diff options
Diffstat (limited to 'en_US.ISO8859-1/books/developers-handbook/x86/chapter.sgml')
| -rw-r--r-- | en_US.ISO8859-1/books/developers-handbook/x86/chapter.sgml | 6488 |
1 files changed, 0 insertions, 6488 deletions
diff --git a/en_US.ISO8859-1/books/developers-handbook/x86/chapter.sgml b/en_US.ISO8859-1/books/developers-handbook/x86/chapter.sgml deleted file mode 100644 index c7a74aa21d..0000000000 --- a/en_US.ISO8859-1/books/developers-handbook/x86/chapter.sgml +++ /dev/null @@ -1,6488 +0,0 @@ -<!-- - The FreeBSD Documentation Project - - This file is automatically generated. Please do not make commits - to this file. Updates should be sent to the author : - - G. Adam Stanislav (adam@whizkidtech.net) - - This chapter is an exception to our general rule, and the author - retains the copyright. Among other things, this means that this - chapter should not be included in any printed version of the - Developer's Handbook without Adam's explicit permission. - - Eventually we will have to replace this chapter or convince the - author to assign us the copyright. For now, it is valuable - content so it should stay. - - $FreeBSD: doc/en_US.ISO8859-1/books/developers-handbook/x86/chapter.sgml,v 1.4 2001/06/23 06:56:59 dd Exp $ ---> - -<chapter id="x86"> - -<title>x86 Assembly Language Programming</title> -<para> -<emphasis> -This chapter was written by G. Adam Stanislav. -<ulink url="http://www.whizkidtech.net/">Whiz Kid Technomagic</ulink> -</emphasis></para> - - - -<sect1 id="x86-intro"> -<title>Synopsis</title> - -<para> -Assembly language programing under Unix is highly undocumented. It -is generally assumed that no one would ever want to use it because -various Unix systems run on different microprocessors, so everything -should be written in C for portability. -</para> - -<para> -In reality, C portability is quite a myth. Even C programs need -to be modified when ported from one Unix to another, regardless of -what processor each runs on. Typically, such a program is full -of conditional statements depending on the system it is -compiled for. -</para> - -<para> -Even if we believe that all of Unix software should be written in C, -or some other high-level language, we still need assembly language -programmers: Who else would write the section of C library -that accesses the kernel? -</para> - -<para> -In this chapter I will attempt to show you -how you can use assembly language writing -Unix programs, specifically under FreeBSD. -</para> - -<para> -This chapter does not explain the basics of assembly language. -There are enough resources about that (for a complete -online course in assembly language, see Randall Hyde's -<ulink url="http://webster.cs.ucr.edu/">Art -of Assembly Language</ulink>; or if you prefer -a printed book, take a look at Jeff Duntemann's -<ulink url="http://www.int80h.org/cgi-bin/isbn?isbn=0471375233">Assembly -Language Step-by-Step</ulink>). However, -once the chapter is finished, any assembly language programmer -will be able to write programs for FreeBSD -quickly and efficiently. -</para> - -<para> -Copyright © 2000-2001 G. Adam Stanislav. All rights reserved. -</para> - -</sect1> - -<sect1 id="x86-the-tools"> -<title>The Tools</title> - -<sect2 id="x86-the-assembler"> -<title>The Assembler</title> - -<para> -The most important tool for assembly language programming is the -assembler, the software that converts assembly language code -into machine language. -</para> - -<para> -Two very different assemblers are available for FreeBSD. One is -<citerefentry><refentrytitle>as</refentrytitle><manvolnum>1</manvolnum></citerefentry>, -which uses the traditional Unix assembly language syntax. It -comes with the system. -</para> - -<para> -The other is <application>/usr/ports/devel/nasm</application>. -It uses the Intel syntax. Its main advantage is that it -can assemble code for many operating systems. It needs -to be installed separately, but is completely free. -</para> - -<para> -This chapter uses <application>nasm</application> -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. -</para> - -</sect2> - -<sect2 id="x86-the-linker"> -<title>The Linker</title> - -<para> -The output of the assembler, like that of any -compiler, needs to be linked to form an executable file. -</para> - -<para> -The standard -<citerefentry><refentrytitle>ld</refentrytitle><manvolnum>1</manvolnum></citerefentry> -linker comes with FreeBSD. It works with the -code assembled with either assembler. -</para> - -</sect2> -</sect1> - -<sect1 id="x86-system-calls"> -<title>System Calls</title> - -<sect2 id="x86-default-calling-convention"> -<title>Default Calling Convention</title> - -<para> -By default, the FreeBSD kernel uses the C calling -convention. Further, although the kernel is accessed -using <function role="opcode">int 80h</function>, -it is assumed the program will call a function that -issues <function role="opcode">int 80h</function>, rather than -issuing <function role="opcode">int 80h</function> directly. -</para> - -<para> -This convention is very convenient, and quite superior to the -Microsoft convention used by <acronym>MS DOS</acronym>. -Why? Because the Unix convention allows any program written in -any language to access the kernel. -</para> - -<para> -An assembly language program can do that as well. -For example, we could open a file: -</para> - -<programlisting> -kernel: - int 80h ; Call kernel - ret - -open: - push dword mode - push dword flags - push dword path - mov eax, 5 - call kernel - add esp, byte 12 - ret -</programlisting> - -<para> -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. -</para> - -<para> -But assembly language programmers like to shave off cycles. The above example -requires a <function role="opcode">call/ret</function> combination. -We can eliminate it by -<function role="opcode">push</function>ing an extra dword: -</para> - -<programlisting> -open: - push dword mode - push dword flags - push dword path - mov eax, 5 - push eax ; Or any other dword - int 80h - add esp, byte 16 -</programlisting> - -<para> -The <constant>5</constant> that we have placed in -<varname role="register">EAX</varname> identifies -the kernel function, in this case <function role="syscall">open</function>. -</para> - -</sect2> -<sect2 id="x86-alternate-calling-convention"> -<title>Alternate Calling Convention</title> -<para> -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. -</para> - -<para> -Linux is a Unix-like system. However, its kernel uses the same -system-call convention of passing parameters in registers -<acronym>MS DOS</acronym> does. As with the Unix convention, -the function number is placed in <varname role="register">EAX</varname>. -The parameters, however, are not passed on the stack but in -<varname role="register">EBX, ECX, EDX, ESI, EDI, EBP</varname>: -</para> - -<programlisting> -open: - mov eax, 5 - mov ebx, path - mov ecx, flags - mov edx, mode - int 80h -</programlisting> - -<para> -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 <function role="opcode">push</function> the registers, then -<function role="opcode">pop</function> them later. This makes your code -bulkier and slower. Nevertheless, FreeBSD gives -you a choice. -</para> - -<para> -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: -</para> - -<screen>&prompt.user; <userinput>brandelf -f Linux <replaceable>filename</replaceable></userinput></screen> - -</sect2> - -<sect2 id="x86-use-geneva"> -<title>Which Convention Should You Use?</title> - -<para> -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. -</para> - -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-call-numbers"> -<title>Call Numbers</title> - -<para> -To tell the kernel which system service you are calling, -place its number in <varname role="register">EAX</varname>. Of course, you need -to know what the number is. -</para> - -<sect3 id="x86-the-syscalls-file"> -<title>The <filename>syscalls</filename> File</title> - -<para> -The numbers are listed in <filename>syscalls</filename>. -<command>locate syscalls</command> finds this file -in several different formats, all produced automatically -from <filename>syscalls.master</filename>. -</para> - -<para> -You can find the master file for the default Unix calling -convention in -<filename>/usr/src/sys/kern/syscalls.master</filename>. -If you need to use the other convention implemented -in the Linux emulation mode, read -<filename>/usr/src/sys/i386/linux/syscalls.master</filename>. -</para> - -<note> -<para> -Not only do FreeBSD and Linux use different calling -conventions, they sometimes use different numbers for -the same functions. -</para> -</note> - -<para> -<filename>syscalls.master</filename> describes how -the call is to be made: -</para> - -<programlisting> -0 STD NOHIDE { int nosys(void); } syscall nosys_args int -1 STD NOHIDE { void exit(int rval); } exit rexit_args void -2 STD POSIX { int fork(void); } -3 STD POSIX { ssize_t read(int fd, void *buf, size_t nbyte); } -4 STD POSIX { ssize_t write(int fd, const void *buf, size_t nbyte); } -5 STD POSIX { int open(char *path, int flags, int mode); } -6 STD POSIX { int close(int fd); } -etc... -</programlisting> -<para> -It is the leftmost column that tells us the number to place in -<varname role="register">EAX</varname>. -</para> - -<para> -The rightmost column tells us what parameters to -<function role="opcode">push</function>. They are <function role="opcode">push</function>ed -<emphasis>from right to left</emphasis>. -</para> - -<informalexample> -<para> -For example, to <function>open</function> a file, we need -to <function role="opcode">push</function> the <varname>mode</varname> first, -then <varname>flags</varname>, then the address at which -the <varname>path</varname> is stored. -</para> -</informalexample> - -</sect3> - -</sect2> - -</sect1> - -<sect1 id="x86-return-values"> -<title>Return Values</title> - -<para> -A system call would not be useful most of the time -if it did not return some kind of a value: The file -descriptor of an open file, the number of bytes read -to a buffer, the system time, etc. -</para> - -<para> -Additionally, the system needs to inform us if an error -occurs: A file does not exist, system resources are exhausted, -we passed an invalid parameter, etc. -</para> - -<sect2 id="x86-man-pages"> -<title>Man Pages</title> - -<para> -The traditional place to look for information about various -system calls under Unix systems are the man pages. -FreeBSD describes its system calls in section 2, sometimes -in section 3. -</para> - -<para> -For example, <citerefentry><refentrytitle>open</refentrytitle><manvolnum>2</manvolnum></citerefentry> says: -</para> - -<blockquote> -<para> -If successful, <function>open()</function> returns a non-negative -integer, termed a file descriptor. It returns <varname>-1</varname> on failure, -and sets <varname>errno</varname> to indicate the error. -</para> - -</blockquote> -<para> -The assembly language programmer new to Unix and FreeBSD will -immediately ask the puzzling question: Where is -<varname>errno</varname> and how do I get to it? -</para> - -<note> -<para> -The information presented in the man pages applies -to C programs. The assembly language programmer needs additional -information. -</para> -</note> - -</sect2> - -<sect2 id="x86-where-return-values"> -<title>Where Are the Return Values?</title> - -<para> -Unfortunately, it depends... For most system calls it is -in <varname role="register">EAX</varname>, 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 <varname role="register">EAX</varname>. -If it is not there, you -need further research. -</para> - -<note> -<para> -I am aware of one system call that returns the value in -<varname role="register">EDX</varname>: <function role="syscall">SYS_fork</function>. All others -I have worked with use <varname role="register">EAX</varname>. -But I have not worked with them all yet. -</para> -</note> - -<tip> -<para> -If you cannot find the answer here or anywhere else, -study <application>libc</application> source code and see how it -interfaces with the kernel. -</para> -</tip> - -</sect2> -<sect2 id="x86-where-errno"> -<title>Where Is <varname>errno</varname>?</title> - -<para> -Actually, nowhere... -</para> - -<para> -<varname>errno</varname> is part of the C language, not the -Unix kernel. When accessing kernel services directly, the -error code is returned in <varname role="register">EAX</varname>, -the same register the proper -return value generally ends up in. -</para> - -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-how-to-know-error"> -<title>Determining an Error Occurred</title> - -<para> -When using the standard FreeBSD calling convention, -the <varname role="register">carry flag</varname> is cleared upon success, -set upon failure. -</para> - -<para> -When using the Linux emulation mode, the signed -value in <varname role="register">EAX</varname> is non-negative upon success, -and contains the return value. In case of an error, the value -is negative, i.e., <varname>-errno</varname>. -</para> - -</sect2> - -</sect1> - -<sect1 id="x86-portable-code"> -<title>Creating Portable Code</title> - -<para> -Portability is generally not one of the strengths of assembly language. -Yet, writing assembly language programs for different platforms is -possible, especially with <application>nasm</application>. I have written -assembly language libraries that can be assembled for such different -operating systems as Windows and FreeBSD. -</para> - -<para> -It is all the more possible when you want your code to run -on two platforms which, while different, are based on -similar architectures. -</para> - -<para> -For example, FreeBSD is Unix, Linux is Unix-like. I only -mentioned three differences between them (from an assembly language -programmer's perspective): The calling convention, the -function numbers, and the way of returning values. -</para> - -<sect2 id="x86-deal-with-function-numbers"><title>Dealing with Function Numbers</title> - -<para> -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: -</para> - -<programlisting> -%ifdef LINUX -%define SYS_execve 11 -%else -%define SYS_execve 59 -%endif -</programlisting> -</sect2> -<sect2 id="x86-deal-with-geneva"><title>Dealing with Conventions</title> -<para> -Both, the calling convention, and the return value (the -<varname>errno</varname> problem) can be resolved with macros: -</para> - -<programlisting> -%ifdef LINUX - -%macro system 0 - call kernel -%endmacro - -align 4 -kernel: - push ebx - push ecx - push edx - push esi - push edi - push ebp - - mov ebx, [esp+32] - mov ecx, [esp+36] - mov edx, [esp+40] - mov esi, [esp+44] - mov ebp, [esp+48] - int 80h - - pop ebp - pop edi - pop esi - pop edx - pop ecx - pop ebx - - or eax, eax - js .errno - clc - ret - -.errno: - neg eax - stc - ret - -%else - -%macro system 0 - int 80h -%endmacro - -%endif -</programlisting> - -</sect2> - -<sect2 id="x86-deal-with-other-portability"><title>Dealing with Other Portability Issues</title> - -<para> -The above solutions can handle most cases of writing code -portable between FreeBSD and Linux. Nevertheless, with some -kernel services the differences are deeper. -</para> - -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-portable-library"><title>Using a Library</title> - -<para> -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. -</para> - -<para> -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 <varname role="register">EAX</varname> 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: -</para> - -<programlisting> -sys.open: -sys.close: -[etc...] - int 80h - ret -</programlisting> - -<para> -Your Linux library will require more different functions. -But even here you can group system calls using the same -number of parameters: -</para> - -<programlisting> -sys.exit: -sys.close: -[etc... one-parameter functions] - push ebx - mov ebx, [esp+12] - int 80h - pop ebx - jmp sys.return - -... - -sys.return: - or eax, eax - js sys.err - clc - ret - -sys.err: - neg eax - stc - ret -</programlisting> - -<para> -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. -</para> - -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-portable-include"> -<title>Using an Include File</title> - -<para> -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. -</para> - -<para> -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. -</para> - -<note> -<para> -This is the approach we will use throughout this chapter. -We will name our include file <filename>system.inc</filename>, and -add to it whenever we deal with a new system call. -</para> -</note> - -<para> -We can start our <filename>system.inc</filename> by declaring the -standard file descriptors: -</para> - -<programlisting> -%define stdin 0 -%define stdout 1 -%define stderr 2 -</programlisting> - -<para> -Next, we create a symbolic name for each system call: -</para> - -<programlisting> -%define SYS_nosys 0 -%define SYS_exit 1 -%define SYS_fork 2 -%define SYS_read 3 -%define SYS_write 4 -; [etc...] -</programlisting> - -<para> -We add a short, non-global procedure with a long name, -so we do not accidentally reuse the name in our code: -</para> - -<programlisting> -section .text -align 4 -access.the.bsd.kernel: - int 80h - ret -</programlisting> - -<para> -We create a macro which takes one argument, the syscall number: -</para> - -<programlisting> -%macro system 1 - mov eax, %1 - call access.the.bsd.kernel -%endmacro -</programlisting> - -<para> -Finally, we create macros for each syscall. These macros take -no arguments. -</para> - -<programlisting> -%macro sys.exit 0 - system SYS_exit -%endmacro - -%macro sys.fork 0 - system SYS_fork -%endmacro - -%macro sys.read 0 - system SYS_read -%endmacro - -%macro sys.write 0 - system SYS_write -%endmacro - -; [etc...] -</programlisting> - -<para> -Go ahead, enter it into your editor and save it as -<filename>system.inc</filename>. We will add more to it as we -discuss more syscalls. -</para> - -</sect2> - -</sect1> - -<sect1 id="x86-first-program"> -<title>Our First Program</title> - -<para> -We are now ready for our first program, the mandatory -<application>Hello, World!</application> -</para> - -<programlisting> - 1: %include 'system.inc' - 2: - 3: section .data - 4: hello db 'Hello, World!', 0Ah - 5: hbytes equ $-hello - 6: - 7: section .text - 8: global _start - 9: _start: -10: push dword hbytes -11: push dword hello -12: push dword stdout -13: sys.write -14: -15: push dword 0 -16: sys.exit -</programlisting> - -<para> -Here is what it does: Line 1 includes the defines, the macros, -and the code from <filename>system.inc</filename>. -</para> - -<para> -Lines 3-5 are the data: Line 3 starts the data section/segment. -Line 4 contains the string "Hello, World!" followed by a new -line (<constant>0Ah</constant>). Line 5 creates a constant that contains -the length of the string from line 4 in bytes. -</para> - -<para> -Lines 7-16 contain the code. Note that FreeBSD uses the <emphasis>elf</emphasis> -file format for its executables, which requires every -program to start at the point labeled <varname>_start</varname> (or, more -precisely, the linker expects that). This label has to be -global. -</para> - -<para> -Lines 10-13 ask the system to write <varname>hbytes</varname> bytes -of the <varname>hello</varname> string to <varname>stdout</varname>. -</para> - -<para> -Lines 15-16 ask the system to end the program with the return -value of <constant>0</constant>. The <function role="syscall">SYS_exit</function> syscall never -returns, so the code ends there. -</para> - -<note> -<para> -If you have come to Unix from <acronym>MS DOS</acronym> -assembly language background, you may be used to writing directly -to the video hardware. You will never have to worry about -this in FreeBSD, or any other flavor of Unix. As far as -you are concerned, you are writing to a file known as -<filename>stdout</filename>. This can be the video screen, or -a <application>telnet</application> terminal, or an actual file, -or even the input of another program. Which one it is, -is for the system to figure out. -</para> -</note> - -<sect2 id="x86-assemble-1"><title>Assembling the Code</title> - -<para> -Type the code (except the line numbers) in an editor, and save -it in a file named <filename>hello.asm</filename>. You need -<application>nasm</application> to assemble it. -</para> - -<sect3 id="x86-get-nasm"><title>Installing <application>nasm</application></title> - -<para> -If you do not have <application>nasm</application>, type: -</para> - -<screen>&prompt.user; <userinput>su</userinput> -Password:<userinput><replaceable>your root password</replaceable></userinput> -&prompt.root; <userinput>cd /usr/ports/devel/nasm</userinput> -&prompt.root; <userinput>make install</userinput> -&prompt.root; <userinput>exit</userinput> -&prompt.user;</screen> - -<para> -You may type <userinput>make install clean</userinput> instead of just -<userinput>make install</userinput> if you do not want to keep -<application>nasm</application> source code. -</para> - -<para> -Either way, FreeBSD will automatically download -<application>nasm</application> from the Internet, -compile it, and install it on your system. -</para> - -<note> -<para> -If your system is not FreeBSD, you need to get -<application>nasm</application> from its -<ulink url="http://www.web-sites.co.uk/nasm/">home -page</ulink>. You can still use it to assemble FreeBSD code. -</para> -</note> - -<para> -Now you can assemble, link, and run the code: -</para> - -<screen>&prompt.user; <userinput>nasm -f elf hello.asm</userinput> -&prompt.user; <userinput>ld -s -o hello hello.o</userinput> -&prompt.user; <userinput>./hello</userinput> -Hello, World! -&prompt.user;</screen> - -</sect3> - -</sect2> - -</sect1> - -<sect1 id="x86-unix-filters"> -<title>Writing Unix Filters</title> - -<para> -A common type of Unix application is a filter—a program -that reads data from the <filename>stdin</filename>, processes it -somehow, then writes the result to <filename>stdout</filename>. -</para> - -<para> -In this chapter, we shall develop a simple filter, and -learn how to read from <filename>stdin</filename> and write to -<filename>stdout</filename>. This filter will convert each byte -of its input into a hexadecimal number followed by a -blank space. -</para> - -<programlisting> -%include 'system.inc' - -section .data -hex db '0123456789ABCDEF' -buffer db 0, 0, ' ' - -section .text -global _start -_start: - ; read a byte from stdin - push dword 1 - push dword buffer - push dword stdin - sys.read - add esp, byte 12 - or eax, eax - je .done - - ; convert it to hex - movzx eax, byte [buffer] - mov edx, eax - shr dl, 4 - mov dl, [hex+edx] - mov [buffer], dl - and al, 0Fh - mov al, [hex+eax] - mov [buffer+1], al - - ; print it - push dword 3 - push dword buffer - push dword stdout - sys.write - add esp, byte 12 - jmp short _start - -.done: - push dword 0 - sys.exit -</programlisting> -<para> -In the data section we create an array called <varname>hex</varname>. -It contains the 16 hexadecimal digits in ascending order. -The array is followed by a buffer which we will use for -both input and output. The first two bytes of the buffer -are initially set to <constant>0</constant>. This is where we will write -the two hexadecimal digits (the first byte also is -where we will read the input). The third byte is a -space. -</para> - -<para> -The code section consists of four parts: Reading the byte, -converting it to a hexadecimal number, writing the result, -and eventually exiting the program. -</para> - -<para> -To read the byte, we ask the system to read one byte -from <filename>stdin</filename>, and store it in the first byte -of the <varname>buffer</varname>. The system returns the number -of bytes read in <varname role="register">EAX</varname>. This will be <constant>1</constant> -while data is coming, or <constant>0</constant>, when no more input -data is available. Therefore, we check the value of -<varname role="register">EAX</varname>. If it is <constant>0</constant>, -we jump to <varname>.done</varname>, otherwise we continue. -</para> - -<note> -<para> -For simplicity sake, we are ignoring the possibility -of an error condition at this time. -</para> -</note> - -<para> -The hexadecimal conversion reads the byte from the -<varname>buffer</varname> into <varname role="register">EAX</varname>, or actually just -<varname role="register">AL</varname>, while clearing the remaining bits of -<varname role="register">EAX</varname> to zeros. We also copy the byte to -<varname role="register">EDX</varname> because we need to convert the upper -four bits (nibble) separately from the lower -four bits. We store the result in the first two -bytes of the buffer. -</para> - -<para> -Next, we ask the system to write the three bytes -of the buffer, i.e., the two hexadecimal digits and -the blank space, to <filename>stdout</filename>. We then -jump back to the beginning of the program and -process the next byte. -</para> - -<para> -Once there is no more input left, we ask the system -to exit our program, returning a zero, which is -the traditional value meaning the program was -successful. -</para> - -<para> -Go ahead, and save the code in a file named <filename>hex.asm</filename>, -then type the following (the <userinput>^D</userinput> means press the -control key and type <userinput>D</userinput> while holding the -control key down): -</para> - -<screen>&prompt.user; <userinput>nasm -f elf hex.asm</userinput> -&prompt.user; <userinput>ld -s -o hex hex.o</userinput> -&prompt.user; <userinput>./hex</userinput> -<userinput>Hello, World!</userinput> -48 65 6C 6C 6F 2C 20 57 6F 72 6C 64 21 0A <userinput>Here I come!</userinput> -48 65 72 65 20 49 20 63 6F 6D 65 21 0A <userinput>^D</userinput> &prompt.user;</screen> - -<note> -<para> -If you are migrating to Unix from <acronym>MS DOS</acronym>, -you may be wondering why each line ends with <constant>0A</constant> -instead of <constant>0D 0A</constant>. -This is because Unix does not use the cr/lf convention, but -a "new line" convention, which is <constant>0A</constant> in hexadecimal. -</para> -</note> - -<para> -Can we improve this? Well, for one, it is a bit confusing because -once we have converted a line of text, our input no longer -starts at the begining of the line. We can modify it to print -a new line instead of a space after each <constant>0A</constant>: -</para> - -<programlisting> -%include 'system.inc' - -section .data -hex db '0123456789ABCDEF' -buffer db 0, 0, ' ' - -section .text -global _start -_start: - mov cl, ' ' - -.loop: - ; read a byte from stdin - push dword 1 - push dword buffer - push dword stdin - sys.read - add esp, byte 12 - or eax, eax - je .done - - ; convert it to hex - movzx eax, byte [buffer] - mov [buffer+2], cl - cmp al, 0Ah - jne .hex - mov [buffer+2], al - -.hex: - mov edx, eax - shr dl, 4 - mov dl, [hex+edx] - mov [buffer], dl - and al, 0Fh - mov al, [hex+eax] - mov [buffer+1], al - - ; print it - push dword 3 - push dword buffer - push dword stdout - sys.write - add esp, byte 12 - jmp short .loop - -.done: - push dword 0 - sys.exit -</programlisting> -<para> -We have stored the space in the <varname role="register">CL</varname> register. We can -do this safely because, unlike Microsoft Windows, Unix system -calls do not modify the value of any register they do not use -to return a value in. -</para> - -<para> -That means we only need to set <varname role="register">CL</varname> once. We have, therefore, -added a new label <varname>.loop</varname> and jump to it for the next byte -instead of jumping at <varname>_start</varname>. We have also added the -<varname>.hex</varname> label so we can either have a blank space or a -new line as the third byte of the <varname>buffer</varname>. -</para> - -<para> -Once you have changed <filename>hex.asm</filename> to reflect -these changes, type: -</para> - -<screen>&prompt.user; <userinput>nasm -f elf hex.asm</userinput> -&prompt.user; <userinput>ld -s -o hex hex.o</userinput> -&prompt.user; <userinput>./hex</userinput> -<userinput>Hello, World!</userinput> -48 65 6C 6C 6F 2C 20 57 6F 72 6C 64 21 0A -<userinput>Here I come!</userinput> -48 65 72 65 20 49 20 63 6F 6D 65 21 0A -<userinput>^D</userinput> &prompt.user;</screen> - -<para> -That looks better. But this code is quite inefficient! We -are making a system call for every single byte twice (once -to read it, another time to write the output). -</para> - -</sect1> - -<sect1 id="x86-buffered-io"> -<title>Buffered Input and Output</title> - -<para> -We can improve the efficiency of our code by buffering our -input and output. We create an input buffer and read a whole -sequence of bytes at one time. Then we fetch them one by one -from the buffer. -</para> - -<para> -We also create an output buffer. We store our output in it until -it is full. At that time we ask the kernel to write the contents -of the buffer to <filename>stdout</filename>. -</para> - -<para> -The program ends when there is no more input. But we still need -to ask the kernel to write the contents of our output buffer -to <filename>stdout</filename> one last time, otherwise some of our output -would make it to the output buffer, but never be sent out. -Do not forget that, or you will be wondering why some of your -output is missing. -</para> - -<programlisting> -%include 'system.inc' - -%define BUFSIZE 2048 - -section .data -hex db '0123456789ABCDEF' - -section .bss -ibuffer resb BUFSIZE -obuffer resb BUFSIZE - -section .text -global _start -_start: - sub eax, eax - sub ebx, ebx - sub ecx, ecx - mov edi, obuffer - -.loop: - ; read a byte from stdin - call getchar - - ; convert it to hex - mov dl, al - shr al, 4 - mov al, [hex+eax] - call putchar - - mov al, dl - and al, 0Fh - mov al, [hex+eax] - call putchar - - mov al, ' ' - cmp dl, 0Ah - jne .put - mov al, dl - -.put: - call putchar - jmp short .loop - -align 4 -getchar: - or ebx, ebx - jne .fetch - - call read - -.fetch: - lodsb - dec ebx - ret - -read: - push dword BUFSIZE - mov esi, ibuffer - push esi - push dword stdin - sys.read - add esp, byte 12 - mov ebx, eax - or eax, eax - je .done - sub eax, eax - ret - -align 4 -.done: - call write ; flush output buffer - push dword 0 - sys.exit - -align 4 -putchar: - stosb - inc ecx - cmp ecx, BUFSIZE - je write - ret - -align 4 -write: - sub edi, ecx ; start of buffer - push ecx - push edi - push dword stdout - sys.write - add esp, byte 12 - sub eax, eax - sub ecx, ecx ; buffer is empty now - ret -</programlisting> -<para> -We now have a third section in the source code, named -<varname>.bss</varname>. This section is not included in our -executable file, and, therefore, cannot be initialized. We use -<function role="opcode">resb</function> instead of <function role="opcode">db</function>. -It simply reserves the requested size of uninitialized memory -for our use. -</para> - -<para> -We take advantage of the fact that the system does not modify the -registers: We use registers for what, otherwise, would have to be -global variables stored in the <varname>.data</varname> section. This is -also why the Unix convention of passing parameters to system calls -on the stack is superior to the Microsoft convention of passing -them in the registers: We can keep the registers for our own use. -</para> - -<para> -We use <varname role="register">EDI</varname> and <varname role="register">ESI</varname> as pointers to the next byte -to be read from or written to. We use <varname role="register">EBX</varname> and -<varname role="register">ECX</varname> to keep count of the number of bytes in the -two buffers, so we know when to dump the output to, or read more -input from, the system. -</para> - -<para> -Let us see how it works now: -</para> - -<screen>&prompt.user; <userinput>nasm -f elf hex.asm</userinput> -&prompt.user; <userinput>ld -s -o hex hex.o</userinput> -&prompt.user; <userinput>./hex</userinput> -<userinput>Hello, World!</userinput> -<userinput>Here I come!</userinput> -48 65 6C 6C 6F 2C 20 57 6F 72 6C 64 21 0A -48 65 72 65 20 49 20 63 6F 6D 65 21 0A -<userinput>^D</userinput> &prompt.user;</screen> - -<para> -Not what you expected? The program did not print the output -until we pressed <userinput>^D</userinput>. That is easy to fix by -inserting three lines of code to write the output every time -we have converted a new line to <constant>0A</constant>. I have marked -the three lines with > (do not copy the > in your -<filename>hex.asm</filename>). -</para> - -<programlisting> -%include 'system.inc' - -%define BUFSIZE 2048 - -section .data -hex db '0123456789ABCDEF' - -section .bss -ibuffer resb BUFSIZE -obuffer resb BUFSIZE - -section .text -global _start -_start: - sub eax, eax - sub ebx, ebx - sub ecx, ecx - mov edi, obuffer - -.loop: - ; read a byte from stdin - call getchar - - ; convert it to hex - mov dl, al - shr al, 4 - mov al, [hex+eax] - call putchar - - mov al, dl - and al, 0Fh - mov al, [hex+eax] - call putchar - - mov al, ' ' - cmp dl, 0Ah - jne .put - mov al, dl - -.put: - call putchar -> cmp al, 0Ah -> jne .loop -> call write - jmp short .loop - -align 4 -getchar: - or ebx, ebx - jne .fetch - - call read - -.fetch: - lodsb - dec ebx - ret - -read: - push dword BUFSIZE - mov esi, ibuffer - push esi - push dword stdin - sys.read - add esp, byte 12 - mov ebx, eax - or eax, eax - je .done - sub eax, eax - ret - -align 4 -.done: - call write ; flush output buffer - push dword 0 - sys.exit - -align 4 -putchar: - stosb - inc ecx - cmp ecx, BUFSIZE - je write - ret - -align 4 -write: - sub edi, ecx ; start of buffer - push ecx - push edi - push dword stdout - sys.write - add esp, byte 12 - sub eax, eax - sub ecx, ecx ; buffer is empty now - ret -</programlisting> - -<para> -Now, let us see how it works: -</para> - -<screen>&prompt.user; <userinput>nasm -f elf hex.asm</userinput> -&prompt.user; <userinput>ld -s -o hex hex.o</userinput> -&prompt.user; <userinput>./hex</userinput> -<userinput>Hello, World!</userinput> -48 65 6C 6C 6F 2C 20 57 6F 72 6C 64 21 0A -<userinput>Here I come!</userinput> -48 65 72 65 20 49 20 63 6F 6D 65 21 0A -<userinput>^D</userinput> &prompt.user;</screen> - -<para> -Not bad for a 644-byte executable, is it! -</para> - -<note> -<para> -This approach to buffered input/output still -contains a hidden danger. I will discuss—and -fix—it later, when I talk about the -<link linkend="x86-buffered-dark-side">dark -side of buffering</link>.</para> -</note> - -<sect2 id="x86-ungetc"> -<title>How to Unread a Character</title> - -<warning><para> -This may be a somewhat advanced topic, mostly of interest to -programmers familiar with the theory of compilers. If you wish, -you may <link linkend="x86-command-line">skip to the next -section</link>, and perhaps read this later. -</para> -</warning> -<para> -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. -</para> - -<para> -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. -</para> - -<para> -This presents an interesting problem: How to return the next -character back to the input stream, so it can be read again -later? -</para> - -<para> -One possible solution is to store it in a character variable, -then set a flag. We can modify <function>getchar</function> 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. -</para> - -<para> -The C language has an <function>ungetc()</function> 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 -<function>getchar</function> 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. -</para> - -<para> -The key to returning a character back to the stream is in how -we are getting the characters to start with: -</para> - -<para> -First we check if the buffer is empty by testing the value -of <varname role="register">EBX</varname>. If it is zero, we call the -<function>read</function> procedure. -</para> - -<para> -If we do have a character available, we use <function role="opcode">lodsb</function>, then -decrease the value of <varname role="register">EBX</varname>. The <function role="opcode">lodsb</function> -instruction is effectively identical to: -</para> - -<programlisting> - mov al, [esi] - inc esi -</programlisting> - -<para> -The byte we have fetched remains in the buffer until the next -time <function>read</function> is called. We do not know when that happens, -but we do know it will not happen until the next call to -<function>getchar</function>. Hence, to "return" the last-read byte back -to the stream, all we have to do is decrease the value of -<varname role="register">ESI</varname> and increase the value of <varname role="register">EBX</varname>: -</para> - -<programlisting> -ungetc: - dec esi - inc ebx - ret -</programlisting> - -<para> -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 <function>ungetc</function> several times -in a row, it will work most of the time, but not all the time -(and will be tough to debug). Why? -</para> - -<para> -Because as long as <function>getchar</function> does not have to call -<function>read</function>, all of the pre-read bytes are still in the buffer, -and our <function>ungetc</function> works without a glitch. But the moment -<function>getchar</function> calls <function>read</function>, -the contents of the buffer change. -</para> - -<para> -We can always rely on <function>ungetc</function> working properly on the last -character we have read with <function>getchar</function>, but not on anything -we have read before that. -</para> - -<para> -If your program reads more than one byte ahead, you have at least -two choices: -</para> - -<para> -If possible, modify the program so it only reads one byte ahead. -This is the simplest solution. -</para> - -<para> -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 <varname>.bss</varname> section of your code, and create -a small "spare" buffer right before your input buffer, -something like this: -</para> - -<programlisting> -section .bss - resb 16 ; or whatever the value you came up with -ibuffer resb BUFSIZE -obuffer resb BUFSIZE -</programlisting> - -<para> -You also need to modify your <function>ungetc</function> to pass the value -of the byte to unget in <varname role="register">AL</varname>: -</para> - -<programlisting> -ungetc: - dec esi - inc ebx - mov [esi], al - ret -</programlisting> - -<para> -With this modification, you can call <function>ungetc</function> -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"). -</para> - -</sect2> - -</sect1> - -<sect1 id="x86-command-line"><title>Command Line Arguments</title> - -<para> -Our <application>hex</application> program will be more useful if it can -read the names of an input and output file from its command -line, i.e., if it can process the command line arguments. -But... Where are they? -</para> - -<para> -Before a Unix system starts a program, it <function role="opcode">push</function>es some -data on the stack, then jumps at the <varname>_start</varname> -label of the program. Yes, I said jumps, not calls. That means the -data can be accessed by reading <varname>[esp+offset]</varname>, -or by simply <function role="opcode">pop</function>ping it. -</para> - -<para> -The value at the top of the stack contains the number of -command line arguments. It is traditionally called -<varname>argc</varname>, for "argument count." -</para> - -<para> -Command line arguments follow next, all <varname>argc</varname> of them. -These are typically referred to as <varname>argv</varname>, for -"argument value(s)." That is, we get <varname>argv[0]</varname>, -<varname>argv[1]</varname>, <varname>...</varname>, -<varname>argv[argc-1]</varname>. These are not the actual -arguments, but pointers to arguments, i.e., memory addresses of -the actual arguments. The arguments themselves are -NUL-terminated character strings. -</para> - -<para> -The <varname>argv</varname> list is followed by a NULL pointer, -which is simply a <constant>0</constant>. There is more, but this is -enough for our purposes right now. -</para> - -<note> -<para> -If you have come from the <acronym>MS DOS</acronym> programming -environment, the main difference is that each argument is in -a separate string. The second difference is that there is no -practical limit on how many arguments there can be. -</para> -</note> - -<para> -Armed with this knowledge, we are almost ready for the next -version of <filename>hex.asm</filename>. First, however, we need to -add a few lines to <filename>system.inc</filename>: -</para> - -<para> -First, we need to add two new entries to our list of system -call numbers: -</para> - -<programlisting> -%define SYS_open 5 -%define SYS_close 6 -</programlisting> - -<para> -Then we add two new macros at the end of the file: -</para> - -<programlisting> -%macro sys.open 0 - system SYS_open -%endmacro - -%macro sys.close 0 - system SYS_close -%endmacro -</programlisting> - -<para> -Here, then, is our modified source code: -</para> - -<programlisting> -%include 'system.inc' - -%define BUFSIZE 2048 - -section .data -fd.in dd stdin -fd.out dd stdout -hex db '0123456789ABCDEF' - -section .bss -ibuffer resb BUFSIZE -obuffer resb BUFSIZE - -section .text -align 4 -err: - push dword 1 ; return failure - sys.exit - -align 4 -global _start -_start: - add esp, byte 8 ; discard argc and argv[0] - - pop ecx - jecxz .init ; no more arguments - - ; ECX contains the path to input file - push dword 0 ; O_RDONLY - push ecx - sys.open - jc err ; open failed - - add esp, byte 8 - mov [fd.in], eax - - pop ecx - jecxz .init ; no more arguments - - ; ECX contains the path to output file - push dword 420 ; file mode (644 octal) - push dword 0200h | 0400h | 01h - ; O_CREAT | O_TRUNC | O_WRONLY - push ecx - sys.open - jc err - - add esp, byte 12 - mov [fd.out], eax - -.init: - sub eax, eax - sub ebx, ebx - sub ecx, ecx - mov edi, obuffer - -.loop: - ; read a byte from input file or stdin - call getchar - - ; convert it to hex - mov dl, al - shr al, 4 - mov al, [hex+eax] - call putchar - - mov al, dl - and al, 0Fh - mov al, [hex+eax] - call putchar - - mov al, ' ' - cmp dl, 0Ah - jne .put - mov al, dl - -.put: - call putchar - cmp al, dl - jne .loop - call write - jmp short .loop - -align 4 -getchar: - or ebx, ebx - jne .fetch - - call read - -.fetch: - lodsb - dec ebx - ret - -read: - push dword BUFSIZE - mov esi, ibuffer - push esi - push dword [fd.in] - sys.read - add esp, byte 12 - mov ebx, eax - or eax, eax - je .done - sub eax, eax - ret - -align 4 -.done: - call write ; flush output buffer - - ; close files - push dword [fd.in] - sys.close - - push dword [fd.out] - sys.close - - ; return success - push dword 0 - sys.exit - -align 4 -putchar: - stosb - inc ecx - cmp ecx, BUFSIZE - je write - ret - -align 4 -write: - sub edi, ecx ; start of buffer - push ecx - push edi - push dword [fd.out] - sys.write - add esp, byte 12 - sub eax, eax - sub ecx, ecx ; buffer is empty now - ret -</programlisting> - -<para> -In our <varname>.data</varname> section we now have two new variables, -<varname>fd.in</varname> and <varname>fd.out</varname>. We store the input and -output file descriptors here. -</para> - -<para> -In the <varname>.text</varname> section we have replaced the references -to <varname>stdin</varname> and <varname>stdout</varname> with -<varname>[fd.in]</varname> and <varname>[fd.out]</varname>. -</para> - -<para> -The <varname>.text</varname> section now starts with a simple error -handler, which does nothing but exit the program with a return -value of <constant>1</constant>. -The error handler is before <varname>_start</varname> so we are -within a short distance from where the errors occur. -</para> - -<para> -Naturally, the program execution still begins at <varname>_start</varname>. -First, we remove <varname>argc</varname> and <varname>argv[0]</varname> from the -stack: They are of no interest to us (in this program, that is). -</para> - -<para> -We pop <varname>argv[1]</varname> to <varname role="register">ECX</varname>. This -register is particularly suited for pointers, as we can handle -NULL pointers with <function role="opcode">jecxz</function>. If <varname>argv[1]</varname> -is not NULL, we try to open the file named in the first -argument. Otherwise, we continue the program as before: Reading -from <varname>stdin</varname>, writing to <varname>stdout</varname>. -If we fail to open the input file (e.g., it does not exist), -we jump to the error handler and quit. -</para> - -<para> -If all went well, we now check for the second argument. If -it is there, we open the output file. Otherwise, we send -the output to <varname>stdout</varname>. If we fail to open the output -file (e.g., it exists and we do not have the write permission), -we, again, jump to the error handler. -</para> - -<para> -The rest of the code is the same as before, except we close -the input and output files before exiting, and, as mentioned, -we use <varname>[fd.in]</varname> and <varname>[fd.out]</varname>. -</para> - -<para> -Our executable is now a whopping 768 bytes long. -</para> - -<para> -Can we still improve it? Of course! Every program can be improved. -Here are a few ideas of what we could do: -</para> - -<itemizedlist> -<listitem> -<para> -Have our error handler print a message to -<varname>stderr</varname>. -</para> -</listitem> - -<listitem> -<para> -Add error handlers to the <function>read</function> -and <function>write</function> functions. -</para> -</listitem> - -<listitem> -<para> -Close <varname>stdin</varname> when we open an input file, -<varname>stdout</varname> when we open an output file. -</para> -</listitem> - -<listitem> -<para> -Add command line switches, such as <parameter>-i</parameter> -and <parameter>-o</parameter>, so we can list the input and -output files in any order, or perhaps read from -<varname>stdin</varname> and write to a file. -</para> -</listitem> - -<listitem> -<para> -Print a usage message if command line arguments are incorrect. -</para> -</listitem> - -</itemizedlist> -<para> -I shall leave these enhancements as an exercise to the reader: -You already know everything you need to know to implement them. -</para> - -</sect1> - -<sect1 id="x86-environment"> -<title>Unix Environment</title> - -<para> -An important Unix concept is the environment, which is defined by -<emphasis>environment variables</emphasis>. Some are set by the system, others -by you, yet others by the <application>shell</application>, or any program -that loads another program. -</para> - -<sect2 id="x86-find-environment"> -<title>How to Find Environment Variables</title> - -<para> -I said earlier that when a program starts executing, the stack -contains <varname>argc</varname> followed by the NULL-terminated -<varname>argv</varname> array, followed by something else. The -"something else" is the <emphasis>environment</emphasis>, or, -to be more precise, a NULL-terminated array of pointers to -<emphasis>environment variables</emphasis>. This is often referred -to as <varname>env</varname>. -</para> - -<para> -The structure of <varname>env</varname> is the same as that of -<varname>argv</varname>, a list of memory addresses followed by a -NULL (<constant>0</constant>). In this case, there is no -<varname>"envc"</varname>—we figure out where the array ends -by searching for the final NULL. -</para> - -<para> -The variables usually come in the <varname>name=value</varname> -format, but sometimes the <varname>=value</varname> part -may be missing. We need to account for that possibility. -</para> - -</sect2> - -<sect2 id="x86-webvar"> -<title>webvars</title> - -<para> -I could just show you some code that prints the environment -the same way the Unix <application>env</application> command does. But -I thought it would be more interesting to write a simple -assembly language CGI utility. -</para> - -<sect3 id="x86-cgi"> -<title>CGI: A Quick Overview</title> - -<para> -I have a -<ulink url="http://www.whizkidtech.net/cgi-bin/tutorial">detailed -<acronym>CGI</acronym> tutorial</ulink> on my web site, -but here is a very quick overview of <acronym>CGI</acronym>: -</para> - -<itemizedlist> -<listitem> -<para> -The web server communicates with the <acronym>CGI</acronym> -program by setting <emphasis>environment variables</emphasis>. -</para> -</listitem> - -<listitem> -<para> -The <acronym>CGI</acronym> program -sends its output to <filename>stdout</filename>. -The web server reads it from there. -</para> -</listitem> - -<listitem> -<para> -It must start with an <acronym>HTTP</acronym> -header followed by two blank lines. -</para> -</listitem> - -<listitem> -<para> -It then prints the <acronym>HTML</acronym> -code, or whatever other type of data it is producing. -</para> -</listitem> - -</itemizedlist> -<note> -<para> -While certain <emphasis>environment variables</emphasis> use -standard names, others vary, depending on the web server. That -makes <application>webvars</application> -quite a useful diagnostic tool. -</para> -</note> - -</sect3> - -<sect3 id="x86-webvars-the-code"> -<title>The Code</title> - -<para> -Our <application>webvars</application> program, then, must send out -the <acronym>HTTP</acronym> header followed by some -<acronym>HTML</acronym> mark-up. It then must read -the <emphasis>environment variables</emphasis> one by one -and send them out as part of the -<acronym>HTML</acronym> page. -</para> - -<para> -The code follows. I placed comments and explanations -right inside the code: -</para> - -<programlisting> -;;;;;;; webvars.asm ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -; -; Copyright (c) 2000 G. Adam Stanislav -; All rights reserved. -; -; Redistribution and use in source and binary forms, with or without -; modification, are permitted provided that the following conditions -; are met: -; 1. Redistributions of source code must retain the above copyright -; notice, this list of conditions and the following disclaimer. -; 2. Redistributions in binary form must reproduce the above copyright -; notice, this list of conditions and the following disclaimer in the -; documentation and/or other materials provided with the distribution. -; -; THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND -; ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE -; IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE -; ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE -; FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL -; DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS -; OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) -; HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT -; LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY -; OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF -; SUCH DAMAGE. -;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -; -; Version 1.0 -; -; Started: 8-Dec-2000 -; Updated: 8-Dec-2000 -; -;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -%include 'system.inc' - -section .data -http db 'Content-type: text/html', 0Ah, 0Ah - db '<?xml version="1.0" encoding="UTF-8"?>', 0Ah - db '<!DOCTYPE html PUBLIC "-//W3C/DTD XHTML Strict//EN" ' - db '"DTD/xhtml1-strict.dtd">', 0Ah - db '<html xmlns="http://www.w3.org/1999/xhtml" ' - db 'xml.lang="en" lang="en">', 0Ah - db '<head>', 0Ah - db '<title>Web Environment</title>', 0Ah - db '<meta name="author" content="G. Adam Stanislav" />', 0Ah - db '</head>', 0Ah, 0Ah - db '<body bgcolor="#ffffff" text="#000000" link="#0000ff" ' - db 'vlink="#840084" alink="#0000ff">', 0Ah - db '<div class="webvars">', 0Ah - db '<h1>Web Environment</h1>', 0Ah - db '<p>The following <b>environment variables</b> are defined ' - db 'on this web server:</p>', 0Ah, 0Ah - db '<table align="center" width="80" border="0" cellpadding="10" ' - db 'cellspacing="0" class="webvars">', 0Ah -httplen equ $-http -left db '<tr>', 0Ah - db '<td class="name"><tt>' -leftlen equ $-left -middle db '</tt></td>', 0Ah - db '<td class="value"><tt><b>' -midlen equ $-middle -undef db '<i>(undefined)</i>' -undeflen equ $-undef -right db '</b></tt></td>', 0Ah - db '</tr>', 0Ah -rightlen equ $-right -wrap db '</table>', 0Ah - db '</div>', 0Ah - db '</body>', 0Ah - db '</html>', 0Ah, 0Ah -wraplen equ $-wrap - -section .text -global _start -_start: - ; First, send out all the http and xhtml stuff that is - ; needed before we start showing the environment - push dword httplen - push dword http - push dword stdout - sys.write - - ; Now find how far on the stack the environment pointers - ; are. We have 12 bytes we have pushed before "argc" - mov eax, [esp+12] - - ; We need to remove the following from the stack: - ; - ; The 12 bytes we pushed for sys.write - ; The 4 bytes of argc - ; The EAX*4 bytes of argv - ; The 4 bytes of the NULL after argv - ; - ; Total: - ; 20 + eax * 4 - ; - ; Because stack grows down, we need to ADD that many bytes - ; to ESP. - lea esp, [esp+20+eax*4] - cld ; This should already be the case, but let's be sure. - - ; Loop through the environment, printing it out -.loop: - pop edi - or edi, edi ; Done yet? - je near .wrap - - ; Print the left part of HTML - push dword leftlen - push dword left - push dword stdout - sys.write - - ; It may be tempting to search for the '=' in the env string next. - ; But it is possible there is no '=', so we search for the - ; terminating NUL first. - mov esi, edi ; Save start of string - sub ecx, ecx - not ecx ; ECX = FFFFFFFF - sub eax, eax -repne scasb - not ecx ; ECX = string length + 1 - mov ebx, ecx ; Save it in EBX - - ; Now is the time to find '=' - mov edi, esi ; Start of string - mov al, '=' -repne scasb - not ecx - add ecx, ebx ; Length of name - - push ecx - push esi - push dword stdout - sys.write - - ; Print the middle part of HTML table code - push dword midlen - push dword middle - push dword stdout - sys.write - - ; Find the length of the value - not ecx - lea ebx, [ebx+ecx-1] - - ; Print "undefined" if 0 - or ebx, ebx - jne .value - - mov ebx, undeflen - mov edi, undef - -.value: - push ebx - push edi - push dword stdout - sys.write - - ; Print the right part of the table row - push dword rightlen - push dword right - push dword stdout - sys.write - - ; Get rid of the 60 bytes we have pushed - add esp, byte 60 - - ; Get the next variable - jmp .loop - -.wrap: - ; Print the rest of HTML - push dword wraplen - push dword wrap - push dword stdout - sys.write - - ; Return success - push dword 0 - sys.exit -</programlisting> - -<para> -This code produces a 1,396-byte executable. Most of it is data, -i.e., the <acronym>HTML</acronym> mark-up we need to send out. -</para> - -<para> -Assemble and link it as usual: -</para> - -<screen>&prompt.user; <userinput>nasm -f elf webvars.asm</userinput> -&prompt.user; <userinput>ld -s -o webvars webvars.o</userinput></screen> - -<para> -To use it, you need to upload <filename>webvars</filename> to your -web server. Depending on how your web server is set up, you -may have to store it in a special <filename>cgi-bin</filename> directory, -or perhaps rename it with a <filename>.cgi</filename> extension. -</para> - -<para> -Then you need to use your browser to view its output. -To see its output on my web server, please go to -<ulink url="http://www.int80h.org/webvars/"><filename>http://www.int80h.org/webvars/</filename></ulink>. -If curious about the additional environment variables -present in a password protected web directory, go to -<ulink url="http://www.int80h.org/private/"><filename>http://www.int80h.org/private/</filename></ulink>, -using the name <userinput>asm</userinput> and password -<userinput>programmer</userinput>. -</para> - -</sect3> - -</sect2> - -</sect1> - -<sect1 id="x86-files"> -<title>Working with Files</title> - -<para> -We have already done some basic file work: We know how -to open and close them, how to read and write them using -buffers. But Unix offers much more functionality when it -comes to files. We will examine some of it in this section, -and end up with a nice file conversion utility. -</para> - -<para> -Indeed, let us start at the end, that is, with the file -conversion utility. It always makes programming easier -when we know from the start what the end product is -supposed to do. -</para> - -<para> -One of the first programs I wrote for Unix was -<ulink url="ftp://ftp.int80h.org/unix/tuc/"><application>tuc</application></ulink>, -a text-to-Unix file converter. It converts a text -file from other operating systems to a Unix text file. -In other words, it changes from different kind of line endings -to the newline convention of Unix. It saves the output -in a different file. Optionally, it converts a Unix text -file to a <acronym>DOS</acronym> text file. -</para> - -<para> -I have used <application>tuc</application> extensively, but always -only to convert from some other <acronym>OS</acronym> -to Unix, never the other way. I have always wished -it would just overwrite the file instead of me having -to send the output to a different file. Most of the time, -I end up using it like this: -</para> - -<screen>&prompt.user; <userinput>tuc <replaceable>myfile tempfile</replaceable></userinput> -&prompt.user; <userinput>mv <replaceable>tempfile myfile</replaceable></userinput></screen> - -<para> -It would be nice to have a <application>ftuc</application>, -i.e., <emphasis>fast tuc</emphasis>, and use it like this: -</para> - -<screen>&prompt.user; <userinput>ftuc <replaceable>myfile</replaceable></userinput></screen> - -<para> -In this chapter, then, we will write -<application>ftuc</application> in assembly language -(the original <application>tuc</application> -is in C), and study various -file-oriented kernel services in the process. -</para> - -<para> -At first sight, such a file conversion is very -simple: All you have to do is strip the carriage -returns, right? -</para> - -<para> -If you answered yes, think again: That approach will -work most of the time (at least with <acronym>MS -DOS</acronym> text files), but will fail occasionally. -</para> - -<para> -The problem is that not all non-Unix text files end their -line with the carriage return / line feed sequence. Some -use carriage returns without line feeds. Others combine several -blank lines into a single carriage return followed by several -line feeds. And so on. -</para> - -<para> -A text file converter, then, must be able to handle -any possible line endings: -</para> - -<itemizedlist> -<listitem> -<para> -carriage return / line feed -</para> -</listitem> - -<listitem> -<para> -carriage return -</para> -</listitem> - -<listitem> -<para> -line feed / carriage return -</para> -</listitem> - -<listitem> -<para> -line feed -</para> -</listitem> - -</itemizedlist> -<para> -It should also handle files that use some kind of a -combination of the above (e.g., carriage return followed -by several line feeds). -</para> - -<sect2 id="x86-finite-state-machine"> -<title>Finite State Machine</title> - -<para> -The problem is easily solved by the use of a technique -called <emphasis>finite state machine</emphasis>, originally developed -by the designers of digital electronic circuits. A -<emphasis>finite state machine</emphasis> 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 <emphasis>finite state machine</emphasis>: 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. -</para> - -<para> -Because of that, all software is essentially a sequence of state -instructions for the microprocessor. Nevertheless, the concept -of <emphasis>finite state machine</emphasis> is useful in software design as well. -</para> - -<para> -Our text file converter can be designed as a <emphasis>finite state machine</emphasis> with three -possible states. We could call them states 0-2, -but it will make our life easier if we give them symbolic names: -</para> - -<itemizedlist> -<listitem> -<para> -<symbol>ordinary -</symbol></para> -</listitem> - -<listitem> -<para> -<symbol>cr -</symbol></para> -</listitem> - -<listitem> -<para> -<symbol>lf -</symbol></para> -</listitem> - -</itemizedlist> -<para> -Our program will start in the <symbol>ordinary</symbol> -state. During this state, the program action depends on -its input as follows: -</para> - -<itemizedlist> -<listitem> -<para> -If the input is anything other than a carriage return -or line feed, the input is simply passed on to the output. The -state remains unchanged. -</para> -</listitem> - -<listitem> -<para> -If the input is a carriage return, the state is changed -to <symbol>cr</symbol>. The input is then discarded, i.e., -no output is made. -</para> -</listitem> - -<listitem> -<para> -If the input is a line feed, the state is changed to -<symbol>lf</symbol>. The input is then discarded. -</para> -</listitem> - -</itemizedlist> -<para> -Whenever we are in the <symbol>cr</symbol> 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: -</para> - -<itemizedlist> -<listitem> -<para> -If the input is anything other than a carriage return -or line feed, output a line feed, then output the input, then -change the state to <symbol>ordinary</symbol>. -</para> -</listitem> - -<listitem> -<para> -If the input is a carriage return, we have received -two (or more) carriage returns in a row. We discard the -input, we output a line feed, and leave the state unchanged. -</para> -</listitem> - -<listitem> -<para> -If the input is a line feed, we output the line feed -and change the state to <symbol>ordinary</symbol>. Note that -this is not the same as the first case above – if we tried -to combine them, we would be outputting two line feeds -instead of one. -</para> -</listitem> - -</itemizedlist> -<para> -Finally, we are in the <symbol>lf</symbol> 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: -</para> - -<itemizedlist> -<listitem> -<para> -If the input is anything other than a carriage return or -line feed, we output a line feed, then output the input, then -change the state to <symbol>ordinary</symbol>. This is exactly -the same action as in the <symbol>cr</symbol> state upon -receiving the same kind of input. -</para> -</listitem> - -<listitem> -<para> -If the input is a carriage return, we discard the input, -we output a line feed, then change the state to <symbol>ordinary</symbol>. -</para> -</listitem> - -<listitem> -<para> -If the input is a line feed, we output the line feed, -and leave the state unchanged. -</para> -</listitem> - -</itemizedlist> -<sect3 id="x86-final-state"> -<title>The Final State</title> - -<para> -The above <emphasis>finite state machine</emphasis> works for the entire file, but leaves -the possibility that the final line end will be ignored. That will -happen whenever the file ends with a single carriage return or -a single line feed. I did not think of it when I wrote -<application>tuc</application>, just to discover that -occasionally it strips the last line ending. -</para> - -<para> -This problem is easily fixed by checking the state after the -entire file was processed. If the state is not -<symbol>ordinary</symbol>, we simply -need to output one last line feed. -</para> - -<note> -<para> -Now that we have expressed our algorithm as a <emphasis>finite state machine</emphasis>, -we could easily design a dedicated digital electronic -circuit (a "chip") to do the conversion for us. Of course, -doing so would be considerably more expensive than writing -an assembly language program. -</para> -</note> - -</sect3> - -<sect3 id="x86-tuc-counter"> -<title>The Output Counter</title> - -<para> -Because our file conversion program may be combining two -characters into one, we need to use an output counter. We -initialize it to <constant>0</constant>, and increase it -every time we send a character to the output. At the end of -the program, the counter will tell us what size we need -to set the file to. -</para> - -</sect3> - -</sect2> - -<sect2 id="x86-software-fsm"> -<title>Implementing FSM in Software</title> - -<para> -The hardest part of working with a <emphasis>finite state machine</emphasis> -is analyzing the problem and expressing it as a -<emphasis>finite state machine</emphasis>. That accomplished, -the software almost writes itself. -</para> - -<para> -In a high-level language, such as C, there are several main -approaches. One is to use a <function role="statement">switch</function> statement -which chooses what function should be run. For example, -</para> - -<programlisting> - switch (state) { - default: - case REGULAR: - regular(inputchar); - break; - case CR: - cr(inputchar); - break; - case LF: - lf(inputchar); - break; - } -</programlisting> - -<para> -Another approach is by using an array of function pointers, -something like this: -</para> - -<programlisting> - (output[state])(inputchar); -</programlisting> - -<para> -Yet another is to have <varname>state</varname> be a -function pointer, set to point at the appropriate function: -</para> - -<programlisting> - (*state)(inputchar); -</programlisting> -<para> -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 <varname role="register">EBX</varname>, and then just issue:</para> - -<programlisting> - call ebx -</programlisting> - -<para> -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 -<emphasis>possibly</emphasis> because with the caching modern -microprocessors do, either way may be equally fast. -</para> - -</sect2> - -<sect2 id="memory-mapped-files"> -<title>Memory Mapped Files</title> - -<para> -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. -</para> - -<para> -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 <function role="syscall">mmap</function> -system call to map it into the memory. One nice thing about -<function role="syscall">mmap</function> 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 <function role="opcode">mov</function>, -<function role="opcode">lods</function>, and <function role="opcode">stos</function>. -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. -</para> - -<para> -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. -</para> - -<para> -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 -<application>tuc</application> instead. -</para> - -<para> -If you examine your copy of <filename>syscalls.master</filename>, -you will find two separate syscalls named <function role="syscall">mmap</function>. -This is because of evolution of Unix: There was the traditional -<acronym>BSD</acronym> <function role="syscall">mmap</function>, -syscall 71. That one was superceded by the <acronym>POSIX</acronym> <function role="syscall">mmap</function>, -syscall 197. The FreeBSD system supports both because -older programs were written by using the original <acronym>BSD</acronym> -version. But new software uses the <acronym>POSIX</acronym> version, -which is what we will use. -</para> - -<para> -The <filename>syscalls.master</filename> file lists -the <acronym>POSIX</acronym> version like this: -</para> - -<programlisting> -197 STD BSD { caddr_t mmap(caddr_t addr, size_t len, int prot, \ - int flags, int fd, long pad, off_t pos); } -</programlisting> - -<para> -This differs slightly from what -<citerefentry><refentrytitle>mmap</refentrytitle><manvolnum>2</manvolnum></citerefentry> -says. That is because -<citerefentry><refentrytitle>mmap</refentrytitle><manvolnum>2</manvolnum></citerefentry> -describes the C version. -</para> - -<para> -The difference is in the <varname>long pad</varname> argument, which is not present in the C version. However, the FreeBSD syscalls add a 32-bit pad after <function role="opcode">push</function>ing a 64-bit argument. In this case, <varname>off_t</varname> is a 64-bit value.</para> - -<para> -When we are finished working with a memory-mapped file, -we unmap it with the <function role="syscall">munmap</function> syscall: -</para> - -<tip> -<para> -For an in-depth treatment of <function role="syscall">mmap</function>, see -W. Richard Stevens' -<ulink url="http://www.int80h.org/cgi-bin/isbn?isbn=0130810819">Unix -Network Programming, Volume 2, Chapter 12</ulink>. -</para> -</tip> - -</sect2> - -<sect2 id="x86-file-size"> -<title>Determining File Size</title> - -<para> -Because we need to tell <function role="syscall">mmap</function> 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. -</para> - -<para> -We can use the <function role="syscall">fstat</function> syscall to get all -the information about an open file that the system can give us. -That includes the file size. -</para> - -<para> -Again, <filename>syscalls.master</filename> lists two versions -of <function role="syscall">fstat</function>, a traditional one -(syscall 62), and a <acronym>POSIX</acronym> one -(syscall 189). Naturally, we will use the -<acronym>POSIX</acronym> version: -</para> - -<programlisting> -189 STD POSIX { int fstat(int fd, struct stat *sb); } -</programlisting> - -<para> -This is a very straightforward call: We pass to it the address -of a <structname>stat</structname> structure and the descriptor -of an open file. It will fill out the contents of the -<structname>stat</structname> structure. -</para> - -<para> -I do, however, have to say that I tried to declare the -<structname>stat</structname> structure in the -<varname>.bss</varname> section, and -<function role="syscall">fstat</function> 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. -</para> - -</sect2> - -<sect2 id="x86-ftruncate"> -<title>Changing the File Size</title> - -<para> -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. -</para> - -<para> -The <function role="syscall">ftruncate</function> system call allows us to do -just that. Despite its somewhat misleading name, the -<function role="syscall">ftruncate</function> system call can be used to both -truncate the file (make it smaller) and to grow it. -</para> - -<para> -And yes, we will find two versions of <function role="syscall">ftruncate</function> -in <filename>syscalls.master</filename>, an older one -(130), and a newer one (201). We will use -the newer one: -</para> - -<programlisting> -201 STD BSD { int ftruncate(int fd, int pad, off_t length); } -</programlisting> - -<para> -Please note that this one contains a <varname>int pad</varname> again. -</para> - -</sect2> - -<sect2 id="x86-ftuc"> -<title>ftuc</title> - -<para> -We now know everything we need to write <application>ftuc</application>. -We start by adding some new lines in <filename>system.inc</filename>. -First, we define some constants and structures, somewhere at -or near the beginning of the file: -</para> - -<programlisting> -;;;;;;; open flags -%define O_RDONLY 0 -%define O_WRONLY 1 -%define O_RDWR 2 - -;;;;;;; mmap flags -%define PROT_NONE 0 -%define PROT_READ 1 -%define PROT_WRITE 2 -%define PROT_EXEC 4 -;; -%define MAP_SHARED 0001h -%define MAP_PRIVATE 0002h - -;;;;;;; stat structure -struc stat -st_dev resd 1 ; = 0 -st_ino resd 1 ; = 4 -st_mode resw 1 ; = 8, size is 16 bits -st_nlink resw 1 ; = 10, ditto -st_uid resd 1 ; = 12 -st_gid resd 1 ; = 16 -st_rdev resd 1 ; = 20 -st_atime resd 1 ; = 24 -st_atimensec resd 1 ; = 28 -st_mtime resd 1 ; = 32 -st_mtimensec resd 1 ; = 36 -st_ctime resd 1 ; = 40 -st_ctimensec resd 1 ; = 44 -st_size resd 2 ; = 48, size is 64 bits -st_blocks resd 2 ; = 56, ditto -st_blksize resd 1 ; = 64 -st_flags resd 1 ; = 68 -st_gen resd 1 ; = 72 -st_lspare resd 1 ; = 76 -st_qspare resd 4 ; = 80 -endstruc -</programlisting> - -<para> -We define the new syscalls: -</para> - -<programlisting> -%define SYS_mmap 197 -%define SYS_munmap 73 -%define SYS_fstat 189 -%define SYS_ftruncate 201 -</programlisting> - -<para> -We add the macros for their use: -</para> - -<programlisting> -%macro sys.mmap 0 - system SYS_mmap -%endmacro - -%macro sys.munmap 0 - system SYS_munmap -%endmacro - -%macro sys.ftruncate 0 - system SYS_ftruncate -%endmacro - -%macro sys.fstat 0 - system SYS_fstat -%endmacro -</programlisting> - -<para> -And here is our code: -</para> - -<programlisting> -;;;;;;; Fast Text-to-Unix Conversion (ftuc.asm) ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -;; -;; Started: 21-Dec-2000 -;; Updated: 22-Dec-2000 -;; -;; Copyright 2000 G. Adam Stanislav. -;; All rights reserved. -;; -;;;;;;; v.1 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -%include 'system.inc' - -section .data - db 'Copyright 2000 G. Adam Stanislav.', 0Ah - db 'All rights reserved.', 0Ah -usg db 'Usage: ftuc filename', 0Ah -usglen equ $-usg -co db "ftuc: Can't open file.", 0Ah -colen equ $-co -fae db 'ftuc: File access error.', 0Ah -faelen equ $-fae -ftl db 'ftuc: File too long, use regular tuc instead.', 0Ah -ftllen equ $-ftl -mae db 'ftuc: Memory allocation error.', 0Ah -maelen equ $-mae - -section .text - -align 4 -memerr: - push dword maelen - push dword mae - jmp short error - -align 4 -toolong: - push dword ftllen - push dword ftl - jmp short error - -align 4 -facerr: - push dword faelen - push dword fae - jmp short error - -align 4 -cantopen: - push dword colen - push dword co - jmp short error - -align 4 -usage: - push dword usglen - push dword usg - -error: - push dword stderr - sys.write - - push dword 1 - sys.exit - -align 4 -global _start -_start: - pop eax ; argc - pop eax ; program name - pop ecx ; file to convert - jecxz usage - - pop eax - or eax, eax ; Too many arguments? - jne usage - - ; Open the file - push dword O_RDWR - push ecx - sys.open - jc cantopen - - mov ebp, eax ; Save fd - - sub esp, byte stat_size - mov ebx, esp - - ; Find file size - push ebx - push ebp ; fd - sys.fstat - jc facerr - - mov edx, [ebx + st_size + 4] - - ; File is too long if EDX != 0 ... - or edx, edx - jne near toolong - mov ecx, [ebx + st_size] - ; ... or if it is above 2 GB - or ecx, ecx - js near toolong - - ; Do nothing if the file is 0 bytes in size - jecxz .quit - - ; Map the entire file in memory - push edx - push edx ; starting at offset 0 - push edx ; pad - push ebp ; fd - push dword MAP_SHARED - push dword PROT_READ | PROT_WRITE - push ecx ; entire file size - push edx ; let system decide on the address - sys.mmap - jc near memerr - - mov edi, eax - mov esi, eax - push ecx ; for SYS_munmap - push edi - - ; Use EBX for state machine - mov ebx, ordinary - mov ah, 0Ah - cld - -.loop: - lodsb - call ebx - loop .loop - - cmp ebx, ordinary - je .filesize - - ; Output final lf - mov al, ah - stosb - inc edx - -.filesize: - ; truncate file to new size - push dword 0 ; high dword - push edx ; low dword - push eax ; pad - push ebp - sys.ftruncate - - ; close it (ebp still pushed) - sys.close - - add esp, byte 16 - sys.munmap - -.quit: - push dword 0 - sys.exit - -align 4 -ordinary: - cmp al, 0Dh - je .cr - - cmp al, ah - je .lf - - stosb - inc edx - ret - -align 4 -.cr: - mov ebx, cr - ret - -align 4 -.lf: - mov ebx, lf - ret - -align 4 -cr: - cmp al, 0Dh - je .cr - - cmp al, ah - je .lf - - xchg al, ah - stosb - inc edx - - xchg al, ah - ; fall through - -.lf: - stosb - inc edx - mov ebx, ordinary - ret - -align 4 -.cr: - mov al, ah - stosb - inc edx - ret - -align 4 -lf: - cmp al, ah - je .lf - - cmp al, 0Dh - je .cr - - xchg al, ah - stosb - inc edx - - xchg al, ah - stosb - inc edx - mov ebx, ordinary - ret - -align 4 -.cr: - mov ebx, ordinary - mov al, ah - ; fall through - -.lf: - stosb - inc edx - ret -</programlisting> - -<warning><para> -Do not use this program on files stored on a disk formated -by <acronym>MS DOS</acronym> or Windows. There seems to be a -subtle bug in the FreeBSD code when using <function role="syscall">mmap</function> -on these drives mounted under FreeBSD: If the file is over -a certain size, <function role="syscall">mmap</function> will just fill the memory -with zeros, and then copy them to the file overwriting -its contents. -</para> -</warning> -</sect2> - -</sect1> - -<sect1 id="x86-one-pointed-mind"> -<title>One-Pointed Mind</title> - -<para> -As a student of Zen, I like the idea of a one-pointed mind: -Do one thing at a time, and do it well. -</para> - -<para> -This, indeed, is very much how Unix works as well. While -a typical Windows application is attempting to do everything -imaginable (and is, therefore, riddled with bugs), a -typical Unix program does only one thing, and it does it -well. -</para> - -<para> -The typical Unix user then essentially assembles his own -applications by writing a shell script which combines the -various existing programs by piping the output of one -program to the input of another. -</para> - -<para> -When writing your own Unix software, it is generally a -good idea to see what parts of the problem you need to -solve can be handled by existing programs, and only -write your own programs for that part of the problem -that you do not have an existing solution for. -</para> - -<sect2 id="x86-csv"><title>CSV</title> - -<para> -I will illustrate this principle with a specific real-life -example I was faced with recently: -</para> - -<para> -I needed to extract the 11th field of each record from a -database I downloaded from a web site. The database was a -<acronym>CSV</acronym> file, i.e., a list of -<emphasis>comma-separated values</emphasis>. That is quite -a standard format for sharing data among people who may be -using different database software. -</para> - -<para> -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. -</para> - -<para> -I tried <application>awk</application>, using the comma as a separator. -But because several lines contained a quoted comma, -<application>awk</application> was extracting the wrong field -from those lines. -</para> - -<para> -Therefore, I needed to write my own software to extract the 11th -field from the <acronym>CSV</acronym> file. However, going with the Unix -spirit, I only needed to write a simple filter that would do the -following: -</para> - -<itemizedlist> -<listitem> -<para> -Remove the first line from the file; -</para> -</listitem> - -<listitem> -<para> -Change all unquoted commas to a different character; -</para> -</listitem> - -<listitem> -<para> -Remove all quotation marks. -</para> -</listitem> - -</itemizedlist> -<para> -Strictly speaking, I could use <application>sed</application> 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. -</para> - -<para> -At any rate, writing a program like this took me about -20 minutes. Writing a program that extracts the 11th field -from the <acronym>CSV</acronym> file would take a lot longer, -and I could not reuse it to extract some other field from some -other database. -</para> - -<para> -This time I decided to let it do a little more work than -a typical tutorial program would: -</para> - -<itemizedlist> -<listitem> -<para> -It parses its command line for options; -</para> -</listitem> - -<listitem> -<para> -It displays proper usage if it finds wrong arguments; -</para> -</listitem> - -<listitem> -<para> -It produces meaningful error messages. -</para> -</listitem> - -</itemizedlist> -<para> -Here is its usage message: -</para> - -<screen>Usage: csv [-t<delim>] [-c<comma>] [-p] [-o <outfile>] [-i <infile>]</screen> - -<para> -All parameters are optional, and can appear in any order. -</para> - -<para> -The <parameter>-t</parameter> parameter declares what to replace -the commas with. The <constant>tab</constant> is the default here. -For example, <parameter>-t;</parameter> will replace all unquoted -commas with semicolons. -</para> - -<para> -I did not need the <parameter>-c</parameter> 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, <parameter>-c@</parameter> will replace all at signs -(useful if you want to split a list of email addresses -to their user names and domains). -</para> - -<para> -The <parameter>-p</parameter> option preserves the first line, i.e., -it does not delete it. By default, we delete the first -line because in a <acronym>CSV</acronym> file it contains the field -names rather than data. -</para> - -<para> -The <parameter>-i</parameter> and <parameter>-o</parameter> -options let me specify the input and the output files. Defaults -are <filename>stdin</filename> and <filename>stdout</filename>, -so this is a regular Unix filter. -</para> - -<para> -I made sure that both <parameter>-i filename</parameter> and -<parameter>-ifilename</parameter> are accepted. I also made -sure that only one input and one output files may be -specified. -</para> - -<para> -To get the 11th field of each record, I can now do: -</para> - -<screen>&prompt.user; <userinput>csv '-t;' <replaceable>data.csv</replaceable> | awk '-F;' '{print $11}'</userinput></screen> - -<para> -The code stores the options (except for the file descriptors) -in <varname role="register">EDX</varname>: The comma in <varname role="register">DH</varname>, the new -separator in <varname role="register">DL</varname>, and the flag for -the <parameter>-p</parameter> option in the highest bit of -<varname role="register">EDX</varname>, so a check for its sign will give us a -quick decision what to do. -</para> - -<para> -Here is the code: -</para> - -<programlisting> -;;;;;;; csv.asm ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -; -; Convert a comma-separated file to a something-else separated file. -; -; Started: 31-May-2001 -; Updated: 1-Jun-2001 -; -; Copyright (c) 2001 G. Adam Stanislav -; All rights reserved. -; -;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; - -%include 'system.inc' - -%define BUFSIZE 2048 - -section .data -fd.in dd stdin -fd.out dd stdout -usg db 'Usage: csv [-t<delim>] [-c<comma>] [-p] [-o <outfile>] [-i <infile>]', 0Ah -usglen equ $-usg -iemsg db "csv: Can't open input file", 0Ah -iemlen equ $-iemsg -oemsg db "csv: Can't create output file", 0Ah -oemlen equ $-oemsg - -section .bss -ibuffer resb BUFSIZE -obuffer resb BUFSIZE - -section .text -align 4 -ierr: - push dword iemlen - push dword iemsg - push dword stderr - sys.write - push dword 1 ; return failure - sys.exit - -align 4 -oerr: - push dword oemlen - push dword oemsg - push dword stderr - sys.write - push dword 2 - sys.exit - -align 4 -usage: - push dword usglen - push dword usg - push dword stderr - sys.write - push dword 3 - sys.exit - -align 4 -global _start -_start: - add esp, byte 8 ; discard argc and argv[0] - mov edx, (',' << 8) | 9 - -.arg: - pop ecx - or ecx, ecx - je near .init ; no more arguments - - ; ECX contains the pointer to an argument - cmp byte [ecx], '-' - jne usage - - inc ecx - mov ax, [ecx] - -.o: - cmp al, 'o' - jne .i - - ; Make sure we are not asked for the output file twice - cmp dword [fd.out], stdout - jne usage - - ; Find the path to output file - it is either at [ECX+1], - ; i.e., -ofile -- - ; or in the next argument, - ; i.e., -o file - - inc ecx - or ah, ah - jne .openoutput - pop ecx - jecxz usage - -.openoutput: - push dword 420 ; file mode (644 octal) - push dword 0200h | 0400h | 01h - ; O_CREAT | O_TRUNC | O_WRONLY - push ecx - sys.open - jc near oerr - - add esp, byte 12 - mov [fd.out], eax - jmp short .arg - -.i: - cmp al, 'i' - jne .p - - ; Make sure we are not asked twice - cmp dword [fd.in], stdin - jne near usage - - ; Find the path to the input file - inc ecx - or ah, ah - jne .openinput - pop ecx - or ecx, ecx - je near usage - -.openinput: - push dword 0 ; O_RDONLY - push ecx - sys.open - jc near ierr ; open failed - - add esp, byte 8 - mov [fd.in], eax - jmp .arg - -.p: - cmp al, 'p' - jne .t - or ah, ah - jne near usage - or edx, 1 << 31 - jmp .arg - -.t: - cmp al, 't' ; redefine output delimiter - jne .c - or ah, ah - je near usage - mov dl, ah - jmp .arg - -.c: - cmp al, 'c' - jne near usage - or ah, ah - je near usage - mov dh, ah - jmp .arg - -align 4 -.init: - sub eax, eax - sub ebx, ebx - sub ecx, ecx - mov edi, obuffer - - ; See if we are to preserve the first line - or edx, edx - js .loop - -.firstline: - ; get rid of the first line - call getchar - cmp al, 0Ah - jne .firstline - -.loop: - ; read a byte from stdin - call getchar - - ; is it a comma (or whatever the user asked for)? - cmp al, dh - jne .quote - - ; Replace the comma with a tab (or whatever the user wants) - mov al, dl - -.put: - call putchar - jmp short .loop - -.quote: - cmp al, '"' - jne .put - - ; Print everything until you get another quote or EOL. If it - ; is a quote, skip it. If it is EOL, print it. -.qloop: - call getchar - cmp al, '"' - je .loop - - cmp al, 0Ah - je .put - - call putchar - jmp short .qloop - -align 4 -getchar: - or ebx, ebx - jne .fetch - - call read - -.fetch: - lodsb - dec ebx - ret - -read: - jecxz .read - call write - -.read: - push dword BUFSIZE - mov esi, ibuffer - push esi - push dword [fd.in] - sys.read - add esp, byte 12 - mov ebx, eax - or eax, eax - je .done - sub eax, eax - ret - -align 4 -.done: - call write ; flush output buffer - - ; close files - push dword [fd.in] - sys.close - - push dword [fd.out] - sys.close - - ; return success - push dword 0 - sys.exit - -align 4 -putchar: - stosb - inc ecx - cmp ecx, BUFSIZE - je write - ret - -align 4 -write: - jecxz .ret ; nothing to write - sub edi, ecx ; start of buffer - push ecx - push edi - push dword [fd.out] - sys.write - add esp, byte 12 - sub eax, eax - sub ecx, ecx ; buffer is empty now -.ret: - ret -</programlisting> - -<para> -Much of it is taken from <filename>hex.asm</filename> above. But there -is one important difference: I no longer call <function>write</function> -whenever I am outputing a line feed. Yet, the code can be -used interactively. -</para> - -<para> -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. -</para> - -<para> -The new solution I use now is to call <function>write</function> 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. -</para> - -<sect3 id="x86-buffered-dark-side"> -<title>The Dark Side of Buffering</title> -<para> -This change prevents a mysterious lockup -in a very specific case. I refer to it as the -<emphasis>dark side of buffering</emphasis>, mostly -because it presents a danger that is not -quite obvious. -</para> - -<para> -It is unlikely to happen with a program like the -<application>csv</application> above, so let us consider yet -another filter: In this case we expect our input -to be raw data representing color values, such as -the <emphasis>red</emphasis>, <emphasis>green</emphasis>, and -<emphasis>blue</emphasis> intensities of a pixel. Our -output will be the negative of our input. -</para> - -<para> -Such a filter would be very simple to write. -Most of it would look just like all the other -filters we have written so far, so I am only -going to show you its inner loop: -</para> - -<programlisting> -.loop: - call getchar - not al ; Create a negative - call putchar - jmp short .loop -</programlisting> -<para> -Because this filter works with raw data, -it is unlikely to be used interactively. -</para> - -<para> -But it could be called by image manipulation software. -And, unless it calls <function>write</function> before each call -to <function>read</function>, chances are it will lock up. -</para> - -<para> -Here is what might happen: -</para> - -<procedure><step><para> -The image editor will load our filter using the -C function <function>popen()</function>. -</para> -</step> -<step><para> -It will read the first row of pixels from -a bitmap or pixmap. -</para> -</step> -<step><para> -It will write the first row of pixels to -the <emphasis>pipe</emphasis> leading to -the <varname>fd.in</varname> of our filter. -</para> -</step> -<step><para> -Our filter will read each pixel -from its input, turn it to a negative, -and write it to its output buffer. -</para> -</step> -<step><para> -Our filter will call <function>getchar</function> -to fetch the next pixel. -</para> -</step> -<step><para> -<function>getchar</function> will find an empty -input buffer, so it will call -<function>read</function>. -</para> -</step> -<step><para> -<function>read</function> will call the -<function role="syscall">SYS_read</function> system call. -</para> -</step> -<step><para> -The <emphasis>kernel</emphasis> will suspend -our filter until the image editor -sends more data to the pipe. -</para> -</step> -<step><para> -The image editor will read from the -other pipe, connected to the -<varname>fd.out</varname> of our filter so it can set the first row of the -output image <emphasis>before</emphasis> -it sends us the second row of the input. -</para> -</step> -<step><para> -The <emphasis>kernel</emphasis> suspends -the image editor until it receives -some output from our filter, so it -can pass it on to the image editor. -</para> -</step> -</procedure> -<para> -At this point our filter waits for the image -editor to send it more data to process, while -the image editor is waiting for our filter -to send it the result of the processing -of the first row. But the result sits in -our output buffer. -</para> - -<para> -The filter and the image editor will continue -waiting for each other forever (or, at least, -until they are killed). Our software has just -entered a -<link linkend="secure-race-conditions">race condition</link>. -</para> - -<para> -This problem does not exist if our filter flushes -its output buffer <emphasis>before</emphasis> asking the -<emphasis>kernel</emphasis> for more input data. -</para> - -</sect3> - -</sect2> - -</sect1> - -<sect1 id="x86-fpu"> -<title>Using the <acronym>FPU</acronym></title> -<para> -Strangely enough, most of assembly language literature does not -even mention the existence of the <acronym>FPU</acronym>, -or <emphasis>floating point unit</emphasis>, let alone discuss -programming it. -</para> - -<para> -Yet, never does assembly language shine more than when -we create highly optimized <acronym>FPU</acronym> -code by doing things that can be done <emphasis>only</emphasis> in assembly language.</para> - -<sect2 id="x86-fpu-organization"><title>Organization of the <acronym>FPU</acronym></title> -<para> -The <acronym>FPU</acronym> consists of 8 80–bit floating–point registers. -These are organized in a stack fashion—you can -<function>push</function> a value on <acronym>TOS</acronym> -(<emphasis>top of stack</emphasis>) and you can -<function>pop</function> it. -</para> - -<para> -That said, the assembly language op codes are not <function role="opcode">push</function> -and <function role="opcode">pop</function> because those are already taken.</para> - -<para> -You can <function>push</function> a value on <acronym>TOS</acronym> -by using <function role="opcode">fld</function>, <function role="opcode">fild</function>, -and <function role="opcode">fbld</function>. Several other op codes -let you <function>push</function> many common -<emphasis>constants</emphasis>—such as <emphasis>pi</emphasis>—on -the <acronym>TOS</acronym>. -</para> - -<para> -Similarly, you can <function>pop</function> a value by -using <function role="opcode">fst</function>, <function role="opcode">fstp</function>, -<function role="opcode">fist</function>, <function role="opcode">fistp</function>, and -<function role="opcode">fbstp</function>. Actually, only the op -codes that end with a <emphasis>p</emphasis> will -literally <function>pop</function> the value, -the rest will <function>store</function> it -somewhere else without removing it from -the <acronym>TOS</acronym>. -</para> - -<para> -We can transfer the data between the -<acronym>TOS</acronym> and the computer memory either as -a 32–bit, 64–bit, or 80–bit <emphasis>real</emphasis>, -a 16–bit, 32–bit, or 64–bit <emphasis>integer</emphasis>, -or an 80–bit <emphasis>packed decimal</emphasis>. -</para> - -<para> -The 80–bit <emphasis>packed decimal</emphasis> is -a special case of <emphasis>binary coded -decimal</emphasis> which is very convenient when -converting between the <acronym>ASCII</acronym> -representation of data and the internal -data of the <acronym>FPU</acronym>. It allows us to use -18 significant digits. -</para> - -<para> -No matter how we represent data in the memory, -the <acronym>FPU</acronym> always stores it in the 80–bit -<emphasis>real</emphasis> format in its registers. -</para> - -<para> -Its internal precision is at least 19 decimal -digits, so even if we choose to display results -as <acronym>ASCII</acronym> in the full -18–digit precision, we are still showing -correct results. -</para> - -<para> -We can perform mathematical operations on the -<acronym>TOS</acronym>: We can calculate its -<emphasis>sine</emphasis>, we can <emphasis>scale</emphasis> it -(i.e., we can multiply or divide it by a power -of 2), we can calculate its base–2 -<emphasis>logarithm</emphasis>, and many other things. -</para> - -<para> -We can also <emphasis>multiply</emphasis> or -<emphasis>divide</emphasis> it by, <emphasis>add</emphasis> -it to, or <emphasis>subtract</emphasis> it from, -any of the <acronym>FPU</acronym> registers (including -itself). -</para> - -<para> -The official Intel op code for the -<acronym>TOS</acronym> is <varname role="register">st</varname>, and -for the <emphasis>registers</emphasis> -<varname role="register">st(0)</varname>–<varname role="register">st(7)</varname>. -<varname role="register">st</varname> and <varname role="register">st(0)</varname>, then, -refer to the same register. -</para> - -<para> -For whatever reasons, the original author of -<application>nasm</application> has decided to use -different op codes, namely -<varname role="register">st0</varname>–<varname role="register">st7</varname>. -In other words, there are no parentheses, -and the <acronym>TOS</acronym> is always -<varname role="register">st0</varname>, never just <function role="opcode">st</function>. -</para> - -<sect3 id="x86-fpu-packed-decimal"> -<title>The Packed Decimal Format</title> -<para> -The <emphasis>packed decimal</emphasis> format -uses 10 bytes (80 bits) of -memory to represent 18 digits. The -number represented there is always an -<emphasis>integer</emphasis>. -</para> - -<tip> -<para> -You can use it to get decimal places -by multiplying the <acronym>TOS</acronym> -by a power of 10 first. -</para> -</tip> - -<para> -The highest bit of the highest byte -(byte 9) is the <emphasis>sign bit</emphasis>: -If it is set, the number is <emphasis>negative</emphasis>, -otherwise, it is <emphasis>positive</emphasis>. -The rest of the bits of this byte are unused/ignored. -</para> - -<para> -The remaining 9 bytes store the 18 digits -of the number: 2 digits per byte.</para> - -<para> -The <emphasis>more significant digit</emphasis> is -stored in the high <emphasis>nibble</emphasis> -(4 bits), the <emphasis>less significant -digit</emphasis> in the low <emphasis>nibble</emphasis>. -</para> - -<para> -That said, you might think that <constant>-1234567</constant> -would be stored in the memory like this (using -hexadecimal notation): -</para> - -<programlisting> -80 00 00 00 00 00 01 23 45 67 -</programlisting> -<para> -Alas it is not! As with everything else of Intel make, -even the <emphasis>packed decimal</emphasis> is -<emphasis>little–endian</emphasis>.</para> - -<para> -That means our <constant>-1234567</constant> -is stored like this: -</para> - -<programlisting> -67 45 23 01 00 00 00 00 00 80 -</programlisting> -<para> -Remember that, or you will be pulling your hair out -in desperation! -</para> - -<note> -<para> -The book to read—if you can find it—is Richard Startz' -<ulink url="http://www.int80h.org/cgi-bin/isbn?isbn=013246604X">8087/80287/80387 -for the IBM PC & Compatibles</ulink>. -Though it does seem to take the fact about the -little–endian storage of the <emphasis>packed -decimal</emphasis> for granted. I kid you not about the -desperation of trying to figure out what was wrong -with the filter I show below <emphasis>before</emphasis> -it occurred to me I should try the -little–endian order even for this type of data. -</para> -</note> - -</sect3> - -</sect2> - -<sect2 id="x86-pinhole-photography"> -<title>Excursion to Pinhole Photography</title> -<para> -To write meaningful software, we must not only -understand our programming tools, but also the -field we are creating software for. -</para> - -<para> -Our next filter will help us whenever we want -to build a <emphasis>pinhole camera</emphasis>, -so, we need some background in <emphasis>pinhole -photography</emphasis> before we can continue. -</para> - -<sect3 id="x86-camera"> -<title>The Camera</title> -<para> -The easiest way to describe any camera ever built -is as some empty space enclosed in some -lightproof material, with a small hole in the -enclosure. -</para> - -<para> -The enclosure is usually sturdy (e.g., a box), -though sometimes it is flexible (the bellows). -It is quite dark inside the camera. However, the -hole lets light rays in through a single point -(though in some cases there may be several). -These light rays form an image, a representation -of whatever is outside the camera, in front of the -hole. -</para> - -<para> -If some light sensitive material (such as film) -is placed inside the camera, it can capture the -image.</para> - -<para> -The hole often contains a <emphasis>lens</emphasis>, or -a lens assembly, often called the <emphasis>objective</emphasis>. -</para> - -</sect3> - -<sect3 id="x86-the-pinhole"> -<title>The Pinhole</title> -<para> -But, strictly speaking, the lens is not necessary: -The original cameras did not use a lens but a -<emphasis>pinhole</emphasis>. Even today, <emphasis>pinholes</emphasis> -are used, both as a tool to study how cameras -work, and to achieve a special kind of image. -</para> - -<para> -The image produced by the <emphasis>pinhole</emphasis> -is all equally sharp. Or <emphasis>blurred</emphasis>. -There is an ideal size for a pinhole: If it is -either larger or smaller, the image loses its -sharpness.</para> - -</sect3> - -<sect3 id="x86-focal-length"> -<title>Focal Length</title> -<para> -This ideal pinhole diameter is a function -of the square root of <emphasis>focal -length</emphasis>, which is the distance of the -pinhole from the film. -</para> - -<programlisting> - D = PC * sqrt(FL) -</programlisting> -<para> -In here, <varname>D</varname> is the -ideal diameter of the pinhole, -<varname>FL</varname> is the focal length, -and <constant>PC</constant> is a pinhole -constant. According to Jay Bender, -its value is <constant>0.04</constant>, while -Kenneth Connors has determined it to -be <constant>0.037</constant>. Others have -proposed other values. Plus, this -value is for the daylight only: Other types -of light will require a different constant, -whose value can only be determined by -experimentation. -</para> - -</sect3> - -<sect3 id="x86-f-number"> -<title>The F–Number</title> -<para> -The f–number is a very useful measure of -how much light reaches the film. A light -meter can determine that, for example, -to expose a film of specific sensitivity -with f5.6 may require the exposure to last -1/1000 sec.</para> - -<para> -It does not matter whether it is a 35–mm -camera, or a 6x9cm camera, etc. -As long as we know the f–number, we can determine -the proper exposure. -</para> - -<para> -The f–number is easy to calculate: -</para> - -<programlisting> - F = FL / D -</programlisting> -<para> -In other words, the f–number equals the focal -length divided by the diameter of the pinhole. -It also means a higher f–number either implies -a smaller pinhole or a larger focal distance, -or both. That, in turn, implies, the higher -the f–number, the longer the exposure has to be. -</para> - -<para> -Furthermore, while pinhole diameter and focal -distance are one–dimensional measurements, -both, the film and the pinhole, are two–dimensional. -That means that -if you have measured the exposure at f–number -<varname>A</varname> as <varname>t</varname>, then the exposure -at f–number <varname>B</varname> is:</para> - -<programlisting> - t * (B / A)² -</programlisting> -</sect3> - -<sect3 id="x86-normalized-f-number"> -<title>Normalized F–Number</title> -<para> -While many modern cameras can change the diameter -of their pinhole, and thus their f–number, quite -smoothly and gradually, such was not always the case. -</para> - -<para> -To allow for different f–numbers, cameras typically -contained a metal plate with several holes of -different sizes drilled to them. -</para> - -<para> -Their sizes were chosen according to the above -formula in such a way that the resultant f–number -was one of standard f–numbers used on all cameras -everywhere. For example, a very old Kodak Duaflex IV -camera in my possession has three such holes for -f–numbers 8, 11, and 16. -</para> - -<para> -A more recently made camera may offer f–numbers of -2.8, 4, 5.6, 8, 11, -16, 22, and 32 (as well as others). -These numbers were not chosen arbitrarily: They all are -powers of the square root of 2, though they may -be rounded somewhat. -</para> - -</sect3> - -<sect3 id="x86-f-stop"> -<title>The F–Stop</title> -<para> -A typical camera is designed in such a way that setting -any of the normalized f–numbers changes the feel of the -dial. It will naturally <emphasis>stop</emphasis> in that -position. Because of that, these positions of the dial -are called f–stops.</para> - -<para> -Since the f–numbers at each stop are powers of the -square root of 2, moving the dial by 1 -stop will double the amount of light required for -proper exposure. Moving it by 2 stops will -quadruple the required exposure. Moving the dial by -3 stops will require the increase in exposure -8 times, etc. -</para> - -</sect3> - -</sect2> - -<sect2 id="x86-pinhole-software"> -<title>Designing the Pinhole Software</title> -<para> -We are now ready to decide what exactly we want our -pinhole software to do. -</para> - -<sect3 id="xpinhole-processing-input"> -<title>Processing Program Input</title> -<para> -Since its main purpose is to help us design a working -pinhole camera, we will use the <emphasis>focal -length</emphasis> as the input to the program. This is something -we can determine without software: Proper focal length -is determined by the size of the film and by the need -to shoot "regular" pictures, wide angle pictures, or -telephoto pictures. -</para> - -<para> -Most of the programs we have written so far worked with -individual characters, or bytes, as their input: The -<application>hex</application> program converted individual bytes -into a hexadecimal number, the <application>csv</application> -program either let a character through, or deleted it, -or changed it to a different character, etc. -</para> - -<para> -One program, <application>ftuc</application> used the state machine -to consider at most two input bytes at a time. -</para> - -<para> -But our <application>pinhole</application> program cannot just -work with individual characters, it has to deal with -larger syntactic units. -</para> - -<para> -For example, if we want the program to calculate the -pinhole diameter (and other values we will discuss -later) at the focal lengths of <constant>100 mm</constant>, -<constant>150 mm</constant>, and <constant>210 mm</constant>, we may want -to enter something like this:</para> - -<screen><userinput>100, 150, 210</userinput></screen> -<para> -Our program needs to consider more than a single byte of -input at a time. When it sees the first <constant>1</constant>, -it must understand it is seeing the first digit of a -decimal number. When it sees the <constant>0</constant> and -the other <constant>0</constant>, it must know it is seeing -more digits of the same number. -</para> - -<para> -When it encounters the first comma, it must know it is -no longer receiving the digits of the first number. -It must be able to convert the digits of the first number -into the value of <constant>100</constant>. And the digits of the -second number into the value of <constant>150</constant>. And, -of course, the digits of the third number into the -numeric value of <constant>210</constant>. -</para> - -<para> -We need to decide what delimiters to accept: Do the -input numbers have to be separated by a comma? If so, -how do we treat two numbers separated by something else? -</para> - -<para> -Personally, I like to keep it simple. Something either -is a number, so I process it. Or it is not a number, -so I discard it. I don't like the computer complaining -about me typing in an extra character when it is -<emphasis>obvious</emphasis> that it is an extra character. Duh! -</para> - -<para> -Plus, it allows me to break up the monotony of computing -and type in a query instead of just a number: -</para> - -<screen><userinput>What is the best pinhole diameter for the focal length of 150?</userinput></screen> -<para> -There is no reason for the computer to spit out -a number of complaints: -</para> - -<screen>Syntax error: What -Syntax error: is -Syntax error: the -Syntax error: best</screen> -<para> -Et cetera, et cetera, et cetera.</para> - -<para> -Secondly, I like the <constant>#</constant> character to denote -the start of a comment which extends to the end of the -line. This does not take too much effort to code, and -lets me treat input files for my software as executable -scripts. -</para> - -<para> -In our case, we also need to decide what units the -input should come in: We choose <emphasis>millimeters</emphasis> -because that is how most photographers measure -the focus length. -</para> - -<para> -Finally, we need to decide whether to allow the use -of the decimal point (in which case we must also -consider the fact that much of the world uses a -decimal <emphasis>comma</emphasis>).</para> - -<para> -In our case allowing for the decimal point/comma -would offer a false sense of precision: There is -little if any noticeable difference between the -focus lengths of <constant>50</constant> and <constant>51</constant>, -so allowing the user to input something like -<constant>50.5</constant> is not a good idea. This is -my opinion, mind you, but I am the one writing -this program. You can make other choices in yours, -of course. -</para> - -</sect3> - -<sect3 id="x86-pinhole-options"> -<title>Offering Options</title> -<para> -The most important thing we need to know when building -a pinhole camera is the diameter of the pinhole. Since -we want to shoot sharp images, we will use the above -formula to calculate the pinhole diameter from focal length. -As experts are offering several different values for the -<constant>PC</constant> constant, we will need to have the choice. -</para> - -<para> -It is traditional in Unix programming to have two main ways -of choosing program parameters, plus to have a default for -the time the user does not make a choice. -</para> - -<para> -Why have two ways of choosing?</para> - -<para> -One is to allow a (relatively) <emphasis>permanent</emphasis> -choice that applies automatically each time the -software is run without us having to tell it over and -over what we want it to do. -</para> - -<para> -The permanent choices may be stored in a configuration -file, typically found in the user's home directory. -The file usually has the same name as the application -but is started with a dot. Often <emphasis>"rc"</emphasis> -is added to the file name. So, ours could be -<filename>~/.pinhole</filename> or <filename>~/.pinholerc</filename>. -(The <filename>~/</filename> means current user's -home directory.) -</para> - -<para> -The configuration file is used mostly by programs -that have many configurable parameters. Those -that have only one (or a few) often use a different -method: They expect to find the parameter in an -<emphasis>environment variable</emphasis>. In our case, -we might look at an environment variable named -<varname>PINHOLE</varname>. -</para> - -<para> -Usually, a program uses one or the other of the -above methods. Otherwise, if a configuration -file said one thing, but an environment variable -another, the program might get confused (or just -too complicated). -</para> - -<para> -Because we only need to choose <emphasis>one</emphasis> -such parameter, we will go with the second method -and search the environment for a variable named -<varname>PINHOLE</varname>.</para> - -<para> -The other way allows us to make <emphasis>ad hoc</emphasis> -decisions: <emphasis>"Though I usually want -you to use 0.039, this time I want 0.03872."</emphasis> -In other words, it allows us to <emphasis>override</emphasis> -the permanent choice. -</para> - -<para> -This type of choice is usually done with command -line parameters. -</para> - -<para> -Finally, a program <emphasis>always</emphasis> needs a -<emphasis>default</emphasis>. The user may not make -any choices. Perhaps he does not know what -to choose. Perhaps he is "just browsing." -Preferably, the default will be the value -most users would choose anyway. That way -they do not need to choose. Or, rather, they -can choose the default without an additional -effort. -</para> - -<para> -Given this system, the program may find conflicting -options, and handle them this way: -</para> - -<procedure><step><para> -If it finds an <emphasis>ad hoc</emphasis> choice -(e.g., command line parameter), it should -accept that choice. It must ignore any permanent -choice and any default. -</para> -</step> -<step><para> -<emphasis>Otherwise</emphasis>, if it finds -a permanent option (e.g., an environment -variable), it should accept it, and ignore -the default.</para> -</step> -<step><para> -<emphasis>Otherwise</emphasis>, it should use -the default. -</para> -</step> -</procedure> -<para> -We also need to decide what <emphasis>format</emphasis> -our <constant>PC</constant> option should have. -</para> - -<para> -At first site, it seems obvious to use the -<varname>PINHOLE=0.04</varname> format for the -environment variable, and <parameter>-p0.04</parameter> -for the command line. -</para> - -<para> -Allowing that is actually a security risk. -The <constant>PC</constant> constant is a very small -number. Naturally, we will test our software -using various small values of <constant>PC</constant>. -But what will happen if someone runs the program -choosing a huge value? -</para> - -<para> -It may crash the program because we have not -designed it to handle huge numbers. -</para> - -<para> -Or, we may spend more time on the program so -it can handle huge numbers. We might do that -if we were writing commercial software for -computer illiterate audience. -</para> - -<para> -Or, we might say, <emphasis>"Tough! -The user should know better.""</emphasis> -</para> - -<para> -Or, we just may make it impossible for the user -to enter a huge number. This is the approach we -will take: We will use an <emphasis>implied 0.</emphasis> -prefix. -</para> - -<para> -In other words, if the user wants <constant>0.04</constant>, -we will expect him to type <parameter>-p04</parameter>, -or set <varname>PINHOLE=04</varname> in his environment. -So, if he says <parameter>-p9999999</parameter>, we will -interpret it as <constant>0.9999999</constant>—still -ridiculous but at least safer. -</para> - -<para> -Secondly, many users will just want to go with either -Bender's constant or Connors' constant. -To make it easier on them, we will interpret -<parameter>-b</parameter> as identical to <parameter>-p04</parameter>, -and <parameter>-c</parameter> as identical to <parameter>-p037</parameter>. -</para> - -</sect3> - -<sect3 id="x86-pinhole-output"> -<title>The Output</title> -<para> -We need to decide what we want our software to -send to the output, and in what format. -</para> - -<para> -Since our input allows for an unspecified number -of focal length entries, it makes sense to use -a traditional database–style output of showing -the result of the calculation for each -focal length on a separate line, while -separating all values on one line by a -<constant>tab</constant> character. -</para> - -<para> -Optionally, we should also allow the user -to specify the use of the <acronym>CSV</acronym> -format we have studied earlier. In this case, -we will print out a line of comma–separated -names describing each field of every line, -then show our results as before, but substituting -a <constant>comma</constant> for the <constant>tab</constant>.</para> - -<para> -We need a command line option for the <acronym>CSV</acronym> -format. We cannot use <parameter>-c</parameter> because -that already means <emphasis>use Connors' constant</emphasis>. -For some strange reason, many web sites refer to -<acronym>CSV</acronym> files as <emphasis>"Excel -spreadsheet"</emphasis> (though the <acronym>CSV</acronym> -format predates Excel). We will, therefore, use -the <parameter>-e</parameter> switch to inform our software -we want the output in the <acronym>CSV</acronym> format. -</para> - -<para> -We will start each line of the output with the -focal length. This may sound repetitious at first, -especially in the interactive mode: The user -types in the focal length, and we are repeating it. -</para> - -<para> -But the user can type several focal lengths on one -line. The input can also come in from a file or -from the output of another program. In that case -the user does not see the input at all. -</para> - -<para> -By the same token, the output can go to a file -which we will want to examine later, or it could -go to the printer, or become the input of another -program. -</para> - -<para> -So, it makes perfect sense to start each line with -the focal length as entered by the user. -</para> - -<para> -No, wait! Not as entered by the user. What if the user -types in something like this:</para> - -<screen><userinput>00000000150</userinput></screen> -<para> -Clearly, we need to strip those leading zeros.</para> - -<para> -So, we might consider reading the user input as is, -converting it to binary inside the <acronym>FPU</acronym>, -and printing it out from there. -</para> - -<para> -But...</para> - -<para> -What if the user types something like this: -</para> - -<screen><userinput>17459765723452353453534535353530530534563507309676764423</userinput></screen> -<para> -Ha! The packed decimal <acronym>FPU</acronym> format -lets us input 18–digit numbers. But the -user has entered more than 18 digits. How -do we handle that? -</para> - -<para> -Well, we <emphasis>could</emphasis> modify our code to read -the first 18 digits, enter it to the <acronym>FPU</acronym>, -then read more, multiply what we already have on the -<acronym>TOS</acronym> by 10 raised to the number -of additional digits, then <function>add</function> to it. -</para> - -<para> -Yes, we could do that. But in <emphasis>this</emphasis> -program it would be ridiculous (in a different one it may be just the thing to do): Even the circumference of the Earth expressed in -millimeters only takes 11 digits. Clearly, -we cannot build a camera that large (not yet, -anyway). -</para> - -<para> -So, if the user enters such a huge number, he is -either bored, or testing us, or trying to break -into the system, or playing games—doing -anything but designing a pinhole camera. -</para> - -<para> -What will we do?</para> - -<para> -We will slap him in the face, in a manner of speaking:</para> - -<screen>17459765723452353453534535353530530534563507309676764423 ??? ??? ??? ??? ???</screen> -<para> -To achieve that, we will simply ignore any leading zeros. -Once we find a non–zero digit, we will initialize a -counter to <constant>0</constant> and start taking three steps: -</para> - -<procedure> -<step><para> -Send the digit to the output. -</para> -</step> -<step><para> -Append the digit to a buffer we will use later to -produce the packed decimal we can send to the -<acronym>FPU</acronym>. -</para> -</step> -<step><para> -Increase the counter. -</para> -</step> -</procedure> -<para> -Now, while we are taking these three steps, -we also need to watch out for one of two -conditions:</para> - -<itemizedlist> -<listitem> -<para> -If the counter grows above 18, -we stop appending to the buffer. We -continue reading the digits and sending -them to the output. -</para> -</listitem> - -<listitem> -<para> -If, or rather <emphasis>when</emphasis>, -the next input character is not -a digit, we are done inputting -for now. -</para> - -<para> -Incidentally, we can simply -discard the non–digit, unless it -is a <constant>#</constant>, which we must -return to the input stream. It -starts a comment, so we must see it -after we are done producing output -and start looking for more input. -</para> -</listitem> - -</itemizedlist> -<para> -That still leaves one possibility -uncovered: If all the user enters -is a zero (or several zeros), we -will never find a non–zero to -display.</para> - -<para> -We can determine this has happened -whenever our counter stays at <constant>0</constant>. -In that case we need to send <constant>0</constant> -to the output, and perform another -"slap in the face": -</para> - -<screen>0 ??? ??? ??? ??? ???</screen> -<para> -Once we have displayed the focal -length and determined it is valid -(greater than <constant>0</constant> -but not exceeding 18 digits), -we can calculate the pinhole diameter. -</para> - -<para> -It is not by coincidence that <emphasis>pinhole</emphasis> -contains the word <emphasis>pin</emphasis>. Indeed, -many a pinhole literally is a <emphasis>pin -hole</emphasis>, a hole carefully punched with the -tip of a pin. -</para> - -<para> -That is because a typical pinhole is very -small. Our formula gets the result in -millimeters. We will multiply it by <constant>1000</constant>, -so we can output the result in <emphasis>microns</emphasis>. -</para> - -<para> -At this point we have yet another trap to face: -<emphasis>Too much precision.</emphasis> -</para> - -<para> -Yes, the <acronym>FPU</acronym> was designed -for high precision mathematics. But we -are not dealing with high precision -mathematics. We are dealing with physics -(optics, specifically). -</para> - -<para> -Suppose we want to convert a truck into -a pinhole camera (we would not be the -first ones to do that!). Suppose its box is -<constant>12</constant> -meters long, so we have the focal length -of <constant>12000</constant>. Well, using Bender's constant, it gives us square root of -<constant>12000</constant> multiplied by <constant>0.04</constant>, -which is <constant>4.381780460</constant> millimeters, -or <constant>4381.780460</constant> microns. -</para> - -<para> -Put either way, the result is absurdly precise. -Our truck is not <emphasis>exactly</emphasis> <constant>12000</constant> -millimeters long. We did not measure its length -with such a precision, so stating we need a pinhole -with the diameter of <constant>4.381780460</constant> -millimeters is, well, deceiving. <constant>4.4</constant> -millimeters would do just fine. -</para> - -<note> -<para> -I "only" used ten digits in the above example. -Imagine the absurdity of going for all 18! -</para> -</note> - -<para> -We need to limit the number of significant -digits of our result. One way of doing it -is by using an integer representing microns. -So, our truck would need a pinhole with the diameter -of <constant>4382</constant> microns. Looking at that number, we still decide that <constant>4400</constant> microns, -or <constant>4.4</constant> millimeters is close enough. -</para> - -<para> -Additionally, we can decide that no matter how -big a result we get, we only want to display four -siginificant digits (or any other number -of them, of course). Alas, the <acronym>FPU</acronym> -does not offer rounding to a specific number -of digits (after all, it does not view the -numbers as decimal but as binary). -</para> - -<para> -We, therefore, must devise an algorithm to reduce -the number of significant digits. -</para> - -<para> -Here is mine (I think it is awkward—if -you know a better one, <emphasis>please</emphasis>, let me know):</para> - -<procedure> -<step><para> -Initialize a counter to <constant>0</constant>. -</para> -</step> -<step><para> -While the number is greater than or equal to -<constant>10000</constant>, divide it by -<constant>10</constant> and increase the counter. -</para> -</step> -<step><para> -Output the result.</para> -</step> -<step><para> -While the counter is greater than <constant>0</constant>, -output <constant>0</constant> and decrease the counter. -</para> -</step> -</procedure> -<note> -<para> -The <constant>10000</constant> is only good if you want -<emphasis>four</emphasis> significant digits. For any other -number of significant digits, replace -<constant>10000</constant> with <constant>10</constant> -raised to the number of significant digits. -</para> -</note> - -<para> -We will, then, output the pinhole diameter -in microns, rounded off to four significant -digits. -</para> - -<para> -At this point, we know the <emphasis>focal -length</emphasis> and the <emphasis>pinhole -diameter</emphasis>. That means we have enough -information to also calculate the -<emphasis>f–number</emphasis>. -</para> - -<para> -We will display the f–number, rounded to -four significant digits. Chances are the -f–number will tell us very little. To make -it more meaningful, we can find the nearest -<emphasis>normalized f–number</emphasis>, i.e., -the nearest power of the square root -of 2. -</para> - -<para> -We do that by multiplying the actual f–number -by itself, which, of course, will give us -its <function>square</function>. We will then calculate -its base–2 logarithm, which is much -easier to do than calculating the -base–square–root–of–2 logarithm! -We will round the result to the nearest integer. -Next, we will raise 2 to the result. Actually, -the <acronym>FPU</acronym> gives us a good shortcut -to do that: We can use the <function role="opcode">fscale</function> -op code to "scale" 1, which is -analogous to <function role="opcode">shift</function>ing an -integer left. Finally, we calculate the square -root of it all, and we have the nearest -normalized f–number. -</para> - -<para> -If all that sounds overwhelming—or too much -work, perhaps—it may become much clearer -if you see the code. It takes 9 op -codes altogether:</para> - -<programlisting> - fmul st0, st0 - fld1 - fld st1 - fyl2x - frndint - fld1 - fscale - fsqrt - fstp st1 -</programlisting> -<para> -The first line, <function role="opcode">fmul st0, st0</function>, squares -the contents of the <acronym>TOS</acronym> -(top of the stack, same as <varname role="register">st</varname>, -called <varname role="register">st0</varname> by <application>nasm</application>). -The <function role="opcode">fld1</function> pushes <constant>1</constant> -on the <acronym>TOS</acronym>.</para> - -<para> -The next line, <function role="opcode">fld st1</function>, pushes -the square back to the <acronym>TOS</acronym>. -At this point the square is both in <varname role="register">st</varname> -and <varname role="register">st(2)</varname> (it will become -clear why we leave a second copy on the stack -in a moment). <varname role="register">st(1)</varname> contains -<constant>1</constant>. -</para> - -<para> -Next, <function role="opcode">fyl2x</function> calculates base–2 -logarithm of <varname role="register">st</varname> multiplied by -<varname role="register">st(1)</varname>. That is why we placed <constant>1</constant> on <varname role="register">st(1)</varname> before.</para> - -<para> -At this point, <varname role="register">st</varname> contains -the logarithm we have just calculated, -<varname role="register">st(1)</varname> contains the square -of the actual f–number we saved for later. -</para> - -<para> -<function role="opcode">frndint</function> rounds the <acronym>TOS</acronym> -to the nearest integer. <function role="opcode">fld1</function> pushes -a <constant>1</constant>. <function role="opcode">fscale</function> shifts the -<constant>1</constant> we have on the <acronym>TOS</acronym> -by the value in <varname role="register">st(1)</varname>, -effectively raising 2 to <varname role="register">st(1)</varname>. -</para> - -<para> -Finally, <function role="opcode">fsqrt</function> calculates -the square root of the result, i.e., -the nearest normalized f–number. -</para> - -<para> -We now have the nearest normalized -f–number on the <acronym>TOS</acronym>, -the base–2 logarithm rounded to the -nearest integer in <varname role="register">st(1)</varname>, -and the square of the actual f–number -in <varname role="register">st(2)</varname>. We are saving -the value in <varname role="register">st(2)</varname> for later. -</para> - -<para> -But we do not need the contents of -<varname role="register">st(1)</varname> anymore. The last -line, <function role="opcode">fstp st1</function>, places the -contents of <varname role="register">st</varname> to -<varname role="register">st(1)</varname>, and pops. As a -result, what was <varname role="register">st(1)</varname> -is now <varname role="register">st</varname>, what was <varname role="register">st(2)</varname> -is now <varname role="register">st(1)</varname>, etc. -The new <varname role="register">st</varname> contains the -normalized f–number. The new -<varname role="register">st(1)</varname> contains the square -of the actual f–number we have -stored there for posterity. -</para> - -<para> -At this point, we are ready to output -the normalized f–number. Because it is -normalized, we will not round it off to -four significant digits, but will -send it out in its full precision. -</para> - -<para> -The normalized f-number is useful as long -as it is reasonably small and can be found -on our light meter. Otherwise we need a -different method of determining proper -exposure. -</para> - -<para> -Earlier we have figured out the formula -of calculating proper exposure at an arbitrary -f–number from that measured at a different -f–number. -</para> - -<para> -Every light meter I have ever seen can determine -proper exposure at f5.6. We will, therefore, -calculate an <emphasis>"f5.6 multiplier,"</emphasis> -i.e., by how much we need to multiply the exposure measured -at f5.6 to determine the proper exposure -for our pinhole camera. -</para> - -<para> -From the above formula we know this factor can be -calculated by dividing our f–number (the -actual one, not the normalized one) by -<constant>5.6</constant>, and squaring the result. -</para> - -<para> -Mathematically, dividing the square of our -f–number by the square of <constant>5.6</constant> -will give us the same result. -</para> - -<para> -Computationally, we do not want to square -two numbers when we can only square one. -So, the first solution seems better at first. -</para> - -<para> -But...</para> - -<para> -<constant>5.6</constant> is a <emphasis>constant</emphasis>. -We do not have to have our <acronym>FPU</acronym> -waste precious cycles. We can just tell it -to divide the square of the f–number by -whatever <constant>5.6²</constant> equals to. -Or we can divide the f–number by <constant>5.6</constant>, -and then square the result. The two ways -now seem equal. -</para> - -<para> -But, they are not!</para> - -<para> -Having studied the principles of photography -above, we remember that the <constant>5.6</constant> -is actually square root of 2 raised to -the fifth power. An <emphasis>irrational</emphasis> -number. The square of this number is -<emphasis>exactly</emphasis> <constant>32</constant>. -</para> - -<para> -Not only is <constant>32</constant> an integer, -it is a power of 2. We do not need -to divide the square of the f–number by -<constant>32</constant>. We only need to use -<function role="opcode">fscale</function> to shift it right by -five positions. In the <acronym>FPU</acronym> -lingo it means we will <function role="opcode">fscale</function> it -with <varname role="register">st(1)</varname> equal to -<constant>-5</constant>. That is <emphasis>much -faster</emphasis> than a division. -</para> - -<para> -So, now it has become clear why we have -saved the square of the f–number on the -top of the <acronym>FPU</acronym> stack. -The calculation of the f5.6 multiplier -is the easiest calculation of this -entire program! We will output it rounded -to four significant digits. -</para> - -<para> -There is one more useful number we can calculate: -The number of stops our f–number is from f5.6. -This may help us if our f–number is just outside -the range of our light meter, but we have -a shutter which lets us set various speeds, -and this shutter uses stops. -</para> - -<para> -Say, our f–number is 5 stops from -f5.6, and the light meter says -we should use 1/1000 sec. -Then we can set our shutter speed to 1/1000 -first, then move the dial by 5 stops. -</para> - -<para> -This calculation is quite easy as well. All -we have to do is to calculate the base-2 -logarithm of the f5.6 multiplier -we had just calculated (though we need its -value from before we rounded it off). We then -output the result rounded to the nearest integer. -We do not need to worry about having more than -four significant digits in this one: The result -is most likely to have only one or two digits -anyway.</para> - -</sect3> - -</sect2> - -<sect2 id="x86-fpu-optimizations"> -<title>FPU Optimizations</title> -<para> -In assembly language we can optimize the <acronym>FPU</acronym> -code in ways impossible in high languages, -including C. -</para> - -<para> -Whenever a C function needs to calculate -a floating–point value, it loads all necessary -variables and constants into <acronym>FPU</acronym> -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. -</para> - -<para> -It "returns" the value by leaving -the result on the <acronym>TOS</acronym>. -However, before it returns, it cleans up. -Any variables and constants it used in its -calculation are now gone from the <acronym>FPU</acronym>. -</para> - -<para> -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. -</para> - -<para> -We <emphasis>knew</emphasis> 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. -</para> - -<para> -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. -</para> - -<para> -Of course, the C programmer may know it. -But the only recourse he has is to store the -value in a memory variable. -</para> - -<para> -That means, for one, the value will be changed -from the 80-bit precision used internally -by the <acronym>FPU</acronym> to a C <emphasis>double</emphasis> -(64 bits) or even <emphasis>single</emphasis> (32 -bits). -</para> - -<para> -That also means that the value must be moved -from the <acronym>TOS</acronym> into the memory, -and then back again. Alas, of all <acronym>FPU</acronym> -operations, the ones that access the computer -memory are the slowest. -</para> - -<para> -So, whenever programming the <acronym>FPU</acronym> -in assembly language, look for the ways of keeping -intermediate results on the <acronym>FPU</acronym> -stack. -</para> - -<para> -We can take that idea even further! In our -program we are using a <emphasis>constant</emphasis> -(the one we named <constant>PC</constant>). -</para> - -<para> -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. -</para> - -<para> -Early on in our program, we are calculating the -value of the above constant. We need to divide -our input by <constant>10</constant> for every digit in the -constant. -</para> - -<para> -It is much faster to multiply than to divide. -So, at the start of our program, we divide <constant>10</constant> -into <constant>1</constant> to obtain <constant>0.1</constant>, which we -then keep on the stack: Instead of dividing the -input by <constant>10</constant> for every digit, -we multiply it by <constant>0.1</constant>. -</para> - -<para> -By the way, we do not input <constant>0.1</constant> directly, -even though we could. We have a reason for that: -While <constant>0.1</constant> can be expressed with just one -decimal place, we do not know how many <emphasis>binary</emphasis> -places it takes. We, therefore, let the <acronym>FPU</acronym> -calculate its binary value to its own high precision. -</para> - -<para> -We are using other constants: We multiply the pinhole -diameter by <constant>1000</constant> to convert it from -millimeters to microns. We compare numbers to -<constant>10000</constant> when we are rounding them off to -four significant digits. So, we keep both, <constant>1000</constant> -and <constant>10000</constant>, on the stack. And, of course, -we reuse the <constant>0.1</constant> when rounding off numbers -to four digits. -</para> - -<para> -Last but not least, we keep <constant>-5</constant> on the stack. -We need it to scale the square of the f–number, -instead of dividing it by <constant>32</constant>. 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 <constant>-5</constant> is at <varname role="register">st(1)</varname>, -precisely where <function role="opcode">fscale</function> expects it to be. -</para> - -<para> -It is common to create certain constants from -scratch instead of loading them from the memory. -That is what we are doing with <constant>-5</constant>: -</para> - -<programlisting> - fld1 ; TOS = 1 - fadd st0, st0 ; TOS = 2 - fadd st0, st0 ; TOS = 4 - fld1 ; TOS = 1 - faddp st1, st0 ; TOS = 5 - fchs ; TOS = -5 -</programlisting> -<para> -We can generalize all these optimizations into one rule: -<emphasis>Keep repeat values on the stack!</emphasis> -</para> - -<tip> -<para> -<emphasis>PostScript</emphasis> is a stack–oriented -programming language. There are many more books -available about PostScript than about the -<acronym>FPU</acronym> assembly language: Mastering -PostScript will help you master the <acronym>FPU</acronym>. -</para> -</tip> - -</sect2> - -<sect2 id="x86-pinhole-the-code"> -<title><application>pinhole</application>—The Code</title> -<programlisting> -;;;;;;; pinhole.asm ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; -; -; Find various parameters of a pinhole camera construction and use -; -; Started: 9-Jun-2001 -; Updated: 10-Jun-2001 -; -; Copyright (c) 2001 G. Adam Stanislav -; All rights reserved. -; -;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; - -%include 'system.inc' - -%define BUFSIZE 2048 - -section .data -align 4 -ten dd 10 -thousand dd 1000 -tthou dd 10000 -fd.in dd stdin -fd.out dd stdout -envar db 'PINHOLE=' ; Exactly 8 bytes, or 2 dwords long -pinhole db '04,', ; Bender's constant (0.04) -connors db '037', 0Ah ; Connors' constant -usg db 'Usage: pinhole [-b] [-c] [-e] [-p <value>] [-o <outfile>] [-i <infile>]', 0Ah -usglen equ $-usg -iemsg db "pinhole: Can't open input file", 0Ah -iemlen equ $-iemsg -oemsg db "pinhole: Can't create output file", 0Ah -oemlen equ $-oemsg -pinmsg db "pinhole: The PINHOLE constant must not be 0", 0Ah -pinlen equ $-pinmsg -toobig db "pinhole: The PINHOLE constant may not exceed 18 decimal places", 0Ah -biglen equ $-toobig -huhmsg db 9, '???' -separ db 9, '???' -sep2 db 9, '???' -sep3 db 9, '???' -sep4 db 9, '???', 0Ah -huhlen equ $-huhmsg -header db 'focal length in millimeters,pinhole diameter in microns,' - db 'F-number,normalized F-number,F-5.6 multiplier,stops ' - db 'from F-5.6', 0Ah -headlen equ $-header - -section .bss -ibuffer resb BUFSIZE -obuffer resb BUFSIZE -dbuffer resb 20 ; decimal input buffer -bbuffer resb 10 ; BCD buffer - -section .text -align 4 -huh: - call write - push dword huhlen - push dword huhmsg - push dword [fd.out] - sys.write - add esp, byte 12 - ret - -align 4 -perr: - push dword pinlen - push dword pinmsg - push dword stderr - sys.write - push dword 4 ; return failure - sys.exit - -align 4 -consttoobig: - push dword biglen - push dword toobig - push dword stderr - sys.write - push dword 5 ; return failure - sys.exit - -align 4 -ierr: - push dword iemlen - push dword iemsg - push dword stderr - sys.write - push dword 1 ; return failure - sys.exit - -align 4 -oerr: - push dword oemlen - push dword oemsg - push dword stderr - sys.write - push dword 2 - sys.exit - -align 4 -usage: - push dword usglen - push dword usg - push dword stderr - sys.write - push dword 3 - sys.exit - -align 4 -global _start -_start: - add esp, byte 8 ; discard argc and argv[0] - sub esi, esi - -.arg: - pop ecx - or ecx, ecx - je near .getenv ; no more arguments - - ; ECX contains the pointer to an argument - cmp byte [ecx], '-' - jne usage - - inc ecx - mov ax, [ecx] - inc ecx - -.o: - cmp al, 'o' - jne .i - - ; Make sure we are not asked for the output file twice - cmp dword [fd.out], stdout - jne usage - - ; Find the path to output file - it is either at [ECX+1], - ; i.e., -ofile -- - ; or in the next argument, - ; i.e., -o file - - or ah, ah - jne .openoutput - pop ecx - jecxz usage - -.openoutput: - push dword 420 ; file mode (644 octal) - push dword 0200h | 0400h | 01h - ; O_CREAT | O_TRUNC | O_WRONLY - push ecx - sys.open - jc near oerr - - add esp, byte 12 - mov [fd.out], eax - jmp short .arg - -.i: - cmp al, 'i' - jne .p - - ; Make sure we are not asked twice - cmp dword [fd.in], stdin - jne near usage - - ; Find the path to the input file - or ah, ah - jne .openinput - pop ecx - or ecx, ecx - je near usage - -.openinput: - push dword 0 ; O_RDONLY - push ecx - sys.open - jc near ierr ; open failed - - add esp, byte 8 - mov [fd.in], eax - jmp .arg - -.p: - cmp al, 'p' - jne .c - or ah, ah - jne .pcheck - - pop ecx - or ecx, ecx - je near usage - - mov ah, [ecx] - -.pcheck: - cmp ah, '0' - jl near usage - cmp ah, '9' - ja near usage - mov esi, ecx - jmp .arg - -.c: - cmp al, 'c' - jne .b - or ah, ah - jne near usage - mov esi, connors - jmp .arg - -.b: - cmp al, 'b' - jne .e - or ah, ah - jne near usage - mov esi, pinhole - jmp .arg - -.e: - cmp al, 'e' - jne near usage - or ah, ah - jne near usage - mov al, ',' - mov [huhmsg], al - mov [separ], al - mov [sep2], al - mov [sep3], al - mov [sep4], al - jmp .arg - -align 4 -.getenv: - ; If ESI = 0, we did not have a -p argument, - ; and need to check the environment for "PINHOLE=" - or esi, esi - jne .init - - sub ecx, ecx - -.nextenv: - pop esi - or esi, esi - je .default ; no PINHOLE envar found - - ; check if this envar starts with 'PINHOLE=' - mov edi, envar - mov cl, 2 ; 'PINHOLE=' is 2 dwords long -rep cmpsd - jne .nextenv - - ; Check if it is followed by a digit - mov al, [esi] - cmp al, '0' - jl .default - cmp al, '9' - jbe .init - ; fall through - -align 4 -.default: - ; We got here because we had no -p argument, - ; and did not find the PINHOLE envar. - mov esi, pinhole - ; fall through - -align 4 -.init: - sub eax, eax - sub ebx, ebx - sub ecx, ecx - sub edx, edx - mov edi, dbuffer+1 - mov byte [dbuffer], '0' - - ; Convert the pinhole constant to real -.constloop: - lodsb - cmp al, '9' - ja .setconst - cmp al, '0' - je .processconst - jb .setconst - - inc dl - -.processconst: - inc cl - cmp cl, 18 - ja near consttoobig - stosb - jmp short .constloop - -align 4 -.setconst: - or dl, dl - je near perr - - finit - fild dword [tthou] - - fld1 - fild dword [ten] - fdivp st1, st0 - - fild dword [thousand] - mov edi, obuffer - - mov ebp, ecx - call bcdload - -.constdiv: - fmul st0, st2 - loop .constdiv - - fld1 - fadd st0, st0 - fadd st0, st0 - fld1 - faddp st1, st0 - fchs - - ; If we are creating a CSV file, - ; print header - cmp byte [separ], ',' - jne .bigloop - - push dword headlen - push dword header - push dword [fd.out] - sys.write - -.bigloop: - call getchar - jc near done - - ; Skip to the end of the line if you got '#' - cmp al, '#' - jne .num - call skiptoeol - jmp short .bigloop - -.num: - ; See if you got a number - cmp al, '0' - jl .bigloop - cmp al, '9' - ja .bigloop - - ; Yes, we have a number - sub ebp, ebp - sub edx, edx - -.number: - cmp al, '0' - je .number0 - mov dl, 1 - -.number0: - or dl, dl ; Skip leading 0's - je .nextnumber - push eax - call putchar - pop eax - inc ebp - cmp ebp, 19 - jae .nextnumber - mov [dbuffer+ebp], al - -.nextnumber: - call getchar - jc .work - cmp al, '#' - je .ungetc - cmp al, '0' - jl .work - cmp al, '9' - ja .work - jmp short .number - -.ungetc: - dec esi - inc ebx - -.work: - ; Now, do all the work - or dl, dl - je near .work0 - - cmp ebp, 19 - jae near .toobig - - call bcdload - - ; Calculate pinhole diameter - - fld st0 ; save it - fsqrt - fmul st0, st3 - fld st0 - fmul st5 - sub ebp, ebp - - ; Round off to 4 significant digits -.diameter: - fcom st0, st7 - fstsw ax - sahf - jb .printdiameter - fmul st0, st6 - inc ebp - jmp short .diameter - -.printdiameter: - call printnumber ; pinhole diameter - - ; Calculate F-number - - fdivp st1, st0 - fld st0 - - sub ebp, ebp - -.fnumber: - fcom st0, st6 - fstsw ax - sahf - jb .printfnumber - fmul st0, st5 - inc ebp - jmp short .fnumber - -.printfnumber: - call printnumber ; F number - - ; Calculate normalized F-number - fmul st0, st0 - fld1 - fld st1 - fyl2x - frndint - fld1 - fscale - fsqrt - fstp st1 - - sub ebp, ebp - call printnumber - - ; Calculate time multiplier from F-5.6 - - fscale - fld st0 - - ; Round off to 4 significant digits -.fmul: - fcom st0, st6 - fstsw ax - sahf - - jb .printfmul - inc ebp - fmul st0, st5 - jmp short .fmul - -.printfmul: - call printnumber ; F multiplier - - ; Calculate F-stops from 5.6 - - fld1 - fxch st1 - fyl2x - - sub ebp, ebp - call printnumber - - mov al, 0Ah - call putchar - jmp .bigloop - -.work0: - mov al, '0' - call putchar - -align 4 -.toobig: - call huh - jmp .bigloop - -align 4 -done: - call write ; flush output buffer - - ; close files - push dword [fd.in] - sys.close - - push dword [fd.out] - sys.close - - finit - - ; return success - push dword 0 - sys.exit - -align 4 -skiptoeol: - ; Keep reading until you come to cr, lf, or eof - call getchar - jc done - cmp al, 0Ah - jne .cr - ret - -.cr: - cmp al, 0Dh - jne skiptoeol - ret - -align 4 -getchar: - or ebx, ebx - jne .fetch - - call read - -.fetch: - lodsb - dec ebx - clc - ret - -read: - jecxz .read - call write - -.read: - push dword BUFSIZE - mov esi, ibuffer - push esi - push dword [fd.in] - sys.read - add esp, byte 12 - mov ebx, eax - or eax, eax - je .empty - sub eax, eax - ret - -align 4 -.empty: - add esp, byte 4 - stc - ret - -align 4 -putchar: - stosb - inc ecx - cmp ecx, BUFSIZE - je write - ret - -align 4 -write: - jecxz .ret ; nothing to write - sub edi, ecx ; start of buffer - push ecx - push edi - push dword [fd.out] - sys.write - add esp, byte 12 - sub eax, eax - sub ecx, ecx ; buffer is empty now -.ret: - ret - -align 4 -bcdload: - ; EBP contains the number of chars in dbuffer - push ecx - push esi - push edi - - lea ecx, [ebp+1] - lea esi, [dbuffer+ebp-1] - shr ecx, 1 - - std - - mov edi, bbuffer - sub eax, eax - mov [edi], eax - mov [edi+4], eax - mov [edi+2], ax - -.loop: - lodsw - sub ax, 3030h - shl al, 4 - or al, ah - mov [edi], al - inc edi - loop .loop - - fbld [bbuffer] - - cld - pop edi - pop esi - pop ecx - sub eax, eax - ret - -align 4 -printnumber: - push ebp - mov al, [separ] - call putchar - - ; Print the integer at the TOS - mov ebp, bbuffer+9 - fbstp [bbuffer] - - ; Check the sign - mov al, [ebp] - dec ebp - or al, al - jns .leading - - ; We got a negative number (should never happen) - mov al, '-' - call putchar - -.leading: - ; Skip leading zeros - mov al, [ebp] - dec ebp - or al, al - jne .first - cmp ebp, bbuffer - jae .leading - - ; We are here because the result was 0. - ; Print '0' and return - mov al, '0' - jmp putchar - -.first: - ; We have found the first non-zero. - ; But it is still packed - test al, 0F0h - jz .second - push eax - shr al, 4 - add al, '0' - call putchar - pop eax - and al, 0Fh - -.second: - add al, '0' - call putchar - -.next: - cmp ebp, bbuffer - jb .done - - mov al, [ebp] - push eax - shr al, 4 - add al, '0' - call putchar - pop eax - and al, 0Fh - add al, '0' - call putchar - - dec ebp - jmp short .next - -.done: - pop ebp - or ebp, ebp - je .ret - -.zeros: - mov al, '0' - call putchar - dec ebp - jne .zeros - -.ret: - ret -</programlisting> -<para> -The code follows the same format as all the other -filters we have seen before, with one subtle -exception: -</para> - -<blockquote> -<para> -We are no longer assuming that the end of input -implies the end of things to do, something we -took for granted in the <emphasis>character–oriented</emphasis> -filters. -</para> - -<para> -This filter does not process characters. It -processes a <emphasis>language</emphasis> -(albeit a very simple -one, consisting only of numbers). -</para> - -<para> -When we have no more input, it can mean one -of two things:</para> - -<itemizedlist><listitem> -<para> -We are done and can quit. This is the -same as before. -</para> -</listitem> - -<listitem> -<para> -The last character we have read was a digit. -We have stored it at the end of our -<acronym>ASCII</acronym>–to–float conversion -buffer. We now need to convert -the contents of that buffer into a -number and write the last line of our -output. -</para> -</listitem> - -</itemizedlist> -<para> -For that reason, we have modified our <function>getchar</function> -and our <function>read</function> routines to return with -the <varname role="register">carry flag</varname> <emphasis>clear</emphasis> whenever we are -fetching another character from the input, or the -<varname role="register">carry flag</varname> <emphasis>set</emphasis> whenever there is no more -input. -</para> - -<para> -Of course, we are still using assembly language magic -to do that! Take a good look at <function>getchar</function>. -It <emphasis>always</emphasis> returns with the -<varname role="register">carry flag</varname> <emphasis>clear</emphasis>. -</para> - -<para> -Yet, our main code relies on the <varname role="register">carry -flag</varname> to tell it when to quit—and it works. -</para> - -<para> -The magic is in <function>read</function>. Whenever it -receives more input from the system, it just -returns to <function>getchar</function>, which -fetches a character from the input buffer, -<emphasis>clears</emphasis> the <varname role="register">carry flag</varname> -and returns. -</para> - -<para> -But when <function>read</function> receives no more -input from the system, it does <emphasis>not</emphasis> -return to <function>getchar</function> at all. -Instead, the <function role="opcode">add esp, byte 4</function> -op code adds <constant>4</constant> to <varname role="register">ESP</varname>, -<emphasis>sets</emphasis> the <varname role="register">carry -flag</varname>, and returns. -</para> - -<para> -So, where does it return to? Whenever a -program uses the <function role="opcode">call</function> op code, -the microprocessor <function role="opcode">push</function>es the -return address, i.e., it stores it on -the top of the stack (not the <acronym>FPU</acronym> -stack, the system stack, which is in the memory). -When a program uses the <function role="opcode">ret</function> -op code, the microprocessor <function role="opcode">pop</function>s -the return value from the stack, and jumps -to the address that was stored there. -</para> - -<para> -But since we added <constant>4</constant> to -<varname role="register">ESP</varname> (which is the stack -pointer register), we have effectively -given the microprocessor a minor case -of <emphasis>amnesia</emphasis>: It no longer -remembers it was <function>getchar</function> -that <function role="opcode">call</function>ed <function>read</function>. -</para> - -<para> -And since <function>getchar</function> never -<function role="opcode">push</function>ed anything before -<function role="opcode">call</function>ing <function>read</function>, -the top of the stack now contains the -return address to whatever or whoever -<function role="opcode">call</function>ed <function>getchar</function>. -As far as that caller is concerned, -he <function role="opcode">call</function>ed <function>getchar</function>, -which <function role="opcode">ret</function>urned with the -<varname role="register">carry flag</varname> set! -</para> - -</blockquote> -<para> -Other than that, the <function>bcdload</function> -routine is caught up in the middle of a -Lilliputian conflict between the Big–Endians -and the Little–Endians. -</para> - -<para> -It is converting the text representation -of a number into that number: The text -is stored in the big–endian order, but -the <emphasis>packed decimal</emphasis> is little–endian. -</para> - -<para> -To solve the conflict, we use the <function>std</function> -op code early on. We cancel it with <function>cld</function> -later on: It is quite important we do not -<function>call</function> anything that may depend on -the default setting of the <emphasis>direction -flag</emphasis> while <function>std</function> is active. -</para> - -<para> -Everything else in this code should be quite -clear, providing you have read the entire chapter -that precedes it. -</para> - -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-pinhole-using"> -<title>Using <application>pinhole</application></title> -<para> -Because we have decided to make the program -<emphasis>ignore</emphasis> any input except for numbers -(and even those inside a comment), we can -actually perform <emphasis>textual queries</emphasis>. -We do not <emphasis>have to</emphasis>, but we <emphasis>can</emphasis>. -</para> - -<para> -In my humble opinion, forming a textual query, -instead of having to follow a very strict -syntax, makes software much more user friendly. -</para> - -<para> -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 <emphasis>fine–tune</emphasis> 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 <emphasis>ask</emphasis> a couple of questions. -</para> - -<para> -Our session might look like this:</para> - -<screen>&prompt.user; <userinput>pinhole - -Computer, - -What size pinhole do I need for the focal length of 150?</userinput> -150 490 306 362 2930 12 -<userinput>Hmmm... How about 160?</userinput> -160 506 316 362 3125 12 -<userinput>Let's make it 155, please.</userinput> -155 498 311 362 3027 12 -<userinput>Ah, let's try 157...</userinput> -157 501 313 362 3066 12 -<userinput>156?</userinput> -156 500 312 362 3047 12 -<userinput>That's it! Perfect! Thank you very much! -^D</userinput></screen> -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-pinhole-scripting"> -<title>Scripting</title> -<para> -Because we have chosen the <constant>#</constant> -character to denote the start of a comment, -we can treat our <application>pinhole</application> -software as a <emphasis>scripting language</emphasis>. -</para> - -<para> -You have probably seen <application>shell</application> -<emphasis>scripts</emphasis> that start with:</para> - -<programlisting> -#! /bin/sh -</programlisting> -<para> -...or...</para> - -<programlisting> -#!/bin/sh -</programlisting> <para> -...because the blank space after the <function>#!</function> -is optional. -</para> - -<para> -Whenever Unix is asked to run an executable -file which starts with the <function>#!</function>, -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. -</para> - -<para> -Suppose now that we have installed <application>pinhole</application> -in <application>/usr/local/bin/</application>, we can now -write a script to calculate various pinhole -diameters suitable for various focal lengths -commonly used with the 120 film.</para> - -<para> -The script might look something like this:</para> - -<programlisting> -#! /usr/local/bin/pinhole -b -i -# Find the best pinhole diameter -# for the 120 film - -### Standard -80 - -### Wide angle -30, 40, 50, 60, 70 - -### Telephoto -100, 120, 140 -</programlisting> -<para> -Because 120 is a medium size film, -we may name this file <application>medium</application>. -</para> - -<para> -We can set its permissions to execute, -and run it as if it were a program: -</para> - -<screen>&prompt.user; <userinput>chmod 755 medium</userinput> -&prompt.user; <userinput>./medium</userinput></screen> -<para> -Unix will interpret that last command as:</para> - -<screen>&prompt.user; <userinput>/usr/local/bin/pinhole -b -i ./medium</userinput></screen> -<para> -It will run that command and display: -</para> - -<screen>80 358 224 256 1562 11 -30 219 137 128 586 9 -40 253 158 181 781 10 -50 283 177 181 977 10 -60 310 194 181 1172 10 -70 335 209 181 1367 10 -100 400 250 256 1953 11 -120 438 274 256 2344 11 -140 473 296 256 2734 11</screen> -<para> - -Now, let us enter:</para> - -<screen>&prompt.user; <userinput>./medium -c</userinput></screen> -<para> -Unix will treat that as:</para> - -<screen>&prompt.user; <userinput>/usr/local/bin/pinhole -b -i ./medium -c</userinput></screen> -<para> -That gives it two conflicting options: -<parameter>-b</parameter> and <parameter>-c</parameter> -(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: -</para> - -<screen>80 331 242 256 1826 11 -30 203 148 128 685 9 -40 234 171 181 913 10 -50 262 191 181 1141 10 -60 287 209 181 1370 10 -70 310 226 256 1598 11 -100 370 270 256 2283 11 -120 405 296 256 2739 11 -140 438 320 362 3196 12</screen> -<para> -We decide we want to go with Bender's -constant after all. We want to save its -values as a comma–separated file: -</para> - -<screen>&prompt.user; <userinput>./medium -b -e > bender</userinput> -&prompt.user; <userinput>cat bender</userinput> -focal length in millimeters,pinhole diameter in microns,F-number,normalized F-number,F-5.6 multiplier,stops from F-5.6 -80,358,224,256,1562,11 -30,219,137,128,586,9 -40,253,158,181,781,10 -50,283,177,181,977,10 -60,310,194,181,1172,10 -70,335,209,181,1367,10 -100,400,250,256,1953,11 -120,438,274,256,2344,11 -140,473,296,256,2734,11 -&prompt.user;</screen> -</sect2> - -</sect1> - -<sect1 id="x86-caveats"> -<title>Caveats</title> - -<para> -Assembly language programmers who "grew up" under -<acronym>MS DOS</acronym> and Windows often tend to take shortcuts. -Reading the keyboard scan codes and writing directly to video -memory are two classical examples of practices which, under -<acronym>MS DOS</acronym> are not frowned upon but considered the -right thing to do. -</para> - -<para> -The reason? Both the <acronym>PC BIOS</acronym> and -<acronym>MS DOS</acronym> are notoriously -slow when performing these operations. -</para> - -<para> -You may be tempted to continue similar practices in the -Unix environment. For example, I have seen a web site which -explains how to access the keyboard scan codes on a popular Unix clone. -</para> - -<para> -That is generally a <emphasis>very bad idea</emphasis> -in Unix environment! Let me explain why. -</para> - -<sect2 id="x86-protected"> -<title>Unix Is Protected</title> - -<para> -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. -</para> - -</sect2> - -<sect2 id="x86-abstraction"> -<title>Unix Is an Abstraction</title> - -<para> -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: -</para> - -<para> -<emphasis>Unix is an abstraction! -</emphasis></para> - -<para> -There is a major difference in the philosophy of design -between <acronym>MS DOS</acronym> and Unix. -<acronym>MS DOS</acronym> 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. -</para> - -<para> -This is NEVER guaranteed under Unix. It is quite common -for a Unix user to pipe and redirect program input and output: -</para> - -<screen>&prompt.user; <userinput>program1 | program2 | program3 > file1</userinput></screen> - -<para> -If you have written <application>program2</application>, your input -does not come from the keyboard but from the output of -<application>program1</application>. Similarly, your output does not -go to the screen but becomes the input for -<application>program3</application> whose output, in turn, -goes to <filename>file1</filename>. -</para> - -<para> -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 -<acronym>PC</acronym>-style scan codes. It may be a Macintosh, -or any other computer. -</para> - -<para> -Now you may be shaking your head: My software is in -<acronym>PC</acronym> 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. -</para> - -<para> -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 <application>telnet</application>. 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! -</para> - -<para> -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. -</para> - -<para> -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. -</para> - -<note> -<para> -These are caveats, not absolute rules. Exceptions are possible. -For example, if a text editor has determined it is running -on a local machine, it may want to read the scan codes -directly for improved control. I am not mentioning these caveats -to tell you what to do or what not to do, just to make you aware -of certain pitfalls that await you if you have just arrived to Unix -form <acronym>MS DOS</acronym>. Of course, creative people often break -rules, and it is OK as long as they know they are breaking -them and why. -</para> -</note> - -</sect2> - -</sect1> - - -<sect1 id="x86-acknowledgements"> -<title>Acknowledgements</title> - -<para> -This tutorial would never have been possible without the -help of many experienced FreeBSD programmers from the -<ulink url="mailto:freebsd-hackers@FreeBSD.org">FreeBSD -hackers</ulink> mailing list, many of whom have patiently -answered my questions, and pointed me in the right direction -in my attempts to explore the inner workings of Unix -system programming in general and FreeBSD in particular. -</para> - -<para> -Thomas M. Sommers opened the door for me. His -<ulink url="http://home.ptd.net/~tms2/hello.html">How -do I write "Hello, world" in FreeBSD assembler?</ulink> -web page was my first encounter with an example of -assembly language programming under FreeBSD. -</para> - -<para> -Jake Burkholder has kept the door open by willingly -answering all of my questions and supplying me with -example assembly language source code. -</para> - -<para> -Copyright © 2000-2001 G. Adam Stanislav. All rights reserved. -</para> - -</sect1> - - -</chapter> |