With an interpreter, the language comes as an environment, where you type in commands at a prompt and the environment executes them for you. For more complicated programs, you can type the commands into a file and get the interpreter to load the file and execute the commands in it. If anything goes wrong, many interpreters will drop you into a debugger to help you track down the problem.

The advantage of this is that you can see the results of your commands immediately, and mistakes can be corrected readily. The biggest disadvantage comes when you want to share your programs with someone. They must have the same interpreter, or you must have some way of giving it to them, and they need to understand how to use it. Also users may not appreciate being thrown into a debugger if they press the wrong key! From a performance point of view, interpreters can use up a lot of memory, and generally do not generate code as efficiently as compilers.

In my opinion, interpreted languages are the best way to start if you have not done any programming before. This kind of environment is typically found with languages like Lisp, Smalltalk, Perl and Basic. It could also be argued that the UNIX® shell (sh, csh) is itself an interpreter, and many people do in fact write shell "scripts" to help with various "housekeeping" tasks on their machine. Indeed, part of the original UNIX® philosophy was to provide lots of small utility programs that could be linked together in shell scripts to perform useful tasks.

Here is a list of interpreters that are available from the FreeBSD Ports Collection, with a brief discussion of some of the more popular interpreted languages.

Instructions on how to get and install applications from the Ports Collection can be found in the Ports section of the handbook.

Compilers are rather different. First of all, you write your code in a file (or files) using an editor. You then run the compiler and see if it accepts your program. If it did not compile, grit your teeth and go back to the editor; if it did compile and gave you a program, you can run it either at a shell command prompt or in a debugger to see if it works properly.^[1]

Obviously, this is not quite as direct as using an interpreter. However it allows you to do a lot of things which are very difficult or even impossible with an interpreter, such as writing code which interacts closely with the operating system-or even writing your own operating system! It is also useful if you need to write very efficient code, as the compiler can take its time and optimize the code, which would not be acceptable in an interpreter. Moreover, distributing a program written for a compiler is usually more straightforward than one written for an interpreter-you can just give them a copy of the executable, assuming they have the same operating system as you.

As the edit-compile-run-debug cycle is rather tedious when using separate programs, many commercial compiler makers have produced Integrated Development Environments (IDEs for short). FreeBSD does not include an IDE in the base system, but devel/kdevelop is available in the Ports Collection and many use Emacs for this purpose. Using Emacs as an IDE is discussed in Using Emacs as a Development Environment.

Most UNIX® systems have a program called test in /usr/bin and the shell is picking that one up before it gets to checking the current directory. Either type:

or choose a better name for your program!

When you are working on a simple program with only one or two source files, typing in

is not too bad, but it quickly becomes very tedious when there are several files-and it can take a while to compile, too.

One way to get around this is to use object files and only recompile the source file if the source code has changed. So we could have something like:

if we had changed file37.c, but not any of the others, since the last time we compiled. This may speed up the compilation quite a bit, but does not solve the typing problem.

Or we could write a shell script to solve the typing problem, but it would have to re-compile everything, making it very inefficient on a large project.

What happens if we have hundreds of source files lying about? What if we are working in a team with other people who forget to tell us when they have changed one of their source files that we use?

Perhaps we could put the two solutions together and write something like a shell script that would contain some kind of magic rule saying when a source file needs compiling. Now all we need now is a program that can understand these rules, as it is a bit too complicated for the shell.

This program is called make. It reads in a file, called a makefile, that tells it how different files depend on each other, and works out which files need to be re-compiled and which ones do not. For example, a rule could say something like "if fromboz.o is older than fromboz.c, that means someone must have changed fromboz.c, so it needs to be re-compiled." The makefile also has rules telling make how to re-compile the source file, making it a much more powerful tool.

Makefiles are typically kept in the same directory as the source they apply to, and can be called makefile, Makefile or MAKEFILE. Most programmers use the name Makefile, as this puts it near the top of a directory listing, where it can easily be seen.^[5]

Here is a very simple make file:

It consists of two lines, a dependency line and a creation line.

The dependency line here consists of the name of the program (known as the target), followed by a colon, then whitespace, then the name of the source file. When make reads this line, it looks to see if foo exists; if it exists, it compares the time foo was last modified to the time foo.c was last modified. If foo does not exist, or is older than foo.c, it then looks at the creation line to find out what to do. In other words, this is the rule for working out when foo.c needs to be re-compiled.

The creation line starts with a tab (press tab) and then the command you would type to create foo if you were doing it at a command prompt. If foo is out of date, or does not exist, make then executes this command to create it. In other words, this is the rule which tells make how to re-compile foo.c.

So, when you type make, it will make sure that foo is up to date with respect to your latest changes to foo.c. This principle can be extended to Makefile's with hundreds of targets-in fact, on FreeBSD, it is possible to compile the entire operating system just by typing make world in the appropriate directory!

Another useful property of makefiles is that the targets do not have to be programs. For instance, we could have a make file that looks like this:

We can tell make which target we want to make by typing:

make will then only look at that target and ignore any others. For example, if we type make foo with the makefile above, make will ignore the install target.

If we just type make on its own, make will always look at the first target and then stop without looking at any others. So if we typed make here, it will just go to the foo target, re-compile foo if necessary, and then stop without going on to the install target.

Notice that the install target does not actually depend on anything! This means that the command on the following line is always executed when we try to make that target by typing make install. In this case, it will copy foo into the user’s home directory. This is often used by application makefiles, so that the application can be installed in the correct directory when it has been correctly compiled.

This is a slightly confusing subject to try to explain. If you do not quite understand how make works, the best thing to do is to write a simple program like "hello world" and a make file like the one above and experiment. Then progress to using more than one source file, or having the source file include a header file. touch is very useful here-it changes the date on a file without you having to edit it.

C code often starts with a list of files to include, for example stdio.h. Some of these files are system-include files, some of them are from the project you are now working on:

To make sure that this file is recompiled the moment foo.h is changed, you have to add it in your Makefile:

The moment your project is getting bigger and you have more and more own include-files to maintain, it will be a pain to keep track of all include files and the files which are depending on it. If you change an include-file but forget to recompile all the files which are depending on it, the results will be devastating. clang has an option to analyze your files and to produce a list of include-files and their dependencies: -MM.

If you add this to your Makefile:

and run make depend, the file .depend will appear with a list of object-files, C-files and the include-files:

If you change foo.h, next time you run make all files depending on foo.h will be recompiled.

Do not forget to run make depend each time you add an include-file to one of your files.

Makefiles can be rather complicated to write. Fortunately, BSD-based systems like FreeBSD come with some very powerful ones as part of the system. One very good example of this is the FreeBSD ports system. Here is the essential part of a typical ports Makefile:

Now, if we go to the directory for this port and type make, the following happens:

Now I think you will agree that is rather impressive for a four line script!

The secret lies in the last line, which tells make to look in the system makefile called bsd.port.mk. It is easy to overlook this line, but this is where all the clever stuff comes from-someone has written a makefile that tells make to do all the things above (plus a couple of other things I did not mention, including handling any errors that may occur) and anyone can get access to that just by putting a single line in their own make file!

If you want to have a look at these system makefiles, they are in /usr/share/mk, but it is probably best to wait until you have had a bit of practice with makefiles, as they are very complicated (and if you do look at them, make sure you have a flask of strong coffee handy!)

Make is a very powerful tool, and can do much more than the simple example above shows. Unfortunately, there are several different versions of make, and they all differ considerably. The best way to learn what they can do is probably to read the documentation-hopefully this introduction will have given you a base from which you can do this. The make(1) manual page offers a comprehensive discussion of variables, arguments, and how to use make.

Many applications in the ports use GNU make, which has a very good set of "info" pages. If you have installed any of these ports, GNU make will automatically have been installed as gmake. It is also available as a port and package in its own right.

To view the info pages for GNU make, you will have to edit dir in the /usr/local/info directory to add an entry for it. This involves adding a line like

to the file. Once you have done this, you can type info and then select make from the menu (or in Emacs, do C-h i).

Using a debugger allows running the program under more controlled circumstances. Typically, it is possible to step through the program a line at a time, inspect the value of variables, change them, tell the debugger to run up to a certain point and then stop, and so on. It is also possible to attach to a program that is already running, or load a core file to investigate why the program crashed.

This section is intended to be a quick introduction to using debuggers and does not cover specialized topics such as debugging the kernel. For more information about that, refer to Kernel Debugging.

The standard debugger supplied with FreeBSD is called lldb (LLVM debugger). As it is part of the standard installation for that release, there is no need to do anything special to use it. It has good command help, accessible via the help command, as well as a web tutorial and documentation.

The other debugger available with FreeBSD is called gdb (GNU debugger). Unlike lldb, it is not installed by default on FreeBSD; to use it, install devel/gdb from ports or packages. It has excellent on-line help, as well as a set of info pages.

The two debuggers have a similar feature set, so which one to use is largely a matter of taste. If familiar with one only, use that one. People familiar with neither or both but wanting to use one from inside Emacs will need to use gdb as lldb is unsupported by Emacs. Otherwise, try both and see which one you prefer.

Starting with LLDB 12.0.0, remote debugging is supported on FreeBSD. This means that lldb-server can be started to debug a program on one host, while the interactive lldb client connects to it from another one.

To launch a new process to be debugged remotely, run lldb-server on the remote server by typing

The process will be stopped immediately after launching, and lldb-server will wait for the client to connect.

Start lldb locally and type the following command to connect to the remote server:

lldb-server can also attach to a running process. To do that, type the following on the remote server:

Emacs is a highly customizable editor-indeed, it has been customized to the point where it is more like an operating system than an editor! Many developers and sysadmins do in fact spend practically all their time working inside Emacs, leaving it only to log out.

It is impossible even to summarize everything Emacs can do here, but here are some of the features of interest to developers:

And doubtless many more that have been overlooked.

Emacs can be installed on FreeBSD using the editors/emacs port.

Once it is installed, start it up and do C-h t to read an Emacs tutorial-that means hold down control, press h, let go of control, and then press t. (Alternatively, you can use the mouse to select Emacs Tutorial from the Help menu.)

Although Emacs does have menus, it is well worth learning the key bindings, as it is much quicker when you are editing something to press a couple of keys than to try to find the mouse and then click on the right place. And, when you are talking to seasoned Emacs users, you will find they often casually throw around expressions like “M-x replace-s RET foo RET bar RET” so it is useful to know what they mean. And in any case, Emacs has far too many useful functions for them to all fit on the menu bars.

Fortunately, it is quite easy to pick up the key-bindings, as they are displayed next to the menu item. My advice is to use the menu item for, say, opening a file until you understand how it works and feel confident with it, then try doing C-x C-f. When you are happy with that, move on to another menu command.

If you cannot remember what a particular combination of keys does, select Describe Key from the Help menu and type it in-Emacs will tell you what it does. You can also use the Command Apropos menu item to find out all the commands which contain a particular word in them, with the key binding next to it.

By the way, the expression above means hold down the Meta key, press x, release the Meta key, type replace-s (short for replace-string-another feature of Emacs is that you can abbreviate commands), press the return key, type foo (the string you want replaced), press the return key, type bar (the string you want to replace foo with) and press return again. Emacs will then do the search-and-replace operation you have just requested.

If you are wondering what on earth Meta is, it is a special key that many UNIX® workstations have. Unfortunately, PC’s do not have one, so it is usually alt (or if you are unlucky, the escape key).

Oh, and to get out of Emacs, do C-x C-c (that means hold down the control key, press x, press c and release the control key). If you have any unsaved files open, Emacs will ask you if you want to save them. (Ignore the bit in the documentation where it says C-z is the usual way to leave Emacs-that leaves Emacs hanging around in the background, and is only really useful if you are on a system which does not have virtual terminals).

Emacs does many wonderful things; some of them are built in, some of them need to be configured.

Instead of using a proprietary macro language for configuration, Emacs uses a version of Lisp specially adapted for editors, known as Emacs Lisp. Working with Emacs Lisp can be quite helpful if you want to go on and learn something like Common Lisp. Emacs Lisp has many features of Common Lisp, although it is considerably smaller (and thus easier to master).

The best way to learn Emacs Lisp is to read the online Emacs Reference manual.

However, there is no need to actually know any Lisp to get started with configuring Emacs, as I have included a sample .emacs, which should be enough to get you started. Just copy it into your home directory and restart Emacs if it is already running; it will read the commands from the file and (hopefully) give you a useful basic setup.

Unfortunately, there is far too much here to explain it in detail; however there are one or two points worth mentioning.

Now, this is all very well if you only want to program in the languages already catered for in .emacs (C, C++, Perl, Lisp and Scheme), but what happens if a new language called "whizbang" comes out, full of exciting features?

The first thing to do is find out if whizbang comes with any files that tell Emacs about the language. These usually end in .el, short for "Emacs Lisp". For example, if whizbang is a FreeBSD port, we can locate these files by doing

and install them by copying them into the Emacs site Lisp directory. On FreeBSD, this is /usr/local/share/emacs/site-lisp.

So for example, if the output from the find command was

we would do

Next, we need to decide what extension whizbang source files have. Let us say for the sake of argument that they all end in .wiz. We need to add an entry to our .emacs to make sure Emacs will be able to use the information in whizbang.el.

Find the auto-mode-alist entry in .emacs and add a line for whizbang, such as:

This means that Emacs will automatically go into whizbang-mode when you edit a file ending in .wiz.

Just below this, you will find the font-lock-auto-mode-list entry. Add whizbang-mode to it like so:

This means that Emacs will always enable font-lock-mode (ie syntax highlighting) when editing a .wiz file.

And that is all that is needed. If there is anything else you want done automatically when you open up .wiz, you can add a whizbang-mode hook (see my-scheme-mode-hook for a simple example that adds auto-indent).

The following example code contains a buffer overflow designed to overwrite the return address and skip the instruction immediately following the function call. (Inspired by 4)

Let us examine what the memory image of this process would look like if we were to input 160 spaces into our little program before hitting return.

Obviously more malicious input can be devised to execute actual compiled instructions (such as exec(/bin/sh)).

The most straightforward solution to the problem of stack-overflows is to always use length restricted memory and string copy functions. strncpy and strncat are part of the standard C library. These functions accept a length value as a parameter which should be no larger than the size of the destination buffer. These functions will then copy up to 'length' bytes from the source to the destination. However there are a number of problems with these functions. Neither function guarantees NUL termination if the size of the input buffer is as large as the destination. The length parameter is also used inconsistently between strncpy and strncat so it is easy for programmers to get confused as to their proper usage. There is also a significant performance loss compared to strcpy when copying a short string into a large buffer since strncpy NUL fills up the size specified.

Another memory copy implementation exists to get around these problems. The strlcpy and strlcat functions guarantee that they will always null terminate the destination string when given a non-zero length argument.

The concept of a Jail extends upon the chroot() by limiting the powers of the superuser to create a true `virtual server'. Once a prison is set up all network communication must take place through the specified IP address, and the power of "root privilege" in this jail is severely constrained.

While in a prison, any tests of superuser power within the kernel using the suser() call will fail. However, some calls to suser() have been changed to a new interface suser_xxx(). This function is responsible for recognizing or denying access to superuser power for imprisoned processes.

A superuser process within a jailed environment has the power to:

Jail is a very useful tool for running applications in a secure environment but it does have some shortcomings. Currently, the IPC mechanisms have not been converted to the suser_xxx so applications such as MySQL cannot be run within a jail. Superuser access may have a very limited meaning within a jail, but there is no way to specify exactly what "very limited" means.

POSIX® has released a working draft that adds event auditing, access control lists, fine grained privileges, information labeling, and mandatory access control.

This is a work in progress and is the focus of the TrustedBSD project. Some of the initial work has been committed to FreeBSD-CURRENT (cap_set_proc(3)).

It has come to our attention that the individual I18N/L10N efforts for each country has been repeating each others' efforts. Many of us have been reinventing the wheel repeatedly and inefficiently. We hope that the various major groups in I18N could congregate into a group effort similar to the Core Team’s responsibility.

Currently, we hope that, when you write or port I18N programs, you would send it out to each country’s related FreeBSD mailing list for testing. In the future, we hope to create applications that work in all the languages out-of-the-box without dirty hacks.

The FreeBSD internationalization mailing list has been established. If you are an I18N/L10N developer, please send your comments, ideas, questions, and anything you deem related to it.

Perl and Python have I18N and wide character handling libraries. Please use them for I18N compliance.

POSIX.1 NLS is based on catalog files, which contain the localized messages in the desired encoding. The messages are organized into sets and each message is identified by an integer number in the containing set. The catalog files are conventionally named after the locale they contain localized messages for, followed by the .msg extension. For instance, the Hungarian messages for ISO8859-2 encoding should be stored in a file called hu_HU.ISO8859-2.

These catalog files are common text files that contain the numbered messages. It is possible to write comments by starting the line with a $ sign. Set boundaries are also separated by special comments, where the keyword set must directly follow the $ sign. The set keyword is then followed by the set number. For example:

The actual message entries start with the message number and followed by the localized message. The well-known modifiers from printf(3) are accepted:

The language catalog files have to be compiled into a binary form before they can be opened from the program. This conversion is done with the gencat(1) utility. Its first argument is the filename of the compiled catalog and its further arguments are the input catalogs. The localized messages can also be organized into more catalog files and then all of them can be processed with gencat(1).

Using the catalog files is simple. To use the related functions, nl_types.h must be included. Before using a catalog, it has to be opened with catopen(3). The function takes two arguments. The first parameter is the name of the installed and compiled catalog. Usually, the name of the program is used, such as grep. This name will be used when looking for the compiled catalog file. The catopen(3) call looks for this file in /usr/share/nls/locale/catname and in /usr/local/share/nls/locale/catname, where locale is the locale set and catname is the catalog name being discussed. The second parameter is a constant, which can have two values:

The catopen(3) call returns a catalog identifier of type nl_catd. Please refer to the manual page for a list of possible returned error codes.

After opening a catalog catgets(3) can be used to retrieve a message. The first parameter is the catalog identifier returned by catopen(3), the second one is the number of the set, the third one is the number of the messages, and the fourth one is a fallback message, which will be returned if the requested message cannot be retrieved from the catalog file.

After using the catalog file, it must be closed by calling catclose(3), which has one argument, the catalog id.

The following example will demonstrate an easy solution on how to use NLS catalogs in a flexible way.

The below lines need to be put into a common header file of the program, which is included into all source files where localized messages are necessary:

Next, put these lines into the global declaration part of the main source file:

Next come the real code snippets, which open, read, and close the catalog:

Using the catalog files requires few repeatable steps, such as compiling the catalogs and installing them to the proper location. In order to simplify this process even more, bsd.nls.mk introduces some macros. It is not necessary to include bsd.nls.mk explicitly, it is pulled in from the common Makefiles, such as bsd.prog.mk or bsd.lib.mk.

Usually it is enough to define NLSNAME, which should have the catalog name mentioned as the first argument of catopen(3) and list the catalog files in NLS without their .msg extension. Here is an example, which makes it possible to to disable NLS when used with the code examples before. The WITHOUT_NLS make(1) variable has to be defined in order to build the program without NLS support.

Conventionally, the catalog files are placed under the nls subdirectory and this is the default behavior of bsd.nls.mk. It is possible, though to override the location of the catalogs with the NLSSRCDIR make(1) variable. The default name of the precompiled catalog files also follow the naming convention mentioned before. It can be overridden by setting the NLSNAME variable. There are other options to fine tune the processing of the catalog files but usually it is not needed, thus they are not described here. For further information on bsd.nls.mk, please refer to the file itself, it is short and easy to understand.

The standard process for managing contributed software and vendor branches is described in detail by the Committer’s Guide.

Typically, one of the ends of a socket-based data communication is a server, the other is a client.

While there is no way to pass the domain name directly to any of the sockets functions, the FreeBSD C library comes with the gethostbyname(3) and gethostbyname2(3) functions, declared in netdb.h.

Both return a pointer to the hostent structure, with much information about the domain. For our purposes, the h_addr_list[0] field of the structure points at h_length bytes of the correct address, already stored in the network byte order.

This allows us to create a much more flexible-and much more useful-version of our daytime program:

We now can type a domain name (or an IP address, it works both ways) on the command line, and the program will try to connect to its daytime server. Otherwise, it will still default to time.nist.gov. However, even in this case we will use gethostbyname rather than hard coding 192.43.244.18. That way, even if its IP address changes in the future, we will still find it.

Since it takes virtually no time to get the time from your local server, you could run daytime twice in a row: First to get the time from time.nist.gov, the second time from your own system. You can then compare the results and see how exact your system clock is:

As you can see, my system was two seconds ahead of the NIST time.

Sometimes you may not be sure what port a certain service uses. The getservbyname(3) function, also declared in netdb.h comes in very handy in those cases:

The servent structure contains the s_port, which contains the proper port, already in network byte order.

Had we not known the correct port for the daytime service, we could have found it this way:

You usually do know the port. But if you are developing a new protocol, you may be testing it on an unofficial port. Some day, you will register the protocol and its port (if nowhere else, at least in your /etc/services, which is where getservbyname looks). Instead of returning an error in the above code, you just use the temporary port number. Once you have listed the protocol in /etc/services, your software will find its port without you having to rewrite the code.

Now following two items are required to be supported by standard drivers:

If any of the drivers do not support the requirements, then the drivers cannot be used for IPv6 and/or IPsec communication. If you find any problem with your card using IPv6/IPsec, then, please report it to the FreeBSD problem reports mailing list.

(NOTE: In the past we required all PCMCIA drivers to have a call to in6_ifattach(). We have no such requirement any more)

We categorize IPv4/IPv6 translator into 4 types:

IPsec is mainly organized by three components.

Before the kernel will dump the contents of its physical memory to a dump device, a dump device must be configured. A dump device is specified by using the dumpon(8) command to tell the kernel where to save kernel crash dumps. The dumpon(8) program must be called after the swap partition has been configured with swapon(8). This is normally handled by setting the dumpdev variable in rc.conf(5) to the path of the swap device (the recommended way to extract a kernel dump) or AUTO to use the first configured swap device. The default for dumpdev is AUTO in HEAD, and changed to NO on RELENG_* branches (except for RELENG_7, which was left set to AUTO). On FreeBSD 9.0-RELEASE and later versions, bsdinstall will ask whether crash dumps should be enabled on the target system during the install process.

Once a dump has been written to a dump device, the dump must be extracted before the swap device is mounted. To extract a dump from a dump device, use the savecore(8) program. If dumpdev has been set in rc.conf(5), savecore(8) will be called automatically on the first multi-user boot after the crash and before the swap device is mounted. The location of the extracted core is placed in the rc.conf(5) value dumpdir, by default /var/crash and will be named vmcore.0.

In the event that there is already a file called vmcore.0 in /var/crash (or whatever dumpdir is set to), the kernel will increment the trailing number for every crash to avoid overwriting an existing vmcore (e.g., vmcore.1). savecore(8) will always create a symbolic link to named vmcore.last in /var/crash after a dump is saved. This symbolic link can be used to locate the name of the most recent dump.

The crashinfo(8) utility generates a text file containing a summary of information from a full memory dump or minidump. If dumpdev has been set in rc.conf(5), crashinfo(8) will be invoked automatically after savecore(8). The output is saved to a file in dumpdir named core.txt.N.

The kernel includes a sysctl(8) node that requests a kernel panic. This can be used to verify that your system is properly configured to save kernel crash dumps. You may wish to remount existing file systems as read-only in single user mode before triggering the crash to avoid data loss.

After rebooting, your system should save a dump in /var/crash along with a matching summary from crashinfo(8).

Most FireWire® (IEEE1394) host controllers are based on the OHCI specification that supports physical access to the host memory. This means that once the host controller is initialized, we can access the host memory without the help of software (kernel). We can exploit this facility for interaction with dcons(4). dcons(4) provides similar functionality as a serial console. It emulates two serial ports, one for the console and DDB, the other for GDB. Since remote memory access is fully handled by the hardware, the dcons(4) buffer is accessible even when the system crashes.

FireWire® devices are not limited to those integrated into motherboards. PCI cards exist for desktops, and a cardbus interface can be purchased for laptops.

We can directly read the dcons(4) buffer via /dev/mem for live systems, and in the core dump for crashed systems. These give you similar output to dmesg -a, but the dcons(4) buffer includes more information.

The most important tool for assembly language programming is the assembler, the software that converts assembly language code into machine language.

Three very different assemblers are available for FreeBSD. Both llvm-as(1) (included in devel/llvm) and as(1) (included in devel/binutils) use the traditional UNIX® assembly language syntax.

On the other hand, nasm(1) (installed through devel/nasm) uses the Intel syntax. Its main advantage is that it can assemble code for many operating systems.

This chapter uses nasm syntax because most assembly language programmers coming to FreeBSD from other operating systems will find it easier to understand. And, because, quite frankly, that is what I am used to.

The output of the assembler, like that of any compiler, needs to be linked to form an executable file.

The standard ld(1) linker comes with FreeBSD. It works with the code assembled with either assembler.

By default, the FreeBSD kernel uses the C calling convention. Further, although the kernel is accessed using int 80h, it is assumed the program will call a function that issues int 80h, rather than issuing int 80h directly.

This convention is very convenient, and quite superior to the Microsoft® convention used by MS-DOS®. Why? Because the UNIX® convention allows any program written in any language to access the kernel.

An assembly language program can do that as well. For example, we could open a file:

This is a very clean and portable way of coding. If you need to port the code to a UNIX® system which uses a different interrupt, or a different way of passing parameters, all you need to change is the kernel procedure.

But assembly language programmers like to shave off cycles. The above example requires a call/ret combination. We can eliminate it by pushing an extra dword:

The 5 that we have placed in EAX identifies the kernel function, in this case open.

FreeBSD is an extremely flexible system. It offers other ways of calling the kernel. For it to work, however, the system must have Linux emulation installed.

Linux is a UNIX® like system. However, its kernel uses the same system-call convention of passing parameters in registers MS-DOS® does. As with the UNIX® convention, the function number is placed in EAX. The parameters, however, are not passed on the stack but in EBX, ECX, EDX, ESI, EDI, EBP:

This convention has a great disadvantage over the UNIX® way, at least as far as assembly language programming is concerned: Every time you make a kernel call you must push the registers, then pop them later. This makes your code bulkier and slower. Nevertheless, FreeBSD gives you a choice.

If you do choose the Linux convention, you must let the system know about it. After your program is assembled and linked, you need to brand the executable:

If you are coding specifically for FreeBSD, you should always use the UNIX® convention: It is faster, you can store global variables in registers, you do not have to brand the executable, and you do not impose the installation of the Linux emulation package on the target system.

If you want to create portable code that can also run on Linux, you will probably still want to give the FreeBSD users as efficient a code as possible. I will show you how you can accomplish that after I have explained the basics.

To tell the kernel which system service you are calling, place its number in EAX. Of course, you need to know what the number is.

The traditional place to look for information about various system calls under UNIX® systems are the manual pages. FreeBSD describes its system calls in section 2, sometimes in section 3.

For example, open(2) says:

The assembly language programmer new to UNIX® and FreeBSD will immediately ask the puzzling question: Where is errno and how do I get to it?

Unfortunately, it depends…​ For most system calls it is in EAX, but not for all. A good rule of thumb, when working with a system call for the first time, is to look for the return value in EAX. If it is not there, you need further research.

Actually, nowhere…​

errno is part of the C language, not the UNIX® kernel. When accessing kernel services directly, the error code is returned in EAX, the same register the proper return value generally ends up in.

This makes perfect sense. If there is no error, there is no error code. If there is an error, there is no return value. One register can contain either.

When using the standard FreeBSD calling convention, the carry flag is cleared upon success, set upon failure.

When using the Linux emulation mode, the signed value in EAX is non-negative upon success, and contains the return value. In case of an error, the value is negative, i.e., -errno.

In many cases the function numbers are the same. However, even when they are not, the problem is easy to deal with: Instead of using numbers in your code, use constants which you have declared differently depending on the target architecture:

Both, the calling convention, and the return value (the errno problem) can be resolved with macros:

The above solutions can handle most cases of writing code portable between FreeBSD and Linux. Nevertheless, with some kernel services the differences are deeper.

In that case, you need to write two different handlers for those particular system calls, and use conditional assembly. Luckily, most of your code does something other than calling the kernel, so usually you will only need a few such conditional sections in your code.

You can avoid portability issues in your main code altogether by writing a library of system calls. Create a separate library for FreeBSD, a different one for Linux, and yet other libraries for more operating systems.

In your library, write a separate function (or procedure, if you prefer the traditional assembly language terminology) for each system call. Use the C calling convention of passing parameters. But still use EAX to pass the call number in. In that case, your FreeBSD library can be very simple, as many seemingly different functions can be just labels to the same code:

Your Linux library will require more different functions. But even here you can group system calls using the same number of parameters:

The library approach may seem inconvenient at first because it requires you to produce a separate file your code depends on. But it has many advantages: For one, you only need to write it once and can use it for all your programs. You can even let other assembly language programmers use it, or perhaps use one written by someone else. But perhaps the greatest advantage of the library is that your code can be ported to other systems, even by other programmers, by simply writing a new library without any changes to your code.

If you do not like the idea of having a library, you can at least place all your system calls in a separate assembly language file and link it with your main program. Here, again, all porters have to do is create a new object file to link with your main program.

If you are releasing your software as (or with) source code, you can use macros and place them in a separate file, which you include in your code.

Porters of your software will simply write a new include file. No library or external object file is necessary, yet your code is portable without any need to edit the code.

We can start our system.inc by declaring the standard file descriptors:

Next, we create a symbolic name for each system call:

We add a short, non-global procedure with a long name, so we do not accidentally reuse the name in our code:

We create a macro which takes one argument, the syscall number:

Finally, we create macros for each syscall. These macros take no arguments.

Go ahead, enter it into your editor and save it as system.inc. We will add more to it as we discuss more syscalls.

Type the code in an editor, and save it in a file named hello.asm. You need nasm to assemble it.

While our sample program does not require it, more sophisticated filters often need to look ahead. In other words, they may need to see what the next character is (or even several characters). If the next character is of a certain value, it is part of the token currently being processed. Otherwise, it is not.

For example, you may be parsing the input stream for a textual string (e.g., when implementing a language compiler): If a character is followed by another character, or perhaps a digit, it is part of the token you are processing. If it is followed by white space, or some other value, then it is not part of the current token.

This presents an interesting problem: How to return the next character back to the input stream, so it can be read again later?

One possible solution is to store it in a character variable, then set a flag. We can modify getchar to check the flag, and if it is set, fetch the byte from that variable instead of the input buffer, and reset the flag. But, of course, that slows us down.

The C language has an ungetc() function, just for that purpose. Is there a quick way to implement it in our code? I would like you to scroll back up and take a look at the getchar procedure and see if you can find a nice and fast solution before reading the next paragraph. Then come back here and see my own solution.

The key to returning a character back to the stream is in how we are getting the characters to start with:

First we check if the buffer is empty by testing the value of EBX. If it is zero, we call the read procedure.

If we do have a character available, we use lodsb, then decrease the value of EBX. The lodsb instruction is effectively identical to:

The byte we have fetched remains in the buffer until the next time read is called. We do not know when that happens, but we do know it will not happen until the next call to getchar. Hence, to "return" the last-read byte back to the stream, all we have to do is decrease the value of ESI and increase the value of EBX:

But, be careful! We are perfectly safe doing this if our look-ahead is at most one character at a time. If we are examining more than one upcoming character and call ungetc several times in a row, it will work most of the time, but not all the time (and will be tough to debug). Why?

Because as long as getchar does not have to call read, all of the pre-read bytes are still in the buffer, and our ungetc works without a glitch. But the moment getchar calls read, the contents of the buffer change.

We can always rely on ungetc working properly on the last character we have read with getchar, but not on anything we have read before that.

If your program reads more than one byte ahead, you have at least two choices:

If possible, modify the program so it only reads one byte ahead. This is the simplest solution.

If that option is not available, first of all determine the maximum number of characters your program needs to return to the input stream at one time. Increase that number slightly, just to be sure, preferably to a multiple of 16-so it aligns nicely. Then modify the .bss section of your code, and create a small "spare" buffer right before your input buffer, something like this:

You also need to modify your ungetc to pass the value of the byte to unget in AL:

With this modification, you can call ungetc up to 17 times in a row safely (the first call will still be within the buffer, the remaining 16 may be either within the buffer or within the "spare").

I said earlier that when a program starts executing, the stack contains argc followed by the NULL-terminated argv array, followed by something else. The "something else" is the environment, or, to be more precise, a NULL-terminated array of pointers to environment variables. This is often referred to as env.

The structure of env is the same as that of argv, a list of memory addresses followed by a NULL (0). In this case, there is no "envc"-we figure out where the array ends by searching for the final NULL.

The variables usually come in the name=value format, but sometimes the =value part may be missing. We need to account for that possibility.

I could just show you some code that prints the environment the same way the UNIX® env command does. But I thought it would be more interesting to write a simple assembly language CGI utility.

The problem is easily solved by the use of a technique called finite state machine, originally developed by the designers of digital electronic circuits. A finite state machine is a digital circuit whose output is dependent not only on its input but on its previous input, i.e., on its state. The microprocessor is an example of a finite state machine: Our assembly language code is assembled to machine language in which some assembly language code produces a single byte of machine language, while others produce several bytes. As the microprocessor fetches the bytes from the memory one by one, some of them simply change its state rather than produce some output. When all the bytes of the op code are fetched, the microprocessor produces some output, or changes the value of a register, etc.

Because of that, all software is essentially a sequence of state instructions for the microprocessor. Nevertheless, the concept of finite state machine is useful in software design as well.

Our text file converter can be designer as a finite state machine with three possible states. We could call them states 0-2, but it will make our life easier if we give them symbolic names:

Our program will start in the ordinary state. During this state, the program action depends on its input as follows:

Whenever we are in the cr state, it is because the last input was a carriage return, which was unprocessed. What our software does in this state again depends on the current input:

Finally, we are in the lf state after we have received a line feed that was not preceded by a carriage return. This will happen when our file already is in UNIX® format, or whenever several lines in a row are expressed by a single carriage return followed by several line feeds, or when line ends with a line feed / carriage return sequence. Here is how we need to handle our input in this state:

The hardest part of working with a finite state machine is analyzing the problem and expressing it as a finite state machine. That accomplished, the software almost writes itself.

In a high-level language, such as C, there are several main approaches. One is to use a switch statement which chooses what function should be run. For example,

Another approach is by using an array of function pointers, something like this:

Yet another is to have state be a function pointer, set to point at the appropriate function:

This is the approach we will use in our program because it is very easy to do in assembly language, and very fast, too. We will simply keep the address of the right procedure in EBX, and then just issue:

This is possibly faster than hardcoding the address in the code because the microprocessor does not have to fetch the address from the memory-it is already stored in one of its registers. I said possibly because with the caching modern microprocessors do, either way may be equally fast.

Because our program works on a single file, we cannot use the approach that worked for us before, i.e., to read from an input file and to write to an output file.

UNIX® allows us to map a file, or a section of a file, into memory. To do that, we first need to open the file with the appropriate read/write flags. Then we use the mmap system call to map it into the memory. One nice thing about mmap is that it automatically works with virtual memory: We can map more of the file into the memory than we have physical memory available, yet still access it through regular memory op codes, such as mov, lods, and stos. Whatever changes we make to the memory image of the file will be written to the file by the system. We do not even have to keep the file open: As long as it stays mapped, we can read from it and write to it.

The 32-bit Intel microprocessors can access up to four gigabytes of memory - physical or virtual. The FreeBSD system allows us to use up to a half of it for file mapping.

For simplicity sake, in this tutorial we will only convert files that can be mapped into the memory in their entirety. There are probably not too many text files that exceed two gigabytes in size. If our program encounters one, it will simply display a message suggesting we use the original tuc instead.

If you examine your copy of syscalls.master, you will find two separate syscalls named mmap. This is because of evolution of UNIX®: There was the traditional BSD mmap, syscall 71. That one was superseded by the POSIX® mmap, syscall 197. The FreeBSD system supports both because older programs were written by using the original BSD version. But new software uses the POSIX® version, which is what we will use.

The syscalls.master lists the POSIX® version like this:

This differs slightly from what mmap(2) says. That is because mmap(2) describes the C version.

The difference is in the long pad argument, which is not present in the C version. However, the FreeBSD syscalls add a 32-bit pad after pushing a 64-bit argument. In this case, off_t is a 64-bit value.

When we are finished working with a memory-mapped file, we unmap it with the munmap syscall:

Because we need to tell mmap how many bytes of the file to map into the memory, and because we want to map the entire file, we need to determine the size of the file.

We can use the fstat syscall to get all the information about an open file that the system can give us. That includes the file size.

Again, syscalls.master lists two versions of fstat, a traditional one (syscall 62), and a POSIX® one (syscall 189). Naturally, we will use the POSIX® version:

This is a very straightforward call: We pass to it the address of a stat structure and the descriptor of an open file. It will fill out the contents of the stat structure.

I do, however, have to say that I tried to declare the stat structure in the .bss section, and fstat did not like it: It set the carry flag indicating an error. After I changed the code to allocate the structure on the stack, everything was working fine.

Because our program may combine carriage return / line feed sequences into straight line feeds, our output may be smaller than our input. However, since we are placing our output into the same file we read the input from, we may have to change the size of the file.

The ftruncate system call allows us to do just that. Despite its somewhat misleading name, the ftruncate system call can be used to both truncate the file (make it smaller) and to grow it.

And yes, we will find two versions of ftruncate in syscalls.master, an older one (130), and a newer one (201). We will use the newer one:

Please note that this one contains a int pad again.

We now know everything we need to write ftuc. We start by adding some new lines in system.inc. First, we define some constants and structures, somewhere at or near the beginning of the file:

We define the new syscalls:

We add the macros for their use:

And here is our code:

I will illustrate this principle with a specific real-life example I was faced with recently:

I needed to extract the 11th field of each record from a database I downloaded from a web site. The database was a CSV file, i.e., a list of comma-separated values. That is quite a standard format for sharing data among people who may be using different database software.

The first line of the file contains the list of various fields separated by commas. The rest of the file contains the data listed line by line, with values separated by commas.

I tried awk, using the comma as a separator. But because several lines contained a quoted comma, awk was extracting the wrong field from those lines.

Therefore, I needed to write my own software to extract the 11th field from the CSV file. However, going with the UNIX® spirit, I only needed to write a simple filter that would do the following:

Strictly speaking, I could use sed to remove the first line from the file, but doing so in my own program was very easy, so I decided to do it and reduce the size of the pipeline.

At any rate, writing a program like this took me about 20 minutes. Writing a program that extracts the 11th field from the CSV file would take a lot longer, and I could not reuse it to extract some other field from some other database.

This time I decided to let it do a little more work than a typical tutorial program would:

Here is its usage message:

All parameters are optional, and can appear in any order.

The -t parameter declares what to replace the commas with. The tab is the default here. For example, -t; will replace all unquoted commas with semicolons.

I did not need the -c option, but it may come in handy in the future. It lets me declare that I want a character other than a comma replaced with something else. For example, -c@ will replace all at signs (useful if you want to split a list of email addresses to their user names and domains).

The -p option preserves the first line, i.e., it does not delete it. By default, we delete the first line because in a CSV file it contains the field names rather than data.

The -i and -o options let me specify the input and the output files. Defaults are stdin and stdout, so this is a regular UNIX® filter.

I made sure that both -i filename and -ifilename are accepted. I also made sure that only one input and one output files may be specified.

To get the 11th field of each record, I can now do:

The code stores the options (except for the file descriptors) in EDX: The comma in DH, the new separator in DL, and the flag for the -p option in the highest bit of EDX, so a check for its sign will give us a quick decision what to do.

Here is the code:

Much of it is taken from hex.asm above. But there is one important difference: I no longer call write whenever I am outputting a line feed. Yet, the code can be used interactively.

I have found a better solution for the interactive problem since I first started writing this chapter. I wanted to make sure each line is printed out separately only when needed. After all, there is no need to flush out every line when used non-interactively.

The new solution I use now is to call write every time I find the input buffer empty. That way, when running in the interactive mode, the program reads one line from the user’s keyboard, processes it, and sees its input buffer is empty. It flushes its output and reads the next line.

The FPU consists of 8 80-bit floating-point registers. These are organized in a stack fashion-you can push a value on TOS (top of stack) and you can pop it.

That said, the assembly language op codes are not push and pop because those are already taken.

You can push a value on TOS by using fld, fild, and fbld. Several other op codes let you push many common constants-such as pi-on the TOS.

Similarly, you can pop a value by using fst, fstp, fist, fistp, and fbstp. Actually, only the op codes that end with a p will literally pop the value, the rest will store it somewhere else without removing it from the TOS.

We can transfer the data between the TOS and the computer memory either as a 32-bit, 64-bit, or 80-bit real, a 16-bit, 32-bit, or 64-bit integer, or an 80-bit packed decimal.

The 80-bit packed decimal is a special case of binary coded decimal which is very convenient when converting between the ASCII representation of data and the internal data of the FPU. It allows us to use 18 significant digits.

No matter how we represent data in the memory, the FPU always stores it in the 80-bit real format in its registers.

Its internal precision is at least 19 decimal digits, so even if we choose to display results as ASCII in the full 18-digit precision, we are still showing correct results.

We can perform mathematical operations on the TOS: We can calculate its sine, we can scale it (i.e., we can multiply or divide it by a power of 2), we can calculate its base-2 logarithm, and many other things.

We can also multiply or divide it by, add it to, or subtract it from, any of the FPU registers (including itself).

The official Intel op code for the TOS is st, and for the registers st(0)-st(7). st and st(0), then, refer to the same register.

For whatever reasons, the original author of nasm has decided to use different op codes, namely st0-st7. In other words, there are no parentheses, and the TOS is always st0, never just st.

To write meaningful software, we must not only understand our programming tools, but also the field we are creating software for.

Our next filter will help us whenever we want to build a pinhole camera, so, we need some background in pinhole photography before we can continue.

We are now ready to decide what exactly we want our pinhole software to do.

In assembly language we can optimize the FPU code in ways impossible in high languages, including C.

Whenever a C function needs to calculate a floating-point value, it loads all necessary variables and constants into FPU registers. It then does whatever calculation is required to get the correct result. Good C compilers can optimize that part of the code really well.

It "returns" the value by leaving the result on the TOS. However, before it returns, it cleans up. Any variables and constants it used in its calculation are now gone from the FPU.

It cannot do what we just did above: We calculated the square of the f-number and kept it on the stack for later use by another function.

We knew we would need that value later on. We also knew we had enough room on the stack (which only has room for 8 numbers) to store it there.

A C compiler has no way of knowing that a value it has on the stack will be required again in the very near future.

Of course, the C programmer may know it. But the only recourse he has is to store the value in a memory variable.

That means, for one, the value will be changed from the 80-bit precision used internally by the FPU to a C double (64 bits) or even single (32 bits).

That also means that the value must be moved from the TOS into the memory, and then back again. Alas, of all FPU operations, the ones that access the computer memory are the slowest.

So, whenever programming the FPU in assembly language, look for the ways of keeping intermediate results on the FPU stack.

We can take that idea even further! In our program we are using a constant (the one we named PC).

It does not matter how many pinhole diameters we are calculating: 1, 10, 20, 1000, we are always using the same constant. Therefore, we can optimize our program by keeping the constant on the stack all the time.

Early on in our program, we are calculating the value of the above constant. We need to divide our input by 10 for every digit in the constant.

It is much faster to multiply than to divide. So, at the start of our program, we divide 10 into 1 to obtain 0.1, which we then keep on the stack: Instead of dividing the input by 10 for every digit, we multiply it by 0.1.

By the way, we do not input 0.1 directly, even though we could. We have a reason for that: While 0.1 can be expressed with just one decimal place, we do not know how many binary places it takes. We, therefore, let the FPU calculate its binary value to its own high precision.

We are using other constants: We multiply the pinhole diameter by 1000 to convert it from millimeters to microns. We compare numbers to 10000 when we are rounding them off to four significant digits. So, we keep both, 1000 and 10000, on the stack. And, of course, we reuse the 0.1 when rounding off numbers to four digits.

Last but not least, we keep -5 on the stack. We need it to scale the square of the f-number, instead of dividing it by 32. It is not by coincidence we load this constant last. That makes it the top of the stack when only the constants are on it. So, when the square of the f-number is being scaled, the -5 is at st(1), precisely where fscale expects it to be.

It is common to create certain constants from scratch instead of loading them from the memory. That is what we are doing with -5:

We can generalize all these optimizations into one rule: Keep repeat values on the stack!

The code follows the same format as all the other filters we have seen before, with one subtle exception:

Other than that, the bcdload routine is caught up in the middle of a Lilliputian conflict between the Big-Endians and the Little-Endians.

It is converting the text representation of a number into that number: The text is stored in the big-endian order, but the packed decimal is little-endian.

To solve the conflict, we use the std op code early on. We cancel it with cld later on: It is quite important we do not call anything that may depend on the default setting of the direction flag while std is active.

Everything else in this code should be quit eclear, providing you have read the entire chapter that precedes it.

It is a classical example of the adage that programming requires a lot of thought and only a little coding. Once we have thought through every tiny detail, the code almost writes itself.

Because we have decided to make the program ignore any input except for numbers (and even those inside a comment), we can actually perform textual queries. We do not have to, but we can.

In my humble opinion, forming a textual query, instead of having to follow a very strict syntax, makes software much more user friendly.

Suppose we want to build a pinhole camera to use the 4x5 inch film. The standard focal length for that film is about 150mm. We want to fine-tune our focal length so the pinhole diameter is as round a number as possible. Let us also suppose we are quite comfortable with cameras but somewhat intimidated by computers. Rather than just have to type in a bunch of numbers, we want to ask a couple of questions.

Our session might look like this:

We have found that while for the focal length of 150, our pinhole diameter should be 490 microns, or 0.49 mm, if we go with the almost identical focal length of 156 mm, we can get away with a pinhole diameter of exactly one half of a millimeter.

Because we have chosen the # character to denote the start of a comment, we can treat our pinhole software as a scripting language.

You have probably seen shell scripts that start with:

…​or…​

…​because the blank space after the #! is optional.

Whenever UNIX® is asked to run an executable file which starts with the #!, it assumes the file is a script. It adds the command to the rest of the first line of the script, and tries to execute that.

Suppose now that we have installed pinhole in /usr/local/bin/, we can now write a script to calculate various pinhole diameters suitable for various focal lengths commonly used with the 120 film.

The script might look something like this:

Because 120 is a medium size film, we may name this file medium.

We can set its permissions to execute, and run it as if it were a program:

UNIX® will interpret that last command as:

It will run that command and display:

Now, let us enter:

UNIX® will treat that as:

That gives it two conflicting options: -b and -c (Use Bender’s constant and use Connors' constant). We have programmed it so later options override early ones-our program will calculate everything using Connors' constant:

We decide we want to go with Bender’s constant after all. We want to save its values as a comma-separated file:

For one thing, it may simply not be possible. UNIX® runs in protected mode. Only the kernel and device drivers are allowed to access hardware directly. Perhaps a particular UNIX® clone will let you read the keyboard scan codes, but chances are a real UNIX® operating system will not. And even if one version may let you do it, the next one may not, so your carefully crafted software may become a dinosaur overnight.

But there is a much more important reason not to try accessing the hardware directly (unless, of course, you are writing a device driver), even on the UNIX® like systems that let you do it:

UNIX® is an abstraction!

There is a major difference in the philosophy of design between MS-DOS® and UNIX®. MS-DOS® was designed as a single-user system. It is run on a computer with a keyboard and a video screen attached directly to that computer. User input is almost guaranteed to come from that keyboard. Your program’s output virtually always ends up on that screen.

This is NEVER guaranteed under UNIX®. It is quite common for a UNIX® user to pipe and redirect program input and output:

If you have written program2, your input does not come from the keyboard but from the output of program1. Similarly, your output does not go to the screen but becomes the input for program3 whose output, in turn, goes to file1.

But there is more! Even if you made sure that your input comes from, and your output goes to, the terminal, there is no guarantee the terminal is a PC: It may not have its video memory where you expect it, nor may its keyboard be producing PC-style scan codes. It may be a Macintosh®, or any other computer.

Now you may be shaking your head: My software is in PC assembly language, how can it run on a Macintosh®? But I did not say your software would be running on a Macintosh®, only that its terminal may be a Macintosh®.

Under UNIX®, the terminal does not have to be directly attached to the computer that runs your software, it can even be on another continent, or, for that matter, on another planet. It is perfectly possible that a Macintosh® user in Australia connects to a UNIX® system in North America (or anywhere else) via telnet. The software then runs on one computer, while the terminal is on a different computer: If you try to read the scan codes, you will get the wrong input!

Same holds true about any other hardware: A file you are reading may be on a disk you have no direct access to. A camera you are reading images from may be on a space shuttle, connected to you via satellites.

That is why under UNIX® you must never make any assumptions about where your data is coming from and going to. Always let the system handle the physical access to the hardware.