callp
FD_ZERO(writeset)
c
callp
FD_ZERO(readset)
c
callp
FD_ZERO(excpset)
c
c
callp
callp
FD_SET(serv: readset)
FD_SET(serv: excpset)
C* the 60 second timeout is just so that we can check for
C* system shutdown periodically.
c
eval
tv_sec = 60
c
eval
tv_usec = 0
c
eval
max = serv
c
do
c
C
occur
c
if
c
callp
c
if
c
callp
c
endif
c
if
c
callp
c
endif
c
if
c
eval
c
endif
c
endif
c
enddo
C*-----------------------CSR
endsr
MAXCLIENTS
C
client
sock <> -1
FD_SET(sock: excpset)
rdbuflen < %size(rdbuf)
FD_SET(sock: readset)
wrbuflen > 0
FD_SET(sock: writeset)
sock > max
max = sock
C*===============================================================
C* Check for a ’shutdown’ condition. If shutdown was requested
C* tell all connected sockets, and then close them.
C*===============================================================
CSR
ChkShutDown
begsr
C*-----------------------c
shtdn
99
107
Chapter 6. Handling many sockets at once using select()
c
c
c
if
eval
endif
*in99 = *on
endpgm = *On
* Note that the ’endpgm’ flag can also be set by the
*
’HandleClient’ subprocedure, not just the code above...
c
if
c
do
c
C
occur
c
if
c
callp
c
c
callp
c
endif
c
enddo
c
callp
c
callp
c
return
c
endif
C*-----------------------CSR
endsr
endpgm = *on
MAXCLIENTS
C
client
sock <> -1
WrLine(sock: ’902 Sorry!
’shutting down now!’)
EndClient(C)
We”re ’ +
close(serv)
Die(’shut down requested...’)
C*===============================================================
C* This reads any data that’s waiting to be read from each
C*
socket, and writes any data that’s waiting to be written.
C*
C* Also disconnects any socket that returns an error, or has
C* and exceptional condition pending.
C*===============================================================
CSR
CheckSockets begsr
C*-----------------------c
do
MAXCLIENTS
C
c
C
occur
client
c
if
sock <> -1
c
c
c
c
c
c
c
if
if
callp
callp
callp
endif
endif
FD_ISSET(sock: readset)
ReadClient(C) < 0
EndClient(C)
FD_CLR(sock: excpset)
FD_CLR(sock: writeset)
c
c
c
c
c
c
if
if
callp
callp
callp
endif
FD_ISSET(sock: writeset)
WriteClient(C) < 0
EndClient(C)
FD_CLR(sock: excpset)
FD_CLR(sock: writeset)
108
Chapter 6. Handling many sockets at once using select()
c
endif
c
c
c
if
callp
endif
c
endif
FD_ISSET(sock: excpset)
EndClient(C)
c
enddo
C*-----------------------CSR
endsr
C*===============================================================
C* This finally gets down to "talking" to the client programs.
C* It switches between each connected client, and then sends
C* data or receives data as appropriate...
C*===============================================================
CSR
DoClients
begsr
C*-----------------------c
do
MAXCLIENTS
C
c
C
occur
client
c
if
sock <> -1
c
callp
HandleClient(C)
c
endif
c
enddo
C*-----------------------CSR
endsr
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Create a new TCP socket that’s listening to a port
*
*
parms:
*
pePort = port to listen to
*
peError = Error message (returned)
*
*
returns: socket descriptor upon success, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewListener
B
D NewListener
PI
10I 0
D
pePort
5U 0 value
D
peError
256A
D
D
D
D
D
D
D
sock
len
bindto
on
linglen
ling
flags
S
S
S
S
S
S
S
10I
10I
*
10I
10I
*
10I
0
0
0 inz(1)
0
0
C*** Create a socket
109
Chapter 6. Handling many sockets at once using select()
c
c
c
c
c
c
eval
if
eval
return
endif
sock = socket(AF_INET:SOCK_STREAM:
IPPROTO_IP)
sock < 0
peError = %str(strerror(errno))
-1
C*** Tell socket that we want to be able to re-use the server
C*** port without waiting for the MSL timeout:
c
callp
setsockopt(sock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C*** create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C*** tell socket to only linger for 2 minutes, then discard:
c
eval
l_onoff = 1
c
eval
l_linger = 120
c
callp
setsockopt(sock: SOL_SOCKET: SO_LINGER:
c
ling: linglen)
C*** free up resources used by linger structure
c
dealloc(E)
ling
C*** tell socket we don’t want blocking...
c
eval
flags = fcntl(sock: F_GETFL)
c
eval
flags = flags + O_NONBLOCK
c
if
fcntl(sock: F_SETFL: flags) < 0
c
eval
peError = %str(strerror(errno))
c
return
-1
c
endif
C*** Create a sockaddr_in structure
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
eval
p_sockaddr = bindto
c
c
c
c
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
INADDR_ANY
pePort
*ALLx’00’
C*** Bind socket to port
c
if
bind(sock: bindto: len) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Listen for a connection
110
Chapter 6. Handling many sockets at once using select()
if
listen(sock: MAXCLIENTS) < 0
eval
peError = %str(strerror(errno))
callp
close(sock)
dealloc(E)
bindto
return
-1
endif
c
c
c
c
c
c
C*** Return newly set-up socket:
c
dealloc(E)
c
return
sock
P
E
bindto
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This accepts a new client connection, and adds him to
*
the ’client’ data structure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewClient
B
D NewClient
PI
10I 0
D
peServ
10I 0 value
D
D
D
D
D
D
D
D
X
S
cl
flags
ling
connfrom
len
Msg
S
S
S
S
S
S
S
S
10I
10I
10I
10I
*
*
10I
52A
0
0
0
0
0
C*************************************************
C* See if there is an empty spot in the data
C* structure.
C*************************************************
c
eval
cl = 0
c
do
MAXCLIENTS
X
c
X
occur
Client
c
if
sock = -1
c
eval
cl = X
c
leave
c
endif
c
enddo
C*************************************************
C* Accept new connection
C*************************************************
c
eval
len = %size(sockaddr_in)
c
alloc
len
connfrom
c
c
c
c
eval
if
return
endif
S = accept(peServ: connfrom: len)
S < 0
-1
111
Chapter 6. Handling many sockets at once using select()
c
dealloc(E)
connfrom
C*************************************************
C* Turn off blocking & limit lingering
C*************************************************
c
eval
flags = fcntl(S: F_GETFL: 0)
c
eval
flags = flags + O_NONBLOCK
c
if
fcntl(S: F_SETFL: flags) < 0
c
eval
Msg = %str(strerror(errno))
c
dsply
Msg
c
return
-1
c
endif
c
c
c
c
c
c
c
c
eval
alloc
eval
eval
eval
callp
len = %size(linger)
len
ling
p_linger = ling
l_onoff = 1
l_linger = 120
setsockopt(S: SOL_SOCKET: SO_LINGER:
ling: len)
dealloc(E)
ling
C*************************************************
C* If we’ve already reached the maximum number
C* of connections, let client know and then
C* get rid of him
C*************************************************
c
if
cl = 0
c
callp
wrline(S: ’901 Maximum number of ’ +
c
’connections has been reached!’)
c
callp
close(s)
c
return
-1
c
endif
C*************************************************
C* Add client into the structure
C*************************************************
c
cl
occur
client
c
eval
sock = S
c
return
0
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* If there is data to be read from a Client’s socket, add it
*
to the client’s buffer, here...
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P ReadClient
B
D ReadClient
PI
10I 0
D
peClient
10I 0 value
D left
S
10I 0
112
Chapter 6. Handling many sockets at once using select()
D p_read
D err
D len
c
S
S
S
peClient
*
10I 0
10I 0
occur
client
c
c
eval
eval
left = %size(rdbuf) - rdbuflen
p_read = %addr(rdbuf) + rdbuflen
c
c
c
c
c
c
c
c
c
eval
if
eval
if
return
else
return
endif
endif
len
len
err
err
0
c
c
P
eval
return
rdbuflen = rdbuflen + len
len
=
<
=
=
recv(sock: p_read: left: 0)
0
errno
EWOULDBLOCK
-1
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* If a client’s socket is ready to write data to, and theres
*
data to write, go ahead and write it...
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P WriteClient
B
D WriteClient
PI
10I 0
D
peClient
10I 0 value
D len
c
c
c
c
S
occur
if
return
endif
client
wrbuflen < 1
0
c
eval
len = send(sock:%addr(wrbuf):wrbuflen:0)
c
c
c
c
if
eval
eval
endif
len > 0
wrbuf = %subst(wrbuf: len+1)
wrbuflen = wrbuflen - len
return
len
c
P
peClient
10I 0
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This disconnects a client and cleans up his spot in the
*
client data structure.
113
Chapter 6. Handling many sockets at once using select()
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P EndClient
B
D EndClient
PI
10I 0
D
peClient
10I 0 value
c
peClient
occur
client
c
if
sock >= 0
c
callp
close(sock)
c
endif
c
eval
sock = -1
c
eval
rdbuf = *Blanks
c
eval
rdbuflen = 0
c
eval
state = 0
c
eval
wrbuflen = 0
c
eval
wrbuf = *Blanks
c
return
0
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* As we’re switching between each different client, this
* routine is called to handle whatever the next ’step’ is
* for a given client.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P HandleClient
B
D HandleClient
PI
10I 0
D
peClient
10I 0 value
D X
D from
D msg
c
S
S
S
peClient
10I 0
24A
256A
occur
client
c
c
c
c
c
select
when
callp
eval
state = 0
PutLine(peClient: ’100 Please enter ’ +
’your name now!’)
state = 1
c
c
when
if
state = 1
GetLine(peClient: msg) > 0
c
c
c
c
c
c
c
c
c
if
eval
callp
eval
else
eval
callp
eval
endif
%trim(msg) = ’quit’
endpgm = *on
PutLine(peClient: ’ ’)
state = 3
c
endif
name = %trim(msg)
PutLine(peClient: ’101 Login Accepted.’)
state = 2
114
Chapter 6. Handling many sockets at once using select()
c
c
c
C*
c
c
c
c
c
c
c
c
when
if
eval
X
copy message to each client:
do
MAXCLIENTS
X
occur
client
if
sock <> -1 and state > 1
callp
PutLine(X: %trimr(from) + ’: ’ + msg)
endif
enddo
peClient
c
c
c
c
P
state = 2
GetLine(peClient: msg) > 0
from = ’200 ’ + name
occur
endif
client
when
callp
endsl
state = 3
EndClient(peClient)
return
0
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This removes one line of data from a client’s read buffer
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P GetLine
B
D GetLine
PI
10I 0
D
peClient
10I 0 value
D
peLine
256A
D pos
S
C*** Load correct client:
c
peClient
occur
c
if
c
return
c
endif
10I 0
client
rdbuflen < 1
0
C*** Look for an end-of-line character:
c
eval
pos = %scan(x’0A’: rdbuf)
C*** If buffer is completely full, take the whole thing
C***
even if there is no end-of-line char.
c
if
pos<1 and rdbuflen>=%size(rdbuf)
c
eval
peLine = rdbuf
c
eval
pos = rdbuflen
c
eval
rdbuf = *blanks
c
eval
rdbuflen = 0
c
callp
Translate(pos: peLine: ’QTCPEBC’)
c
return
pos
115
Chapter 6. Handling many sockets at once using select()
c
endif
C*** otherwise, only return something if end of line existed
c
if
pos < 1 or pos > rdbuflen
c
return
0
c
endif
C*** Add line to peLine variable, and remove from rdbuf:
c
eval
peLine = %subst(rdbuf:1:pos-1)
c
c
c
c
c
if
eval
else
eval
endif
pos < %size(rdbuf)
rdbuf = %subst(rdBuf:pos+1)
c
eval
rdbuflen = rdbuflen - pos
rdbuf = *blanks
C*** If CR character found, remove that too...
c
eval
pos = pos - 1
c
if
%subst(peLine:pos:1) = x’0D’
c
eval
peLine = %subst(peLine:1:pos-1)
c
eval
pos = pos - 1
c
endif
C*** Convert to EBCDIC:
c
if
c
callp
c
endif
pos > 0
Translate(pos: peLine: ’QTCPEBC’)
C*** return length of line:
c
return
P
E
pos
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Add a line of text onto the end of a client’s write buffer
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P PutLine
B
D PutLine
PI
10I 0
D
peClient
10I 0 value
D
peLine
256A
const
D wkLine
D saveme
D len
c
c
peClient
S
S
S
258A
10I 0
10I 0
occur
occur
client
client
saveme
C* Add CRLF & calculate length & translate to ASCII
c
eval
wkLine = %trimr(peLine) + x’0D25’
c
eval
len = %len(%trimr(wkLine))
116
Chapter 6. Handling many sockets at once using select()
c
callp
Translate(len: wkLine: ’QTCPASC’)
C* make sure we don’t overflow buffer
c
if
(wrbuflen+len) > %size(wrbuf)
c
eval
len = %size(wrbuf) - wrbuflen
c
endif
c
if
len < 1
c
saveme
occur
client
c
return
0
c
endif
C* add data onto end of buffer
c
eval
%subst(wrbuf:wrbuflen+1) =
c
%subst(wkLine:1:len)
c
eval
wrbuflen = wrbuflen + len
c
saveme
occur
client
c
return
len
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
c
c
c
c
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
117
Chapter 6. Handling many sockets at once using select()
c
c
c
c
P
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
118
Chapter 7. Handling many sockets by spawning
jobs
Written by Scott Klement.
7.1. Overview of job spawning approach
The last chapter explained in detail how to write server programs that could handle many simultaneous clients, all
being served by one program in one job.
As I explained in the introduction to that chapter, that type of coding has it’s advantages. It’s main advantages are
that the various connected clients can share data very easily through the server, and that it uses less system resources
than some of the other methods of handling multiple clients.
It has some big disadvantages too! The coding can get very complicated! If something that one client is doing should
crash the program, it’ll crash all of the other clients as well! Everything in that server must run with the same user
profile, so all client’s have the same authority to system settings.
This chapter will discuss the opposite method. This method will use the SBMJOB command to "spawn" a new job
for every client that connects. It will then become the new job’s responsibility to send & receive data from that client.
Since the operating system will handle the multitasking for us, and the operating system is undoubtedly better at it
than our programs are, this will give better performance for very network-intensive applications.
It will however, make communications between the various connected clients extremely difficult. You’d have a very
hard time writing a ’chat room’ using this method -- even if you did get it to work, it would almost certainly be less
efficient.
The ’job spawning approach’ requires two (or more) programs in order to do it’s basic duty. The first program is the
’listener’ program. It listens and accepts new client connections, then submits a job to handle each client. The second
program is the ’server instance’ program, it handles all of the socket communications for a single client.
Here’s the basic pseudocode of the ’listener program’:
1.
A socket is created that listens for connections.
2.
accept() is called to receive a new connection.
3.
a new job is submitted to handle this new connection.
4.
We call the givedescriptor() API to tell the operating system that its okay for the new job to take the client’s
socket descriptor from us.
5.
We close() the client’s descriptor, we’re done with it.
6.
Go back to step #2.
Here’s the basic pseudocode of the ’server instance’ program:
1.
Communicate our job information back to the listener program so that it knows who to give a descriptor to.
2.
Call the takedescriptor() API to take the client’s socket descriptor from the listener program.
119
Chapter 7. Handling many sockets by spawning jobs
3.
send() and recv() data to/from the client application.
4.
close the socket
5.
end.
7.2. The takedescriptor() and givedescriptor() API calls
The givedescriptor() and takedescriptor() API calls are vital for implementing a server using the ’job spawning
approach.’ They allow a listener program to pass a descriptor to a server instance program.
At first glance, it sounds as if this should be easy. After all, a socket descriptor is just a number, right? And passing a
number from one program to another isn’t that hard, is it?
Well, unfortunately it’s not that easy. First of all, there’s security. Understandably, you don’t want one job to be able
to read data from a socket that another job has open! If that were allowed, I could write a simple loop to try every
descriptor on the system and peek into the affairs of every network job on the system! Yikes!
Secondly, each open descriptor is an attribute of the job it was opened in. Job A and Job B can both have a descriptor
#5. And they can be different things. The memory that the system uses to keep track of each descriptor is allocated to
each individual job. If it weren’t, then jobs could interfere with each other -- and possibly crash one another. While
this may be acceptable in Windows, it certainly is not in OS/400!
So, in order to pass a descriptor from one job to another, you MUST use the givedescriptor() and takedescriptor()
APIs.
The givedescriptor() API is documented in the IBM manual page at this location:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/gvsoc.htm
This manual lists the C language prototype for the givedescriptor() API as looking like this:
int givedescriptor(int descriptor, char *target_job);
Pretty straight forward, right? It’s called ’givedescriptor’, and it returns an integer. It has two parameters, the first one
is an integer, and the second one is a pointer.
To make the same prototype in RPG, we add this to our SOCKET_H member:
D givedescriptor
D
SockDesc
D
target_job
PR
10I 0 extproc(’givedescriptor’)
10I 0 VALUE
*
VALUE
The "usage notes" tell us that the "target_job" parameter is the ’internal job identifier’ of the job that we want to give
the descriptor to. And that we can look up this job identifier using the QUSRJOBI API.
The IBM manual page for the takedescriptor() API is located here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/tksoc.htm
And it tells us that the C language prototype for the takedescriptor() API looks like this:
int takedescriptor(char *source_job);
120
Chapter 7. Handling many sockets by spawning jobs
So, the API is called ’takedescriptor’. It returns an integer, and accepts one parameter, a pointer. This makes the RPG
prototype look like this:
D takedescriptor
D
source_job
PR
10I 0 extproc(’takedescriptor’)
*
VALUE
So add that prototype to your SOCKET_H member as well.
The usage notes for the takedescriptor() API tell us that the source_job is also the ’internal job identifier’ which we
can retrieve using the QUSRJOBI API. It also tells us that if we set the source_job parameter to NULL, it’ll take any
descriptor that the system tries to pass to it.
You can call these APIs in RPG by doing something like this:
In the "listener" program:
c
if
c*** handle failure
c
endif
givedescriptor(sock: target_int_id) < 0
And in the "server instance" program:
c
eval
sock = takedescriptor(*NULL)
The tricky part, of course, is getting the ’target_int_id’, which we shall discuss in the next section.
7.3. Retrieving the internal job id
The takedescriptor() and getdescriptor() APIs require that you know the internal job ID of the job that you wish to
pass a descriptor to. This topic discusses the QUSRJOBI (Retrieve Job Information) API which we will use to
determine the Internal Job Identifier.
The QUSRJOBI API is an ’Work Management API’ (most of the other APIs we’ve worked with have been
’Unix-Type APIs’, which makes it somewhat different.
The IBM manual page for QUSRJOBI is located here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/qusrjobi.htm
The prototype listed on this page isn’t in C format, but rather is in a somewhat ’language-neutral’ format. It looks
like this:
Parm# Description
Usage
Data Type
Output
Char (*)
1
Receiver variable
2
Length of receiver variable
Input
Binary (4)
3
Format name
Input
Char (8)
4
Qualified job name
Input
Char (26)
5
Internal job identifier
Input
Char (16)
Optional parameter:
121
Chapter 7. Handling many sockets by spawning jobs
Parm# Description
6
Error code
Usage
I/O
Data Type
Char (*)
QUSRJOBI is actually a program on the system that we need to call (as opposed to a procedure in a service program)
The 2nd column of the parameter group listing above tells us whether the parameter is used for "Output" from the
API, "Input" to the API, or "I/O" for both input and output.
The Last column explains the data type of the variable used, and that data type’s length. A (*) under length indicates
that the length can vary, otherwise, the length is the number of bytes to pass.
Since this is a "common gotcha" for people who are calling APIs, I’ll say it again. The length is the number of bytes
to pass . When it says "Binary(4)", this is NOT the RPG ’4B 0’ data type. In RPG, ’4B 0’ means a 4 digit binary
number. But, a 4-digit binary number is only 2 bytes long! Instead, a ’9B 0’ should be passed, or much better yet, a
’10I 0’ should be passed. In general, the RPG ’I’ data type works better for anything called ’binary’ than the legacy
’B’ data type.
So, the prototype for the QUSRJOBI API will look like this:
D RtvJobInf
D
RcvVar
D
RcvVarLen
D
Format
D
JobName
D
IntJobID
D
ErrorCode
PR
ExtPgm(’QUSRJOBI’)
32766A
options(*varsize)
10I 0 CONST
8A
CONST
26A
CONST
16A
CONST
32766A
options(*varsize)
Note: Each parameter that’s to be ’Input’ only to the API is marked as ’CONST’. This is always a good idea, as it
protects your program from accidental changes, helps to document the usage of the API, and allows you to
substitute expressions for the parameters.
The parameters that can vary in size we used the "options(*varsize)" keyword with. This allows us to call the API
using different sized variables for each call.
You’ll note that QUSRJOBI return’s all of it’s data in the ’RcvVar’ parm, so we need an idea of what format the data
will be in. In fact, most APIs can return data in more than one format, and this format is determined by the
’FORMAT’ parameter.
Reading more of the IBM manual page tells us that there are many different formats. Since we only need one that
returns the internal job id, we can use any of the formats that it allows. The shortest, and best performing format is
called ’JOBI0100’, so why don’t we use that?
The manual tells us that the format of the returned data for QUSRJOBI looks like this:
Offset
Dec
Hex
Type
Field
0
0
BINARY(4)
Number of bytes returned
4
4
BINARY(4)
Number of bytes available
8
8
CHAR(10)
Job name
122
Chapter 7. Handling many sockets by spawning jobs
Offset
Dec
18
Hex
12
Type
CHAR(10)
Field
User name
28
1C
CHAR(6)
Job number
34
22
CHAR(16)
Internal job identifier
50
32
CHAR(10)
Job status
60
3C
CHAR(1)
Job type
61
3D
CHAR(1)
Job subtype
62
3E
CHAR(2)
Reserved
64
40
BINARY(4)
Run priority
68
44
BINARY(4)
Time slice
72
48
BINARY(4)
Default wait
76
4C
CHAR(10)
Purge
So, we’ll make a data structure to use as our receiver variable. That’ll help us by separating all of the fields listed
above when we get our returned data from the API.
Here’s what this data structure should look like:
D dsJobI0100
D
JobI_ByteRtn
D
JobI_ByteAvl
D
JobI_JobName
D
JobI_UserID
D
JobI_JobNbr
D
JobI_IntJob
D
JobI_Status
D
JobI_Type
D
JobI_SbType
D
JobI_Reserv1
D
JobI_RunPty
D
JobI_TimeSlc
D
JobI_DftWait
D
JobI_Purge
DS
10I
10I
10A
10A
6A
16A
10A
1A
1A
2A
10I
10I
10I
10A
0
0
0
0
0
And, finally, the ’Error Code’ data structure. This is used in a lot of different APIs, and is a convienient way to get
returned error info.
It looks like this:
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
123
Chapter 7. Handling many sockets by spawning jobs
Since this prototype & data struct definition don’t (directly) have anything to do with sockets, I decided to make a
new header file for this API, called ’JOBINFO_H’
If you want to download my copy of JOBINFO_H, you can get it here:
http://www.scottklement.com/rpg/socktut/qrpglesrc.jobinfo_h
(http://www.scottklement.com/rpg/socktut/qrpglesrc.jobinfo_h)
You call the QUSRJOBI API like this:
C
callp
RtvJobInf(dsJOBI0100: %size(dsJOBI0100):
C
’JOBI0100’: ’*’: *BLANKS: dsEC)
c
if
dsECBytesA > 0
C** Error occurred.
Error message number is in dsECMsgID
C
endif
7.4. Communicating the job information
The QUSRJOBI API can be used to give us the Internal Job ID that the givedescriptor() API requires. But, before we
can call the QUSRJOBI API, we need to know the JobName/Userid/JobNbr of the server instance job.
This leads to a problem. If we haven’t submitted the job yet, how can we possibly know it’s job number? And even
after we’ve submitted it, how do we find out the number?
The easiest way is to submit the server instance job, and then have it look up it’s own job information. Once it has
that information, it should communicate it back to the listener program, so that givedescriptor() can be called.
The easiest way for the server instance to communicate back to the listener program is by using a data queue.
The data queue API’s are described in the Object APIs manual. You can find them online at this URL:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/obj2.htm
In our situation, we are interested in calling two of these APIs, they are the ’Receive From Data Queue’ API, and the
’Send to a Data Queue’ API. (QRCVDTAQ an QSNDDTAQ, respectively)
Here are the parameters listed in the manual for the QSNDDTAQ API:
Required Parameter Group:
1
Data queue name
Input
Char(10)
2
Library name
Input
Char(10)
3
Length of data
Input
Packed(5,0)
4
Data
Input
Char(*)
Optional Parameter Group 1:
5
Length of key data
Input
Packed(3,0)
6
Key data
Input
Char(*)
7
Asynchronous request
Input
Char(10)
Optional Parameter Group 2:
The RPG prototype for QSNDDTAQ looks like this:
124
Chapter 7. Handling many sockets by spawning jobs
D SndDtaQ
D dtaqname
D dtaqlib
D dtaqlen
D data
D keylen
D keydata
D asyncreq
PR
ExtPgm(’QSNDDTAQ’)
10A
const
10A
const
5P 0 const
32766A
const options(*varsize)
3P 0 const options(*nopass)
32766A
const options(*varsize: *nopass)
10A
const options(*nopass)
Here are the parameters listed in the manual for the QRCVDTAQ API:
Required Parameter Group:
1
Data queue name
Input
Char(10)
2
Library name
Input
Char(10)
3
Length of data
Output
Packed(5,0)
4
Data
Output
Char(*)
5
Wait time
Input
Packed(5,0)
Optional Parameter Group 1:
6
Key order
Input
Char(2)
7
Length of key data
Input
Packed(3,0)
8
Key data
9
Length of sender information
I/O
10
Sender information
Input
Char(*)
Packed(3,0)
Output
Char(*)
Optional Parameter Group 2:
11
Remove message
Input
Char(10)
12
Size of data receiver
Input
Packed(5,0)
13
Error code
I/O
Char(*)
So, an RPG prototype for the QRCVDTAQ API looks like this:
D RcvDtaQ
D
DtaqName
D
DtaqLib
D
DtaqLen
D
Data
D
WaitTime
D
KeyOrder
D
KeyLen
D
KeyData
D
SenderLen
D
SenderInfo
D
RmvMsg
D
RcvVarSize
D
ErrorCode
PR
10A
10A
5P
32766A
5P
2A
3P
32766A
3P
32766A
10A
5P
32766A
ExtPgm(’QRCVDTAQ’)
const
const
0
options(*varsize)
0 const
const options(*nopass)
0 const options(*nopass)
options(*varsize: *nopass)
0 const options(*nopass)
options(*varsize: *nopass)
const options(*nopass)
0 const options(*nopass)
options(*varsize: *nopass)
125
Chapter 7. Handling many sockets by spawning jobs
Each time one of our server instances starts, it will call QUSRJOBI to look up it’s internal job ID. It will then send it’s
internal job ID, along with it’s job name, userid, and job number to the data queue by calling QSNDDTAQ like this:
D dsJobInfo
D
MsgType
D
JobName
D
JobUser
D
JobNbr
D
InternalID
c
c
DS
1
11
21
31
65
callp
10A
20A
30A
36A
80A
SndDtaQ(’SVREX6DQ’: ’SOCKTUT’: 80:
dsJobInfo)
Note: We’re making the data queue length be 80, and including a ’MsgType’ field at the start in case we ever
want to make this data queue compatible with those used by display files, icf files, etc.
After the Listener program has submitted its next server instance pgm, it will read an entry off of the data queue to
find out the job info it needs to call givedescriptor(). It will call the QRCVDTAQ API like this:
c
callp
c
c
if
c*** timeout occurred.
c
else
C*** dsJobInfo is populated
c
endif
RcvDtaQ(’SVREX6DQ’: ’SOCKTUT’: Len:
dsJobInfo: 60)
Len < 80
Please add the prototypes shown in this topic, as well as the ’dsJobInfo’ data structure to the JOBINFO_H member
that you created earlier in this chapter.
Or, if you prefer, you can download my copy of JOBINFO_H here:
http://www.scottklement.com/rpg/socktut/qrpglesrc.jobinfo_h
We will use this JOBINFO_H member for all of our ’job spawning approach’ examples.
7.5. Our server as a multi-job server
So, we’ve learned what the "job spawning approach" is, and we’ve learned about the givedescriptor() and
takedescriptor() APIs, how to get the internal job id, and how to communicate that id back to the listener job.
Great. Time to put this knowledge to use!
Our "hello/goodbye" server that we originally wrote in Chapter 5 can now be made into a job spawning server. This
will involve the two programs that we’ve discussed in this chapter, the "listener" and the "server instance" programs.
I’ve decided to name them "svrex6l" (server example 6 -- listener) and "svrex6i" (server example 6 -- instance). And
here they are:
File SOCKTUT/QRPGLESRC, Member SVREX6L:
126
Chapter 7. Handling many sockets by spawning jobs
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy
D/copy
D/copy
D/copy
socktut/qrpglesrc,socket_h
socktut/qrpglesrc,sockutil_h
socktut/qrpglesrc,errno_h
socktut/qrpglesrc,jobinfo_h
*** prototypes for external calls
D Cmd
D
command
D
length
PR
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
*** Prototypes for local subprocedures:
D die
D
peMsg
PR
D NewListener
D
pePort
D
peError
PR
D KillEmAll
PR
256A
const
10I 0
5U 0 value
256A
*** local variables & constants
D MAXCLIENTS
C
D
D
D
D
D
D
D
S
S
S
S
S
S
S
c
svr
cli
msg
err
calen
clientaddr
jilen
CONST(256)
10I
10I
256A
10I
10I
*
5P
eval
0
0
0
0
0
*inlr = *on
C*************************************************
C* Clean up any previous instances of the dtaq
C*************************************************
c
callp(e) Cmd(’DLTDTAQ SOCKTUT/SVREX6DQ’: 200)
c
callp(e) Cmd(’CRTDTAQ DTAQ(SOCKTUT/SVREX6DQ) ’ +
c
’ MAXLEN(80) TEXT(”Data ’ +
c
’ queue for SVREX6L”)’: 200)
c
if
%error
c
callp
Die(’Unable to create data queue!’)
c
return
c
endif
127
Chapter 7. Handling many sockets by spawning jobs
C*************************************************
C* Start listening for connections on port 4000
C*************************************************
c
eval
svr = NewListener(4000: msg)
c
if
svr < 0
c
callp
die(msg)
c
return
c
endif
C*************************************************
C* create a space to put client addr struct into
C*************************************************
c
eval
calen = %size(sockaddr_in)
c
alloc
calen
clientaddr
c
dow
1 = 1
C************************************
C* Get a new server instance ready
C************************************
c
callp(e) Cmd(’SBMJOB CMD(CALL PGM(SVREX6I))’ +
c
’ JOB(SERVERINST) ’ +
c
’ JOBQ(QSYSNOMAX) ’ +
c
’ JOBD(QDFTJOBD) ’ +
c
’ RTGDTA(QCMDB)’: 200)
c
if
%error
c
callp
close(svr)
c
callp
KillEmAll
c
callp
Die(’Unable to submit a new job to ’ +
c
’process clients!’)
c
return
c
endif
C************************************
C* Accept a new client conn
C************************************
c
eval
cli = accept(svr: clientaddr: calen)
c
if
cli < 0
c
eval
err = errno
c
callp
close(svr)
c
callp
KillEmAll
c
callp
die(’accept(): ’ + %str(strerror(err)))
c
return
c
endif
c
c
c
c
c
if
callp
eval
iter
endif
calen <> %size(sockaddr_in)
close(cli)
calen = %size(sockaddr_in)
C************************************
C* get the internal job id of a
128
Chapter 7. Handling many sockets by spawning jobs
C* server instance to handle client
C************************************
c
eval
jilen = %size(dsJobInfo)
c
callp
RcvDtaQ(’SVREX6DQ’: ’SOCKTUT’: jilen:
c
dsJobInfo: 60)
c
if
jilen < 80
c
callp
close(cli)
c
callp
KillEmAll
c
callp
close(svr)
c
callp
die(’No response from server instance!’)
c
return
c
endif
C************************************
C* Pass descriptor to svr instance
C************************************
c
if
givedescriptor(cli: %addr(InternalID))<0
c
eval
err = errno
c
callp
close(cli)
c
callp
KillEmAll
c
callp
close(svr)
c
callp
Die(’givedescriptor(): ’ +
c
%str(strerror(err)))
c
Return
c
endif
c
c
callp
enddo
close(cli)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends any server instances that have been started, but
*
have not been connected with clients.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P KillEmAll
B
D KillEmAll
PI
c
dou
jilen < 80
c
c
c
eval
callp
jilen = %size(dsJobInfo)
RcvDtaQ(’SVREX6DQ’: ’SOCKTUT’: jilen:
dsJobInfo: 1)
c
if
jilen >= 80
c
c
c
c
callp(E)
Cmd(’ENDJOB JOB(’ + %trim(JobNbr) +
’/’ + %trim(JobUser) + ’/’ +
%trim(jobName) + ’) OPTION(*IMMED)’+
’ LOGLMT(0)’: 200)
C
endif
c
P
enddo
E
129
Chapter 7. Handling many sockets by spawning jobs
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Create a new TCP socket that’s listening to a port
*
*
parms:
*
pePort = port to listen to
*
peError = Error message (returned)
*
*
returns: socket descriptor upon success, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewListener
B
D NewListener
PI
10I 0
D
pePort
5U 0 value
D
peError
256A
D
D
D
D
D
D
sock
len
bindto
on
linglen
ling
S
S
S
S
S
S
C*** Create a socket
c
eval
c
c
if
c
eval
c
return
c
endif
10I
10I
*
10I
10I
*
0
0
0 inz(1)
0
sock = socket(AF_INET:SOCK_STREAM:
IPPROTO_IP)
sock < 0
peError = %str(strerror(errno))
-1
C*** Tell socket that we want to be able to re-use the server
C*** port without waiting for the MSL timeout:
c
callp
setsockopt(sock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C*** create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C*** tell socket to only linger for 2 minutes, then discard:
c
eval
l_onoff = 1
c
eval
l_linger = 120
c
callp
setsockopt(sock: SOL_SOCKET: SO_LINGER:
c
ling: linglen)
C*** free up resources used by linger structure
c
dealloc(E)
ling
C*** Create a sockaddr_in structure
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
130
Chapter 7. Handling many sockets by spawning jobs
c
eval
p_sockaddr = bindto
c
c
c
c
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
INADDR_ANY
pePort
*ALLx’00’
C*** Bind socket to port
c
if
bind(sock: bindto: len) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Listen for a connection
c
if
listen(sock: MAXCLIENTS) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Return newly set-up socket:
c
dealloc(E)
c
return
sock
P
E
bindto
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
DS
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
4I 0 INZ(256)
8I 0 INZ(0)
15
16
131
Chapter 7. Handling many sockets by spawning jobs
D
dsECMsgDta
D wwMsgLen
D wwTheKey
17
256
S
S
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
File SOCKTUT/QRPGLESRC, member SVREX6I:
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy
D/copy
D/copy
D/copy
socktut/qrpglesrc,socket_h
socktut/qrpglesrc,sockutil_h
socktut/qrpglesrc,errno_h
socktut/qrpglesrc,jobinfo_h
*** Prototypes for local subprocedures:
D die
D
peMsg
PR
D GetClient
PR
10I 0
D cli
D name
S
S
10I 0
80A
256A
const
c
eval
*inlr = *on
c
c
c
c
c
c
eval
if
callp
cli = GetClient
cli < 0
Die(’Failure retrieving client socket ’+
’descriptor.’)
c
callp
return
endif
WrLine(cli: ’Please enter your ’ +
132
Chapter 7. Handling many sockets by spawning jobs
c
’name now!’)
c
c
c
c
c
if
callp
callp
return
endif
RdLine(cli: %addr(name): 80: *On) < 0
close(cli)
Die(’RdLine() failed!’)
c
c
callp
callp
WrLine(cli: ’Hello ’ + %trim(name) + ’!’)
WrLine(cli: ’Goodbye ’ +%trim(name)+ ’!’)
c
c
callp
return
close(cli)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Get the new client from the listener application
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P GetClient
B
D GetClient
PI
10I 0
D jilen
D sock
S
S
5P 0
10I 0
c
c
c
c
c
callp
c
c
c
c
eval
eval
eval
eval
JobName = JobI_JobName
JobUser = JobI_UserID
JobNbr = JobI_JobNbr
InternalID = JobI_IntJob
c
eval
jilen = %size(dsJobInfo)
c
c
callp
SndDtaq(’SVREX6DQ’: ’SOCKTUT’: jilen:
dsJobInfo)
c
c
P
eval
return
sock = TakeDescriptor(*NULL)
sock
if
return
endif
RtvJobInf(dsJobI0100: %size(dsJobI0100):
’JOBI0100’: ’*’: *BLANKS: dsEC)
dsECBytesA > 0
-1
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
133
Chapter 7. Handling many sockets by spawning jobs
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
7.6. Working towards our next example
If you haven’t yet tried the first "job spawning" example, try it now.
Compile it by typing:
CRTBNDRPG SVREX6L SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
CRTBNDRPG SVREX6I SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
And then start it like so:
SBMJOB CMD(CALL SVREX6L) JOBQ(QSYSNOMAX) JOB(LISTENER)
Test it from your PC by doing a ’telnet as400 4000’, as we’ve discussed in the past few chapters. Look! Nice!
134
Chapter 7. Handling many sockets by spawning jobs
When you’re done playing with it, type ’NETSTAT *CNN’ and find the reference to ’Local Port’ 4000, that’s in
’Listen’ state. Use the ’4’ option to end the server.
Looking back at the sourcecode for SVREX6L, you’ll see that when SVREX6L receives an error on the accept()
command, it goes and does an ’ENDJOB’ command to the server instance. So, in fact, when you entered the 4 in
NETSTAT, SVREX6L ended all of the programs involved. Not too bad, huh?
However, if you tried to end the job using other methods, the listener may end without stopping the server instances.
We should really be checking the SHTDN op-code, and ending all of the instances whenever the ENDJOB,
ENDSBS, ENDSYS or PWRDWNSYS commands are run.
To do this, we need to call select() prior to accept(). Calling select() will allow us to detect when a client connects,
but still provide us with a timeout value, so we can check for the system shutting down periodically.
Add these to the bottom of your D-specs:
D
D
D
D
D
rc
tolen
timeout
readset
excpset
S
S
S
S
S
10I 0
10I 0
*
like(fdset)
like(fdset)
Then, add this code right after the ’alloc calen’ line:
c
c
eval
alloc
tolen = %size(timeval)
tolen
timeout
Then, right before calling accept(), add the code to check for shutdown. It should look something like this:
c
dou
rc > 0
c
c
c
c
c
c
c
callp
callp
callp
callp
eval
eval
eval
FD_ZERO(readset)
FD_ZERO(excpset)
FD_SET(svr: readset)
FD_SET(svr: excpset)
p_timeval = timeout
tv_sec = 20
tv_usec = 0
c
c
eval
rc = select(svr+1: %addr(readset):
*NULL: %addr(excpset): timeout)
c
c
c
c
c
c
c
shtdn
if
callp
callp
callp
return
endif
c
enddo
99
*in99 = *on
close(svr)
KillEmAll
die(’shutdown requested!’)
135
Chapter 7. Handling many sockets by spawning jobs
Now, when you end the job with *CNTRLD, it’ll end all of the server instances as well. We’re making progress!
We also didn’t implement the ’quit’ command that we’ve used in chapters 5 & 6 to end this server from the client
side. If you’re looking for a new challenge, you might want to try to implement this with the new, "job spawning"
approach. However, we won’t do that in this tutorial. (Being realistic, it’s unusual that you want the client to be able
to end your server!)
Another thing that you may note about this new server is that if you start up many clients rapidly, it doesn’t respond
as quickly as the examples from chapter 6 did. (Though, if you have a faster AS/400, this may be hard to detect!)
The reason that it’s slower is that it takes a bit of time for the system to create a new job, make it active, and have it
communicate back to the listener job. The easiest way to improve the performance of this is to "pre-submit" several
different server instances, which can be waiting and ready when a client connects.
To implement this, I’m going to add a new named constant, right after the ’MAXCLIENTS’ constant. It’ll look like
this:
D PRESTART
C
CONST(5)
Then, after we call the ’NewListener’ procedure, we’ll insert this code:
c
c
c
c
c
c
c
c
c
c
c
c
c
c
do
callp(e)
if
callp
callp
callp
PRESTART
Cmd(’SBMJOB CMD(CALL PGM(SVREX6I))’ +
’ JOB(SERVERINST) ’ +
’ JOBQ(QSYSNOMAX) ’ +
’ JOBD(QDFTJOBD) ’ +
’ RTGDTA(QCMDB)’: 200)
%error
close(svr)
KillEmAll
Die(’Unable to submit a new job to ’ +
’process clients!’)
return
endif
enddo
When each of these jobs starts, it’ll put it’s job info onto the data queue. When the listener program goes to read the
data queue, it’ll get them in the order they were added. This should cut the delay between the time it takes to accept
each new client down to 1/5th the time.
Since the ’KillEmAll’ procedure (nice name, huh?) will end any server instance that is currently waiting for a client
to service, it’ll happily end all of the pre-submitted server instances when NETSTAT, ENDJOB, ENDSBS, ENDSYS
or PWRDWNSYS is used to try to shut us down.
Now, we’ll begin working towards a new example server program. This program will ask the client for 3 different
things: a user-id, a password, and a program to call. It will validate the user-id and password, and then use that user’s
authority to call the program.
It will pass the program 2 parameters, the socket descriptor that it can use to talk to the client program and the user-id
of the validated user.
When the called program has completed, the server instance will close the socket and then end.
The next few topics will explain this in more detail.
136
Chapter 7. Handling many sockets by spawning jobs
7.7. Validating a user-id and password
The Get Profile Handle API (QSYGETPH) will be used to check a user-id and password.
The QSYGETPH API is listed under "Security APIs" in the IBM manuals, and you’ll find it at this location:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/QSYGETPH.htm
The manual page tells us that the parameters to the QSYGETPH API look like this:
Required Parameter Group:
1
User ID
Input
Char(10)
2
Password
Input
Char(10)
3
Profile handle
Output
Char(12)
4
Error code
I/O
Char(*)
Optional parameter:
So, our prototype in RPG will look like this:
D GetProfile
D
UserID
D
Password
D
Handle
D
ErrorCode
PR
10A
10A
12A
32766A
ExtPgm(’QSYGETPH’)
const
const
options(*varsize)
Please add that to your JOBINFO_H member.
Or, if you prefer, you could download my copy of JOBINFO_H, available here:
http://www.scottklement.com/rpg/socktut/qrpglesrc.jobinfo_h
When we call this API, we will pass it a user-id and password, and it’ll either return a "profile handle" in the Handle
parameter, or it’ll give us an error in the ErrorCode parameter.
We’ll use this profile handle for the API that we discuss in the next topic.
To call the GetProfile API, we’ll do something like this:
c
callp
GetProfile(UserID: Passwd: Handle:
c
dsEC)
c
if
dsECBytesA > 0
C** handle "invalid user profile" error
c
endif
7.8. Running with the user’s authority
Once we know that a user-id and password as usable, and we’ve received a "profile handle" for them, we can use that
profile handle to make the job "run under that user profile".
We do that by calling the Set Profile (QWTSETP) API. After this API has been called successfully, everything that
new that we try to do will use the authority of the new user-id.
137
Chapter 7. Handling many sockets by spawning jobs
The Set Profile (QWTSETP) API is documented in the IBM manual at this location:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/QWTSETP.htm
The manual tells us that the API accepts these parameters:
Required Parameter Group:
1
Profile handle
Input
Char(12)
I/O
Char(*)
Optional parameter:
2
Error code
The RPG prototype for this API will look like this:
D SetProfile
D
Handle
D
ErrorCode
PR
12A
32766A
ExtPgm(’QWTSETP’)
const
options(*varsize: *nopass)
And we’ll call it like this:
c
callp
GetProfile(UserID: Passwd: Handle:
c
dsEC)
c
if
dsECBytesA > 0
C** handle "invalid user profile" error
c
endif
c
callp
SetProfile(Handle: dsEC)
c
if
dsECBytesA > 0
C** handle "could not set profile" error
C
endif
Please add this prototype to your JOBINFO_H member, as well.
Or, if you prefer, you could download my copy of JOBINFO_H, available here:
http://www.scottklement.com/rpg/socktut/qrpglesrc.jobinfo_h
7.9. A "generic server" example
Are you excited about trying another example program? Are you? Are you?
As I mentioned before, this example program differs from the last one, in that it asks for a userid & password, then
validates them, them changes it’s ’effective user profile’ to the user & password that you’ve signed in as.
Once you’re signed in, it asks for a program name, and then it calls that program, passing the socket descriptor and
user-id as passwords.
This design is very practical, because by using this server program, you can easily write many different client/server
applications without needing to write a separate listener & server instance program for each.
This one involves 3 different programs. The Listener program, which hasn’t changed much since our last example -the only real difference is that the phrase ’SVREX6’ has been changed to ’SVREX76’ throughout the member. The
138
Chapter 7. Handling many sockets by spawning jobs
server instance program, which now validates userid & password, and calls a program. And the ’program to call’, for
which I provide one sample program.
In the next topic, we’ll talk about how to run this program, as well as giving a few samples of what you can do with
this server.
So... here it is!
File: SOCKTUT/QRPGLESRC, Member: SVREX7L
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy
D/copy
D/copy
D/copy
socktut/qrpglesrc,socket_h
socktut/qrpglesrc,sockutil_h
socktut/qrpglesrc,errno_h
socktut/qrpglesrc,jobinfo_h
*** prototypes for external calls
D Cmd
D
command
D
length
PR
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
*** Prototypes for local subprocedures:
D die
D
peMsg
PR
D NewListener
D
pePort
D
peError
PR
D KillEmAll
PR
256A
const
10I 0
5U 0 value
256A
*** local variables & constants
D MAXCLIENTS
D PRESTART
C
C
D
D
D
D
D
D
D
D
D
D
D
D
S
S
S
S
S
S
S
S
S
S
S
S
svr
cli
msg
err
calen
clientaddr
jilen
rc
tolen
timeout
readset
excpset
CONST(256)
CONST(5)
10I
10I
256A
10I
10I
*
5P
10I
10I
*
0
0
0
0
0
0
0
like(fdset)
like(fdset)
139
Chapter 7. Handling many sockets by spawning jobs
c
eval
*inlr = *on
C*************************************************
C* Clean up any previous instances of the dtaq
C*************************************************
c
callp(e) Cmd(’DLTDTAQ SOCKTUT/SVREX7DQ’: 200)
c
callp(e) Cmd(’CRTDTAQ DTAQ(SOCKTUT/SVREX7DQ) ’ +
c
’ MAXLEN(80) TEXT(”Data ’ +
c
’ queue for SVREX7L”)’: 200)
c
if
%error
c
callp
Die(’Unable to create data queue!’)
c
return
c
endif
C*************************************************
C* Start listening for connections on port 4000
C*************************************************
c
eval
svr = NewListener(4000: msg)
c
if
svr < 0
c
callp
die(msg)
c
return
c
endif
C*************************************************
C* Pre-start some server instances
C*************************************************
c
do
PRESTART
c
callp(e) Cmd(’SBMJOB CMD(CALL PGM(SVREX7I))’ +
c
’ JOB(SERVERINST) ’ +
c
’ JOBQ(QSYSNOMAX) ’ +
c
’ JOBD(QDFTJOBD) ’ +
c
’ RTGDTA(QCMDB)’: 200)
c
if
%error
c
callp
close(svr)
c
callp
KillEmAll
c
callp
Die(’Unable to submit a new job to ’ +
c
’process clients!’)
c
return
c
endif
c
enddo
C*************************************************
C* create a space to put client addr struct into
C*************************************************
c
eval
calen = %size(sockaddr_in)
c
alloc
calen
clientaddr
c
c
eval
alloc
tolen = %size(timeval)
tolen
timeout
c
dow
1 = 1
140
Chapter 7. Handling many sockets by spawning jobs
C************************************
C* Get a new server instance ready
C************************************
c
callp(e) Cmd(’SBMJOB CMD(CALL PGM(SVREX7I))’ +
c
’ JOB(SERVERINST) ’ +
c
’ JOBQ(QSYSNOMAX) ’ +
c
’ JOBD(QDFTJOBD) ’ +
c
’ RTGDTA(QCMDB)’: 200)
c
if
%error
c
callp
close(svr)
c
callp
KillEmAll
c
callp
Die(’Unable to submit a new job to ’ +
c
’process clients!’)
c
return
c
endif
C************************************
C* Check every 30 seconds for a
C* system shutdown, until a client
C* connects.
C************************************
c
dou
rc > 0
c
c
c
c
c
c
c
callp
callp
callp
callp
eval
eval
eval
FD_ZERO(readset)
FD_ZERO(excpset)
FD_SET(svr: readset)
FD_SET(svr: excpset)
p_timeval = timeout
tv_sec = 20
tv_usec = 0
c
c
eval
rc = select(svr+1: %addr(readset):
*NULL: %addr(excpset): timeout)
c
c
c
c
c
c
c
shtdn
if
callp
callp
callp
return
endif
c
enddo
99
*in99 = *on
close(svr)
KillEmAll
die(’shutdown requested!’)
C************************************
C* Accept a new client conn
C************************************
c
eval
cli = accept(svr: clientaddr: calen)
c
if
cli < 0
c
eval
err = errno
c
callp
close(svr)
c
callp
KillEmAll
c
callp
die(’accept(): ’ + %str(strerror(err)))
141
Chapter 7. Handling many sockets by spawning jobs
c
c
return
endif
c
c
c
c
c
if
callp
eval
iter
endif
calen <> %size(sockaddr_in)
close(cli)
calen = %size(sockaddr_in)
C************************************
C* get the internal job id of a
C* server instance to handle client
C************************************
c
eval
jilen = %size(dsJobInfo)
c
callp
RcvDtaQ(’SVREX7DQ’: ’SOCKTUT’: jilen:
c
dsJobInfo: 60)
c
if
jilen < 80
c
callp
close(cli)
c
callp
KillEmAll
c
callp
close(svr)
c
callp
die(’No response from server instance!’)
c
return
c
endif
C************************************
C* Pass descriptor to svr instance
C************************************
c
if
givedescriptor(cli: %addr(InternalID))<0
c
eval
err = errno
c
callp
close(cli)
c
callp
KillEmAll
c
callp
close(svr)
c
callp
Die(’givedescriptor(): ’ +
c
%str(strerror(err)))
c
Return
c
endif
c
c
callp
enddo
close(cli)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends any server instances that have been started, but
*
have not been connected with clients.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P KillEmAll
B
D KillEmAll
PI
c
dou
jilen < 80
c
c
c
eval
callp
jilen = %size(dsJobInfo)
RcvDtaQ(’SVREX7DQ’: ’SOCKTUT’: jilen:
dsJobInfo: 1)
142
Chapter 7. Handling many sockets by spawning jobs
c
if
jilen >= 80
c
c
c
c
callp(E)
Cmd(’ENDJOB JOB(’ + %trim(JobNbr) +
’/’ + %trim(JobUser) + ’/’ +
%trim(jobName) + ’) OPTION(*IMMED)’+
’ LOGLMT(0)’: 200)
C
endif
c
P
enddo
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Create a new TCP socket that’s listening to a port
*
*
parms:
*
pePort = port to listen to
*
peError = Error message (returned)
*
*
returns: socket descriptor upon success, or -1 upon error
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P NewListener
B
D NewListener
PI
10I 0
D
pePort
5U 0 value
D
peError
256A
D
D
D
D
D
D
sock
len
bindto
on
linglen
ling
S
S
S
S
S
S
C*** Create a socket
c
eval
c
c
if
c
eval
c
return
c
endif
10I
10I
*
10I
10I
*
0
0
0 inz(1)
0
sock = socket(AF_INET:SOCK_STREAM:
IPPROTO_IP)
sock < 0
peError = %str(strerror(errno))
-1
C*** Tell socket that we want to be able to re-use the server
C*** port without waiting for the MSL timeout:
c
callp
setsockopt(sock: SOL_SOCKET:
c
SO_REUSEADDR: %addr(on): %size(on))
C*** create space for a linger structure
c
eval
linglen = %size(linger)
c
alloc
linglen
ling
c
eval
p_linger = ling
C*** tell socket to only linger for 2 minutes, then discard:
143
Chapter 7. Handling many sockets by spawning jobs
c
c
c
c
eval
eval
callp
l_onoff = 1
l_linger = 120
setsockopt(sock: SOL_SOCKET: SO_LINGER:
ling: linglen)
C*** free up resources used by linger structure
c
dealloc(E)
ling
C*** Create a sockaddr_in structure
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
eval
p_sockaddr = bindto
c
c
c
c
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
INADDR_ANY
pePort
*ALLx’00’
C*** Bind socket to port
c
if
bind(sock: bindto: len) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Listen for a connection
c
if
listen(sock: MAXCLIENTS) < 0
c
eval
peError = %str(strerror(errno))
c
callp
close(sock)
c
dealloc(E)
bindto
c
return
-1
c
endif
C*** Return newly set-up socket:
c
dealloc(E)
c
return
sock
P
E
bindto
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
PR
7A
20A
256A
10I 0
ExtPgm(’QMHSNDPM’)
Const
Const
Const
Const
144
Chapter 7. Handling many sockets by spawning jobs
D
D
D
D
D
MsgType
CallStkEnt
CallStkCnt
MessageKey
ErrorCode
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
File: SOCKTUT/QRPGLESRC, Member: SVREX7I
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy
D/copy
D/copy
D/copy
socktut/qrpglesrc,socket_h
socktut/qrpglesrc,sockutil_h
socktut/qrpglesrc,errno_h
socktut/qrpglesrc,jobinfo_h
*** Prototypes for local subprocedures:
D die
D
peMsg
PR
D GetClient
PR
10I 0
D SignIn
D
sock
PR
10I 0
10I 0 value
256A
const
145
Chapter 7. Handling many sockets by spawning jobs
D
D
D
D
D
userid
cli
rc
usrprf
pgmname
D lower
D upper
c
10A
S
S
S
S
10I 0
10I 0
10A
21A
C
C
’abcdefghijklmnopqrstuvwxyz’
’ABCDEFGHIJKLMNOPQRSTUVWXYZ’
eval
*inlr = *on
C*********************************************************
C* Get socket descriptor from ’listener’ program
C*********************************************************
c
eval
cli = GetClient
c
if
cli < 0
c
callp
Die(’Failure retrieving client socket ’+
c
’descriptor.’)
c
return
c
endif
C*********************************************************
C* Ask user to sign in, and set user profile.
C*********************************************************
c
eval
rc = SignIn(cli: usrprf)
c
select
c
when
rc < 0
c
callp
Die(’Client disconnected during sign-in’)
c
callp
close(cli)
c
return
c
when
rc = 0
c
callp
Die(’Authorization failure!’)
c
callp
close(cli)
c
return
c
endsl
C*********************************************************
C* Ask for the program to be called
C*********************************************************
c
callp
WrLine(cli: ’102 Please enter the ’ +
c
’program you”d like to call’)
c
c
c
c
c
c
lower:upper
if
callp
callp
return
endif
RdLine(cli: %addr(pgmname): 21: *On) < 0
Die(’Error calling RdLine()’)
close(cli)
xlate
pgmname
pgmname
C*********************************************************
C* Call the program, passing the socket desc & profile
146
Chapter 7. Handling many sockets by spawning jobs
C*
as the parameters.
C*********************************************************
c
call(e)
PgmName
c
parm
cli
c
parm
usrprf
c
c
c
c
c
if
callp
else
callp
endif
not %error
WrLine(cli: ’103 Call succeeded.’)
WrLine(cli: ’902 Call failed.’)
C*********************************************************
C* End.
C*********************************************************
c
callp
close(cli)
c
return
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Sign a user-id into the system
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P SignIn
B
D SignIn
PI
10I 0
D
sock
10I 0 value
D
userid
10A
D passwd
D handle
S
S
10A
12A
c
dou
userid <> *blanks
c
c
callp
WrLine(sock: ’100 Please enter your ’ +
’user-id now!’)
c
c
c
if
return
endif
RdLine(sock: %addr(userid): 10: *On) < 0
-1
xlate
userid
c
c
callp
WrLine(sock: ’101 Please enter your ’ +
’password now!’)
c
c
c
if
return
endif
RdLine(sock: %addr(passwd): 10: *On) < 0
-1
xlate
passwd
callp
if
callp
GetProfile(userid: passwd: handle: dsEC)
dsECBytesA > 0
WrLine(sock: ’900 Incorrect userid ’ +
c
c
c
c
c
lower:upper
lower:upper
userid
passwd
147
Chapter 7. Handling many sockets by spawning jobs
c
c
c
eval
endif
c
enddo
c
c
c
c
c
c
callp
if
callp
c
P
return
endif
return
’or password! (’+%trim(dsECMsgID)+’)’)
userid = *blanks
SetProfile(handle: dsEC)
dsECBytesA > 0
WrLine(sock: ’901 Unable to set ’ +
’profile! (’+%trim(dsECMsgID)+’)’)
0
1
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* Get the new client from the listener application
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P GetClient
B
D GetClient
PI
10I 0
D jilen
D sock
S
S
5P 0
10I 0
c
c
c
c
c
callp
c
c
c
c
eval
eval
eval
eval
JobName = JobI_JobName
JobUser = JobI_UserID
JobNbr = JobI_JobNbr
InternalID = JobI_IntJob
c
eval
jilen = %size(dsJobInfo)
c
c
callp
SndDtaq(’SVREX7DQ’: ’SOCKTUT’: jilen:
dsJobInfo)
c
c
P
eval
return
sock = TakeDescriptor(*NULL)
sock
if
return
endif
RtvJobInf(dsJobI0100: %size(dsJobI0100):
’JOBI0100’: ’*’: *BLANKS: dsEC)
dsECBytesA > 0
-1
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
148
Chapter 7. Handling many sockets by spawning jobs
D
peMsg
256A
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
const
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
File: SOCKTUT/QRPGLESRC, Member: TESTPGM
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
*** header files for calling service programs & APIs
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D sock
D user
c
S
S
*entry
10I 0
10A
plist
149
Chapter 7. Handling many sockets by spawning jobs
c
c
parm
parm
sock
user
c
c
callp
callp
WrLine(sock: ’Hello ’ + %trim(user))
WrLine(sock: ’Goodbye ’ + %trim(user))
c
eval
*inlr = *on
7.10. Trying the "generic server" out
If you haven’t already done so, compile the "generic server" example like this:
CRTBNDRPG SVREX7L SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
CRTBNDRPG SVREX7I SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
CRTBNDRPG TESTPGM SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
Then start it up, like so:
SBMJOB CMD(CALL SVREX7L) JOBQ(QSYSNOMAX) JOB(LISTENER)
Try it out by telnetting from your PC:
telnet as400 4000
It’ll ask for your user-id, and password. After you’ve typed both, it’ll set your user profile to the one that you’ve
typed it.
Note: Unless SVREX7L was submitted by a user with *ALLOBJ authority, it will only be able to set the user profile
to the same profile that it’s running under. This is a security safeguard on the AS/400 that prevents programmers
from writing programs that would give them more access than they are allowed to have. The solution, of course,
would be to run SVREX7L from a job with *ALLOBJ authority, so that it can handle sign-ons for any user.
Once it has set the user profile, it’ll ask you for a program to call. You can qualify your program name with a library
if you like. To try my demonstration program, you could type SOCKTUT/TESTPGM
You’ll notice that each response the server program sends to the telnet client is prefixed by a 3-digit number. This is
to make it easy to write a client program that interfaces with the server. The first 3 digits are always a number, so if
the client wanted to display the human-readable portion, it could do a simple substring starting in position 5 to
remove the message number. The message number can be used to discern exactly what the server is expecting. Plus,
any number starting with a ’1’ is a positive response, any number starting with a ’9’ is a negative (error) response.
The TESTPGM program will say ’hello’ & ’goodbye’ just as our other simple server programs have done. Not
particularly practical, but it’s always good to start simple.
Here’s another example of a program you could call. This one asks for the name of a source file, and a member. It
then sends back the member’s contents. Each source line starts with an extra ’.’, which a client program would
normally remove. This is useful, as it allows us to tell the difference between a line of source code and a server’s
response message.
File: SOCKTUT/QRPGLESRC, Member: GETSRC
150
Chapter 7. Handling many sockets by spawning jobs
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’SOCKTUT/SOCKUTIL’) BNDDIR(’QC2LE’)
FSOURCE
IF
F
252
DISK
USROPN INFDS(dsSrc)
*** header files for calling service programs & APIs
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D Cmd
D
command
D
length
PR
D dsSrc
D
dsSrcRLen
DS
D lower
D upper
C
C
D
D
D
D
D
S
S
S
S
S
sock
user
file
mbr
reclen
ISOURCE
I
c
c
c
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
125
126I 0
’abcdefghijklmnopqrstuvwxyz’
’ABCDEFGHIJKLMNOPQRSTUVWXYZ’
10I 0
10A
21A
10A
5I 0
NS
*entry
13
252
plist
parm
parm
SrcDta
sock
user
c
eval
*inlr = *on
c
c
c
c
callp
if
return
endif
WrLine(sock: ’110 Name of source file?’)
RdLine(sock: %addr(file): 21: *On) < 0
c
c
c
c
callp
if
return
endif
WrLine(sock: ’111 Name of member?’)
RdLine(sock: %addr(mbr): 21: *On) < 0
xlate
xlate
file
mbr
callp(e)
cmd(’OVRDBF FILE(SOURCE) ’ +
’TOFILE(’+%trim(file)+’) ’ +
’MBR(’ +%trim(mbr)+ ’)’: 200)
%error
WrLine(sock: ’910 Error calling OVRDBF’)
c
c
c
c
c
c
c
lower:upper
lower:upper
if
callp
file
mbr
151
Chapter 7. Handling many sockets by spawning jobs
c
c
return
endif
c
c
c
c
c
c
open(e)
if
callp
callp
return
endif
SOURCE
%error
WrLine(sock: ’911 Unable to open file!’)
Cmd(’DLTOVR FILE(SOURCE)’: 200)
c
eval
reclen = dsSrcRLen - 12
c
c
c
c
c
c
c
c
read
dow
if
SOURCE
not %eof(SOURCE)
WrLine(sock:
’.’ + %subst(SrcDta:1:reclen)) < 0
leave
endif
read
enddo
c
callp
WrLine(sock: ’112 Download successful!’)
c
c
c
close
callp
return
SOURCE
Cmd(’DLTOVR FILE(SOURCE)’: 200)
SOURCE
Compile it by typing:
CRTBNDRPG GETSRC SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
Now use your PC to telnet to the AS/400’s port 4000 again. Type in your user name and pasword. When it asks for
the program, type socktut/getsrc. For source file, you might type something like SOCKTUT/QRPGLESRC for
source member, maybe type TESTPGM
See how easy it is, using this example server program to write your own TCP servers? Just think, by writing your
own program (perhaps somewhat similar to GETSRC) and by writing a GUI client program, you could write a
client/server application using native RPG database access!
Whew. I think we’ve covered server programs well enough for now. Time to try something new...
152
Chapter 8. The User Datagram Protocol
Written by Scott Klement.
8.1. Overview of using UDP
Like TCP, the User Datagram Protocol (UDP) is a protocol that runs on top of IP. Where TCP is designed to allow
you to communicate using "streams" of data, UDP is designed to let you communicate using "datagrams."
As you may recall from the introduction at the start of this tutorial, all IP information is transmitted in "datagrams".
So why is there a UDP? What is the difference between an "IP datagram" and a "UDP datagram"?
What’s the difference? What’s the difference? Not much.
If you really want to know (you were quite insistent, after all) here’s the difference:
struct udphdr {
u_short
u_short
short
u_short
uh_sport;
uh_dport;
uh_ulen;
uh_sum;
/* UDP header
*/
/*
/*
/*
/*
*/
*/
*/
*/
source port
destination port
UDP length
UDP checksum
};
That’s it. Those 8 bytes are the only difference between UDP and IP datagrams. ("short" is 5I 0 in RPG, and
"u_short" is 5U 0 in RPG... so these are all "short integers") When these 4 items are placed in the first 8 bytes of an
IP datagram, it becomes a UDP datagram!
What these four integers provide is a "source port", "destination port", and some safeguards to ensure that the data is
the same length and content on the receiving side as it was on the sending side.
These "source" and "destination" ports are really very significant. They allow different applications to call bind() and
therefore these applications can be addressed individually. They’re a method of differentiating between different
applications on the same computer.
In other words, if these ports weren’t specified, only one program on the system could work with them -- the
operating system itself -- because there would be no way to differentiate each program’s datagrams.
And this is why it’s called the "User" datagram protocol. It allows "user" applications to send and receive datagrams,
as opposed to system processes.
Unlike TCP, UDP does not ensure that data was received at the other end of the connection. When many datagrams
are sent, UDP does not ensure that they are received in the same order that they were sent like TCP does.
So why would you ever use UDP?
Well, first of all, it’s faster than TCP. The "safeguards" and "datagram sequencing" functions of TCP do add some
overhead to the protocol. UDP does not have that overhead, and therefore UDP is a bit faster.
Secondly, you don’t always WANT the data that gets "lost". An example of this might be a radio broadcast over the
internet. If one datagram in the middle of a stream of audio data gets lost, it might be better to just skip the lost data
153
Chapter 8. The User Datagram Protocol
-- much like receiving static on your traditional radio, rather than lose time in the broadcast. Real Audio is an
application that uses this feature.
Third, you don’t always need to send more than a single datagram of data. If the entire conversation consists of only
a few bytes of data, there’s no need to undergo the additional overhead of negotiating a TCP connection. Just send all
the data in one datagram, and have the other side manually send back an acknowledgement.
So, although there are many advantages to using TCP, there are some situations where UDP makes more sense.
Nearly all of the internet application protocols are run over TCP. But there are still several that use UDP. Some of the
better known ones are:
•
Domain Name System ("DNS") (the protocol that translates names, such as ’www.myhost.com’ to IP addresses)
•
Network Time Protocol ("NTP") (a protocol for synchronizing your system clock over a TCP/IP network)
•
Simple Network Management Protocol ("SNMP") (a protocol for managing your network devices)
•
Traceroute (a program used to see how packets are being routed across the internet)
By comparison to TCP, very few applications actually use UDP. However, it is still an important protcol that deserves
a look in our tutorial!
8.2. Creating a socket for use with UDP
Ya know, there’s nothing like a good cheeseburger.
Like TCP, UDP programming requires the use of a socket. You’ll need to call the socket() API to get one.
Before we can do that, we’ll need to add a new constant to our SOCKET_H member:
D SOCK_DGRAM
C
CONST(2)
This tells the socket() API that you want a datagram socket instead of a TCP socket.
Now, we can call the socket() API like this:
C
C
C**
C
eval
s = socket(AF_INET:SOCK_DGRAM:IPPROTO_IP)
if
s < 0
error occurred, check errno
endif
As you can see, we still tell the socket() API that it’s going to use the "INET" (internet) address family. And that it’s
going to use the "IP" (Internet Protocol) In TCP, we send our data in a stream, but in UDP, we’ll use datagrams, thus
we tell it "SOCK_DGRAM" instead of "SOCK_STREAM".
Makes a lot more sense than that cheeseburger remark, doesn’t it?
154
Chapter 8. The User Datagram Protocol
8.3. The sendto() API call
Each datagram that you send on a UDP socket is sent "by itself". It’s not a part of a larger conversation. There is no
’connection’ to make, since there isn’t a continuous stream of data. It’s "one message at a time."
Therefore, a sending program must send a destination address and port with each and every datagram that gets sent.
The send() API doesn’t give us a place to put an address or port, so a new API is necessary, sendto().
The sendto() API is found in the IBM manuals at this location:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/sendto.htm
It tells us that the C-language prototype for sendto() looks like this:
int sendto(int socket_descriptor,
char *buffer,
int buffer_length,
int flags,
struct sockaddr *destination_address,
int address_length);
Nothing terribly different or exciting about this prototype. A procedure called ’sendto’, with 6 parameters. An
integer, a pointer, an integer, another integer, a pointer and finally an integer. So, lets prototype this in RPG:
D sendto
D
sock_desc
D
buffer
D
buffer_len
D
flags
D
destaddr
D
addrlen
PR
10I
10I
*
10I
10I
*
10I
0 ExtProc(’sendto’)
0 Value
Value
0 Value
0 Value
Value
0 Value
The first four parms are identical to those used in the send() API. The only difference is that we’ve added a sockaddr
structure to contain the address & port number of the destination. And, like every other address structure we’ve done,
we’ll need to supply a length for it.
Please, add this little gem to your SOCKET_H member.
The sendto() API is called like this:
c
c
eval
alloc
len = %size(sockaddr_in)
len
toaddr
c
c
c
c
c
eval
eval
eval
eval
eval
p_sockaddr
sin_family
sin_addr =
sin_port =
sin_zero =
c
if
c
C*** error! check errno!
c
endif
= toaddr
= AF_INET
SOME_IP
SOME_PORT
*ALLx’00’
sendto(s: %addr(buf): buflen: 0:
toaddr: len) < 0
155
Chapter 8. The User Datagram Protocol
8.4. The recvfrom() API call
Just as the sendto() API adds an address and port to the send() API, so does the recvfrom() API add an address and
port to the recv() API.
Sometimes in a TCP socket, you don’t bother to keep track of where the connection comes from. You don’t need to.
You don’t care. With TCP, the API takes care of making sure your responses get where they need to go.
This is not true with UDP. Each datagram stands alone. You need to read in the address that a datagram came from so
that you know who to respond to -- that is, if your application requires responses at all.
So, when you supply an address and length in the recvfrom API, it’s to receive the address and port that the datagram
originated with. You can use it to reply!
The IBM manual page for the recvfrom() API is found right about here:
http://publib.boulder.ibm.com/pubs/html/as400/v4r5/ic2924/info/apis/recvfr.htm
And it tells us that the C-language prototype for the recvfrom() API looks like this:
int recvfrom(int socket_descriptor,
char *buffer,
int buffer_length,
int flags,
struct sockaddr *from_address,
int *address_length);
Yes, it’s just like recv(), except that they’ve added a socket address structure, and an address length. You’ll see that
the address length is passed by pointer -- this is because the length depends on what the sending program sends us,
just like it did with the accept() API.
The RPG version of recvfrom() looks like this:
D recvfrom
D
sock_desc
D
buffer
D
buffer_len
D
flags
D
fromaddr
D
addrlength
PR
10I
10I
*
10I
10I
*
10I
0 ExtProc(’recvfrom’)
0 Value
Value
0 Value
0 Value
Value
0
You’ll note that we’re passing the address length parameter ’by address’ instead of passing a pointer to an integer.
This looks exactly the same to the API that receives the parameter, but allows the compiler to do some better validity
checking.
Add this to your SOCKET_H member, please.
Here’s how we call recvfrom:
C
eval
datalen = recvfrom(s: %addr(buf):
c
%size(buf): 0: fromaddr: fromlen)
c
if
datalen < 0
c*** Error occurred, check errno!
c
endif
156
Chapter 8. The User Datagram Protocol
8.5. Sample UDP server
Time to put try our new found UDP knowledge out with a sample UDP server program. This program will receive
datagrams containing a user-id and some text, and then will send that user a message using the SNDMSG command.
Once the message has been sent, it will send back a datagram that acknowledges that the message was sent correctly.
So, the pseudocode for our server looks like this:
1.
Create a UDP socket
2.
Bind the socket to a port. Maybe port 4000, that’s worked pretty well so far, yes?
3.
Wait for any datagrams to be received. (we can use select() to give us a timeout, so we can check for system shut
down)
4.
convert the datagram to EBCDIC.
5.
substring out a userid & message
6.
Call the SNDMSG command to send the message
7.
Send back an acknowledgement packet
8.
GO back to step 3.
Note: Unlikely TCP, we CANNOT use a simple telnet client to test our server program. We’ll have to write a client
program to play with our server -- which we’ll do in the next topic.
Here’s our server:
File SOCKTUT/QRPGLESRC, Member: UDPSERVER
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’QC2LE’) BNDDIR(’SOCKTUT/SOCKUTIL’)
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D/copy socktut/qrpglesrc,errno_h
D Cmd
D
command
D
length
PR
D translate
D
length
D
data
D
table
PR
D die
D
peMsg
PR
D sock
D err
D bindto
S
S
S
ExtPgm(’QCMDEXC’)
200A
const
15P 5 const
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
256A
const
10I 0
10I 0
*
157
Chapter 8. The User Datagram Protocol
D
D
D
D
D
D
D
D
D
D
D
D
c
len
dtalen
msgfrom
fromlen
timeout
tolen
readset
excpset
user
msg
buf
rc
S
S
S
S
S
S
S
S
S
S
S
S
10I
10I
*
10I
*
10I
0
0
0
0
like(fdset)
like(fdset)
10A
150A
256A
10I 0
eval
*inlr = *on
C* Reserve space for the address that we receive a
C* datagram from.
c
eval
fromlen = %size(sockaddr_in)
c
alloc
fromlen
msgfrom
C* Create a timeval structure so we can check for shutdown
C*
every 25 seconds:
c
eval
tolen = %size(timeval)
c
alloc
tolen
timeout
c
c
c
eval
eval
eval
p_timeval = timeout
tv_sec = 25
tv_usec = 0
C* Create a socket to do our UDP stuff:
c
eval
sock = socket(AF_INET: SOCK_DGRAM:
c
IPPROTO_IP)
c
if
sock < 0
c
callp
die(’socket(): ’+%str(strerror(errno)))
c
return
c
endif
C* Create a sockaddr struct to tell the
C* bind() API which port to use.
c
eval
len = %size(sockaddr_in)
c
alloc
len
bindto
c
c
c
c
c
eval
eval
eval
eval
eval
C* bind to the port.
c
if
c
eval
c
callp
c
callp
p_sockaddr
sin_family
sin_addr =
sin_port =
sin_zero =
= bindto
= AF_INET
INADDR_ANY
4000
*ALLx’00’
bind(sock: bindto: len) < 0
err = errno
close(sock)
die(’bind(): ’+%str(strerror(err)))
158
Chapter 8. The User Datagram Protocol
c
c
return
endif
c
dow
1 = 1
C* Use select to determine when data is found
c
callp
fd_zero(readset)
c
callp
fd_zero(excpset)
c
callp
fd_set(sock: readset)
c
callp
fd_set(sock: excpset)
c
c
eval
rc = select(sock+1: %addr(readset):
*NULL: %addr(excpset): timeout)
C* If shutdown is requested, end program.
c
shtdn
c
if
*in99 = *on
c
callp
close(sock)
c
return
c
endif
99
C* If select timed out, go back to select()...
c
if
rc = 0
c
iter
c
endif
C* Receive a datagram:
c
eval
c
C* Check for errors
c
if
c
eval
c
callp
c
callp
c
return
c
endif
C* Skip any invalid messages
c
if
c
iter
c
endif
dtalen = recvfrom(sock: %addr(buf):
%size(buf): 0: msgfrom: fromlen)
dtalen < 0
err = errno
close(sock)
die(’recvfrom(): ’+%str(strerror(err)))
dtalen < 11
C* Skip any messages from invalid addresses
c
if
fromlen <> %size(sockaddr_in)
c
eval
fromlen = %size(sockaddr_in)
c
iter
c
endif
C* Translate to EBCDIC
c
callp
Translate(dtalen: buf: ’QTCPEBC’)
c* send message to user
159
Chapter 8. The User Datagram Protocol
c
c
c
eval
eval
eval
user = %subst(buf:1: 10)
dtalen = dtalen - 10
msg = %subst(buf:11:dtalen)
c
c
callp(e)
Cmd(’SNDMSG MSG(”’ + %trimr(msg) +
”’) TOUSR(’ + %trim(user) + ’)’: 200)
c* make an acknowledgement
c
if
c
eval
c
eval
c
else
c
eval
c
eval
c
endif
%error
buf = ’failed’
dtalen = 6
buf = ’success’
dtalen = 7
c* convert acknowledgement to ASCII
c
callp
Translate(dtalen: buf: ’QTCPASC’)
c* send acknowledgement to ASCII
c
if
sendto(sock: %addr(buf): dtalen: 0:
c
msgfrom: fromlen) < 0
c
eval
err = errno
c
callp
close(sock)
c
callp
die(’sendto(): ’+%str(strerror(err)))
c
return
c
endif
c
enddo
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
DS
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
4I 0 INZ(256)
8I 0 INZ(0)
160
Chapter 8. The User Datagram Protocol
D
D
D
dsECMsgID
dsECReserv
dsECMsgDta
D wwMsgLen
D wwTheKey
9
16
17
15
16
256
S
S
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
8.6. Sample UDP client
Here is a sample client program that goes with our UDP server. This also, incidentally, demonstrates the ability of the
socket library to enable communications between two different programs on the same computer.
Here’s pseudocode for our client:
1.
We need 3 pieces of data from the command line. The host to send datagrams to, a user-id, and a message.
(Since the CALL command has problems with variables over 32 bytes long, we’ll have to create a *cmd object
as a front-end)
2.
Attempt to treat the hostname as a dotted-decimal IP address, and convert it into a 32-bit address.
3.
If it’s not a dotted-decimal IP address, attempt to look the host up using DNS by calling the gethostbyname()
API.
4.
Create a new socket. Use SOCK_DGRAM to tell the API that we want a UDP socket.
5.
Create a socket address structure, and assign it the address that we calculated in step 2-3. Also assign port 4000.
6.
Format the user-id and message into a buffer. Translate that buffer to ASCII.
7.
Send the buffer as a UDP datagram to the address that we assigned in step 5.
8.
Wait for an acknowledgement datagram.
9.
Convert the acknowledgement data back to EBCDIC.
10. Send back the results as a "completion message".
Here’s the source code for the *cmd object that we’ll use as a front-end:
161
Chapter 8. The User Datagram Protocol
File: SOCKTUT/QCMDSRC, Member: UDPCLIENT
CMD
PROMPT(’Sample UDP client’)
PARM
KWD(RMTSYS) TYPE(*CHAR) LEN(128) MIN(1) +
PROMPT(’Remote System’)
PARM
KWD(USERID) TYPE(*CHAR) LEN(10) MIN(1) +
PROMPT(’UserID to Send Message To’)
PARM
KWD(MSG) TYPE(*CHAR) LEN(150) MIN(1) +
PROMPT(’Message to Send’)
Here’s the RPG source code for our client:
File: SOCKTUT/QRPGLESRC, Member: UDPCLIENT
H DFTACTGRP(*NO) ACTGRP(*NEW)
H BNDDIR(’QC2LE’) BNDDIR(’SOCKTUT/SOCKUTIL’)
D/copy socktut/qrpglesrc,socket_h
D/copy socktut/qrpglesrc,sockutil_h
D/copy socktut/qrpglesrc,errno_h
D translate
D
length
D
data
D
table
PR
D compmsg
D
peMsg
PR
D die
D
peMsg
PR
D
D
D
D
D
D
D
D
D
D
D
D
S
S
S
S
S
S
S
S
S
S
S
S
c
c
c
c
sock
err
len
bindto
addr
buf
buflen
host
user
msg
destlen
destaddr
*entry
ExtPgm(’QDCXLATE’)
5P 0 const
32766A
options(*varsize)
10A
const
256A
const
256A
const
10I
10I
10I
*
10U
256A
10I
128A
10A
150A
10I
*
plist
parm
parm
parm
0
0
0
0
0
0
host
user
msg
162
Chapter 8. The User Datagram Protocol
c
eval
*inlr = *on
C* Get the 32-bit network IP address for the host
C* that was supplied by the user:
c
eval
addr = inet_addr(%trim(host))
c
if
addr = INADDR_NONE
c
eval
p_hostent = gethostbyname(%trim(host))
c
if
p_hostent = *NULL
c
callp
die(’Unable to find that host!’)
c
return
c
endif
c
eval
addr = h_addr
c
endif
C* Create a UDP socket:
c
eval
c
c
if
c
callp
c
return
c
endif
sock = socket(AF_INET: SOCK_DGRAM:
IPPROTO_IP)
sock < 0
die(’socket(): ’+%str(strerror(errno)))
C* Create a socket address struct with destination info
C
eval
destlen = %size(sockaddr_in)
c
alloc
destlen
destaddr
c
eval
p_sockaddr = destaddr
c
c
c
c
eval
eval
eval
eval
sin_family
sin_addr =
sin_port =
sin_zero =
= AF_INET
addr
4000
*ALLx’00’
C* Create a buffer with the userid & password and
C* translate it to ASCII
c
eval
buf = user
c
eval
%subst(buf:11) = msg
c
eval
buflen = %len(%trimr(buf))
c
callp
translate(buflen: buf: ’QTCPASC’)
C* Send the datagram
c
if
c
c
eval
c
callp
c
callp
c
return
c
endif
C* Wait for an ack
c
c
c
eval
if
sendto(sock: %addr(buf): buflen: 0:
destaddr: destlen) < 0
err = errno
close(sock)
die(’sendto(): ’+%str(strerror(err)))
len = recvfrom(sock: %addr(buf): 256: 0:
destaddr: destlen)
len < 6
163
Chapter 8. The User Datagram Protocol
c
c
c
c
callp
callp
return
endif
C* Report status & end
c
callp
c
callp
c
c
c
close(sock)
die(’error receiving ack!’)
Translate(len: buf: ’QTCPEBC’)
compmsg(’Message sent. Server ’ +
’responded with: ’ + %subst(buf:1:len))
callp
return
close(sock)
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This sends a ’completion message’, showing a successful
*
termination to the program.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P compmsg
B
D compmsg
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*COMP’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
return
164
Chapter 8. The User Datagram Protocol
P
E
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
* This ends this program abnormally, and sends back an escape.
*
message explaining the failure.
*+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
P die
B
D die
PI
D
peMsg
256A
const
D SndPgmMsg
D
MessageID
D
QualMsgF
D
MsgData
D
MsgDtaLen
D
MsgType
D
CallStkEnt
D
CallStkCnt
D
MessageKey
D
ErrorCode
PR
D dsEC
D dsECBytesP
D dsECBytesA
D dsECMsgID
D dsECReserv
D dsECMsgDta
DS
D wwMsgLen
D wwTheKey
S
S
ExtPgm(’QMHSNDPM’)
7A
Const
20A
Const
256A
Const
10I 0 Const
10A
Const
10A
Const
10I 0 Const
4A
32766A
options(*varsize)
1
5
9
16
17
4I 0 INZ(256)
8I 0 INZ(0)
15
16
256
10I 0
4A
c
c
c
c
eval
if
return
endif
wwMsgLen = %len(%trimr(peMsg))
wwMsgLen<1
c
c
c
callp
SndPgmMsg(’CPF9897’: ’QCPFMSG
*LIBL’:
peMsg: wwMsgLen: ’*ESCAPE’:
’*PGMBDY’: 1: wwTheKey: dsEC)
c
P
return
E
/define ERRNO_LOAD_PROCEDURE
/copy socktut/qrpglesrc,errno_h
165
Chapter 8. The User Datagram Protocol
8.7. Trying it out
In order to try out our sample UDP programs, we’ll first have to compile them. Let’s build the server first:
CRTBNDRPG UDPSERVER SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
Now the client:
CRTBNDRPG UDPCLIENT SRCFILE(SOCKTUT/QRPGLESRC) DBGVIEW(*LIST)
CRTCMD CMD(UDPCLIENT) PGM(SOCKTUT/UDPCLIENT) SRCFILE(SOCKTUT/QCMDSRC)
Start the server by typing this:
SBMJOB CMD(CALL UDPSERVER) JOBQ(QSYSNOMAX) JOB(UDP)
And then you can try out the client, like this:
UDPCLIENT RMTSYS(’127.0.0.1’) USERID(klemscot) MSG(’this is a fine test!’)
At first glance, this program doesn’t seem very practical. After all, you could’ve done the same thing with the
SNDMSG command. However, if you now go to another AS/400, and put our newly built UDPCLIENT on that
machine, you can use it to send messages over the network! or even over the internet!
Likewise, if you wrote a message client (similar to UDPCLIENT) to run on your PC, you could send messages from
there.
Of course, this is still not very practical :) It is, however, a simple example of how to use UDP datagrams. If you have
ideas for a better UDP program, I certainly encourage you to go ahead and experiment!
This concludes my socket tutorial (for now, anyway) I hope you found it enlightening.
If you’ve found any mistakes in the tutorial, or anything that you feel is not explained as well as it could be, please
send me an E-mail. I’d like this tutorial to be the best that it can be -- and your suggestions will help to make that
happen!
166
Colophon
This tutorial was written by Scott Klement as a free service to the AS400 / iSeries RPG community.
It was originally written as a plain ASCII text file, written using the vi text editor on a FreeBSD system. Before
releasing it to the public, however, I re-formatted it into HTML to match the the theme of the rest of my web page.
After receiving many helpful comments from members of the RPG community, I decided that it would be better to
re-format it into SGML, so that it could be transformed into PDF, RTF and HTML formats for both on-line viewing
and easy printing. After some research, I decided to use the DocBook DTD. This allows me to transform it into
different presentation formats using Jade, an open source DSSSL engine. Norm Walsh’s DSSSL stylesheets were
used with an additional customization layer to provide the presentation instructions for Jade.
All of the software used to re-format this document and transform it into different formats is open-source technology,
and having used it for the first time creating this document, I highly recommend it!
167
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement