2 \# Source code to NASM documentation
4 \M{category}{Programming}
5 \M{title}{NASM - The Netwide Assembler}
7 \M{author}{The NASM Development Team}
8 \M{license}{All rights reserved. This document is redistributable under the license given in the file "COPYING" distributed in the NASM archive.}
9 \M{summary}{This file documents NASM, the Netwide Assembler: an assembler targetting the Intel x86 series of processors, with portable source.}
12 \M{infotitle}{The Netwide Assembler for x86}
13 \M{epslogo}{nasmlogo.eps}
19 \IR{-On} \c{-On} option
37 \IR{!=} \c{!=} operator
38 \IR{$, here} \c{$}, Here token
39 \IR{$, prefix} \c{$}, prefix
42 \IR{%%} \c{%%} operator
43 \IR{%+1} \c{%+1} and \c{%-1} syntax
45 \IR{%0} \c{%0} parameter count
47 \IR{&&} \c{&&} operator
49 \IR{..@} \c{..@} symbol prefix
51 \IR{//} \c{//} operator
53 \IR{<<} \c{<<} operator
54 \IR{<=} \c{<=} operator
55 \IR{<>} \c{<>} operator
57 \IR{==} \c{==} operator
59 \IR{>=} \c{>=} operator
60 \IR{>>} \c{>>} operator
61 \IR{?} \c{?} MASM syntax
63 \IR{^^} \c{^^} operator
65 \IR{||} \c{||} operator
67 \IR{%$} \c{%$} and \c{%$$} prefixes
69 \IR{+ opaddition} \c{+} operator, binary
70 \IR{+ opunary} \c{+} operator, unary
71 \IR{+ modifier} \c{+} modifier
72 \IR{- opsubtraction} \c{-} operator, binary
73 \IR{- opunary} \c{-} operator, unary
74 \IR{! opunary} \c{!} operator, unary
75 \IR{alignment, in bin sections} alignment, in \c{bin} sections
76 \IR{alignment, in elf sections} alignment, in \c{elf} sections
77 \IR{alignment, in win32 sections} alignment, in \c{win32} sections
78 \IR{alignment, of elf common variables} alignment, of \c{elf} common
80 \IR{alignment, in obj sections} alignment, in \c{obj} sections
81 \IR{a.out, bsd version} \c{a.out}, BSD version
82 \IR{a.out, linux version} \c{a.out}, Linux version
83 \IR{autoconf} Autoconf
85 \IR{bitwise and} bitwise AND
86 \IR{bitwise or} bitwise OR
87 \IR{bitwise xor} bitwise XOR
88 \IR{block ifs} block IFs
89 \IR{borland pascal} Borland, Pascal
90 \IR{borland's win32 compilers} Borland, Win32 compilers
91 \IR{braces, after % sign} braces, after \c{%} sign
93 \IR{c calling convention} C calling convention
94 \IR{c symbol names} C symbol names
95 \IA{critical expressions}{critical expression}
96 \IA{command line}{command-line}
97 \IA{case sensitivity}{case sensitive}
98 \IA{case-sensitive}{case sensitive}
99 \IA{case-insensitive}{case sensitive}
100 \IA{character constants}{character constant}
101 \IR{common object file format} Common Object File Format
102 \IR{common variables, alignment in elf} common variables, alignment
104 \IR{common, elf extensions to} \c{COMMON}, \c{elf} extensions to
105 \IR{common, obj extensions to} \c{COMMON}, \c{obj} extensions to
106 \IR{declaring structure} declaring structures
107 \IR{default-wrt mechanism} default-\c{WRT} mechanism
110 \IR{dll symbols, exporting} DLL symbols, exporting
111 \IR{dll symbols, importing} DLL symbols, importing
113 \IR{dos archive} DOS archive
114 \IR{dos source archive} DOS source archive
115 \IA{effective address}{effective addresses}
116 \IA{effective-address}{effective addresses}
118 \IR{elf, 16-bit code and} ELF, 16-bit code and
119 \IR{elf shared libraries} ELF, shared libraries
120 \IR{executable and linkable format} Executable and Linkable Format
121 \IR{extern, obj extensions to} \c{EXTERN}, \c{obj} extensions to
122 \IR{extern, rdf extensions to} \c{EXTERN}, \c{rdf} extensions to
124 \IR{freelink} FreeLink
125 \IR{functions, c calling convention} functions, C calling convention
126 \IR{functions, pascal calling convention} functions, Pascal calling
128 \IR{global, aoutb extensions to} \c{GLOBAL}, \c{aoutb} extensions to
129 \IR{global, elf extensions to} \c{GLOBAL}, \c{elf} extensions to
130 \IR{global, rdf extensions to} \c{GLOBAL}, \c{rdf} extensions to
132 \IR{got relocations} \c{GOT} relocations
133 \IR{gotoff relocation} \c{GOTOFF} relocations
134 \IR{gotpc relocation} \c{GOTPC} relocations
135 \IR{intel number formats} Intel number formats
136 \IR{linux, elf} Linux, ELF
137 \IR{linux, a.out} Linux, \c{a.out}
138 \IR{linux, as86} Linux, \c{as86}
139 \IR{logical and} logical AND
140 \IR{logical or} logical OR
141 \IR{logical xor} logical XOR
143 \IA{memory reference}{memory references}
145 \IA{misc directory}{misc subdirectory}
146 \IR{misc subdirectory} \c{misc} subdirectory
147 \IR{microsoft omf} Microsoft OMF
148 \IR{mmx registers} MMX registers
149 \IA{modr/m}{modr/m byte}
150 \IR{modr/m byte} ModR/M byte
152 \IR{ms-dos device drivers} MS-DOS device drivers
153 \IR{multipush} \c{multipush} macro
155 \IR{nasm version} NASM version
159 \IR{operating system} operating system
161 \IR{pascal calling convention}Pascal calling convention
162 \IR{passes} passes, assembly
167 \IR{plt} \c{PLT} relocations
168 \IA{pre-defining macros}{pre-define}
169 \IA{preprocessor expressions}{preprocessor, expressions}
170 \IA{preprocessor loops}{preprocessor, loops}
171 \IA{preprocessor variables}{preprocessor, variables}
172 \IA{rdoff subdirectory}{rdoff}
173 \IR{rdoff} \c{rdoff} subdirectory
174 \IR{relocatable dynamic object file format} Relocatable Dynamic
176 \IR{relocations, pic-specific} relocations, PIC-specific
177 \IA{repeating}{repeating code}
178 \IR{section alignment, in elf} section alignment, in \c{elf}
179 \IR{section alignment, in bin} section alignment, in \c{bin}
180 \IR{section alignment, in obj} section alignment, in \c{obj}
181 \IR{section alignment, in win32} section alignment, in \c{win32}
182 \IR{section, elf extensions to} \c{SECTION}, \c{elf} extensions to
183 \IR{section, win32 extensions to} \c{SECTION}, \c{win32} extensions to
184 \IR{segment alignment, in bin} segment alignment, in \c{bin}
185 \IR{segment alignment, in obj} segment alignment, in \c{obj}
186 \IR{segment, obj extensions to} \c{SEGMENT}, \c{elf} extensions to
187 \IR{segment names, borland pascal} segment names, Borland Pascal
188 \IR{shift command} \c{shift} command
190 \IR{sib byte} SIB byte
191 \IR{solaris x86} Solaris x86
192 \IA{standard section names}{standardized section names}
193 \IR{symbols, exporting from dlls} symbols, exporting from DLLs
194 \IR{symbols, importing from dlls} symbols, importing from DLLs
195 \IR{test subdirectory} \c{test} subdirectory
197 \IR{underscore, in c symbols} underscore, in C symbols
199 \IA{sco unix}{unix, sco}
200 \IR{unix, sco} Unix, SCO
201 \IA{unix source archive}{unix, source archive}
202 \IR{unix, source archive} Unix, source archive
203 \IA{unix system v}{unix, system v}
204 \IR{unix, system v} Unix, System V
205 \IR{unixware} UnixWare
207 \IR{version number of nasm} version number of NASM
208 \IR{visual c++} Visual C++
209 \IR{www page} WWW page
213 \IR{windows 95} Windows 95
214 \IR{windows nt} Windows NT
215 \# \IC{program entry point}{entry point, program}
216 \# \IC{program entry point}{start point, program}
217 \# \IC{MS-DOS device drivers}{device drivers, MS-DOS}
218 \# \IC{16-bit mode, versus 32-bit mode}{32-bit mode, versus 16-bit mode}
219 \# \IC{c symbol names}{symbol names, in C}
222 \C{intro} Introduction
224 \H{whatsnasm} What Is NASM?
226 The Netwide Assembler, NASM, is an 80x86 and x86-64 assembler designed for
227 portability and modularity. It supports a range of object file
228 formats, including Linux and \c{*BSD} \c{a.out}, \c{ELF}, \c{COFF}, \c{Mach-O},
229 Microsoft 16-bit \c{OBJ}, \c{Win32} and \c{Win64}. It will also output plain
230 binary files. Its syntax is designed to be simple and easy to understand, similar
231 to Intel's but less complex. It supports from the upto and including \c{Pentium},
232 \c{P6}, \c{MMX}, \c{3DNow!}, \c{SSE}, \c{SSE2}, \c{SSE3} and \c{x64} opcodes. NASM has
233 a strong support for macro conventions.
236 \S{yaasm} Why Yet Another Assembler?
238 The Netwide Assembler grew out of an idea on \i\c{comp.lang.asm.x86}
239 (or possibly \i\c{alt.lang.asm} - I forget which), which was
240 essentially that there didn't seem to be a good \e{free} x86-series
241 assembler around, and that maybe someone ought to write one.
243 \b \i\c{a86} is good, but not free, and in particular you don't get any
244 32-bit capability until you pay. It's DOS only, too.
246 \b \i\c{gas} is free, and ports over to DOS and Unix, but it's not
247 very good, since it's designed to be a back end to \i\c{gcc}, which
248 always feeds it correct code. So its error checking is minimal. Also,
249 its syntax is horrible, from the point of view of anyone trying to
250 actually \e{write} anything in it. Plus you can't write 16-bit code in
253 \b \i\c{as86} is specific to Minix and Linux, and (my version at least)
254 doesn't seem to have much (or any) documentation.
256 \b \i\c{MASM} isn't very good, and it's (was) expensive, and it runs only under
259 \b \i\c{TASM} is better, but still strives for MASM compatibility,
260 which means millions of directives and tons of red tape. And its syntax
261 is essentially MASM's, with the contradictions and quirks that
262 entails (although it sorts out some of those by means of Ideal mode.)
263 It's expensive too. And it's DOS-only.
265 So here, for your coding pleasure, is NASM. At present it's
266 still in prototype stage - we don't promise that it can outperform
267 any of these assemblers. But please, \e{please} send us bug reports,
268 fixes, helpful information, and anything else you can get your hands
269 on (and thanks to the many people who've done this already! You all
270 know who you are), and we'll improve it out of all recognition.
274 \S{legal} License Conditions
276 Please see the file \c{COPYING}, supplied as part of any NASM
277 distribution archive, for the \i{license} conditions under which you
278 may use NASM. NASM is now under the so-called GNU Lesser General
279 Public License, LGPL.
282 \H{contact} Contact Information
284 The current version of NASM (since about 0.98.08) is maintained by a
285 team of developers, accessible through the \c{nasm-devel} mailing list
286 (see below for the link).
287 If you want to report a bug, please read \k{bugs} first.
289 NASM has a \i{WWW page} at
290 \W{http://nasm.sourceforge.net}\c{http://nasm.sourceforge.net}. If it's
291 not there, google for us!
294 The original authors are \i{e\-mail}able as
295 \W{mailto:jules@dsf.org.uk}\c{jules@dsf.org.uk} and
296 \W{mailto:anakin@pobox.com}\c{anakin@pobox.com}.
297 The latter is no longer involved in the development team.
299 \i{New releases} of NASM are uploaded to the official sites
300 \W{http://nasm.sourceforge.net}\c{http://nasm.sourceforge.net}
302 \W{ftp://ftp.kernel.org/pub/software/devel/nasm/}\i\c{ftp.kernel.org}
304 \W{ftp://ibiblio.org/pub/Linux/devel/lang/assemblers/}\i\c{ibiblio.org}.
306 Announcements are posted to
307 \W{news:comp.lang.asm.x86}\i\c{comp.lang.asm.x86},
308 \W{news:alt.lang.asm}\i\c{alt.lang.asm} and
309 \W{news:comp.os.linux.announce}\i\c{comp.os.linux.announce}
311 If you want information about NASM beta releases, and the current
312 development status, please subscribe to the \i\c{nasm-devel} email list
314 \W{http://sourceforge.net/projects/nasm}\c{http://sourceforge.net/projects/nasm}.
317 \H{install} Installation
319 \S{instdos} \i{Installing} NASM under MS-\i{DOS} or Windows
321 Once you've obtained the \i{DOS archive} for NASM, \i\c{nasmXXX.zip}
322 (where \c{XXX} denotes the version number of NASM contained in the
323 archive), unpack it into its own directory (for example \c{c:\\nasm}).
325 The archive will contain four executable files: the NASM executable
326 files \i\c{nasm.exe} and \i\c{nasmw.exe}, and the NDISASM executable
327 files \i\c{ndisasm.exe} and \i\c{ndisasmw.exe}. In each case, the
328 file whose name ends in \c{w} is a \I{Win32}\c{Win32} executable,
329 designed to run under \I{Windows 95}\c{Windows 95} or \I{Windows NT}
330 \c{Windows NT} Intel, and the other one is a 16-bit \I{DOS}\c{DOS}
333 The only file NASM needs to run is its own executable, so copy
334 (at least) one of \c{nasm.exe} and \c{nasmw.exe} to a directory on
335 your PATH, or alternatively edit \i\c{autoexec.bat} to add the
336 \c{nasm} directory to your \i\c{PATH}. (If you're only installing the
337 \c{Win32} version, you may wish to rename it to \c{nasm.exe}.)
339 That's it - NASM is installed. You don't need the nasm directory
340 to be present to run NASM (unless you've added it to your \c{PATH}),
341 so you can delete it if you need to save space; however, you may
342 want to keep the documentation or test programs.
344 If you've downloaded the \i{DOS source archive}, \i\c{nasmXXXs.zip},
345 the \c{nasm} directory will also contain the full NASM \i{source
346 code}, and a selection of \i{Makefiles} you can (hopefully) use to
347 rebuild your copy of NASM from scratch.
349 Note that the source files \c{insnsa.c}, \c{insnsd.c}, \c{insnsi.h}
350 and \c{insnsn.c} are automatically generated from the master
351 instruction table \c{insns.dat} by a Perl script; the file
352 \c{macros.c} is generated from \c{standard.mac} by another Perl
353 script. Although the NASM source distribution includes these generated
354 files, you will need to rebuild them (and hence, will need a Perl
355 interpreter) if you change insns.dat, standard.mac or the
356 documentation. It is possible future source distributions may not
357 include these files at all. Ports of \i{Perl} for a variety of
358 platforms, including DOS and Windows, are available from
359 \W{http://www.cpan.org/ports/}\i{www.cpan.org}.
362 \S{instdos} Installing NASM under \i{Unix}
364 Once you've obtained the \i{Unix source archive} for NASM,
365 \i\c{nasm-X.XX.tar.gz} (where \c{X.XX} denotes the version number of
366 NASM contained in the archive), unpack it into a directory such
367 as \c{/usr/local/src}. The archive, when unpacked, will create its
368 own subdirectory \c{nasm-X.XX}.
370 NASM is an \I{Autoconf}\I\c{configure}auto-configuring package: once
371 you've unpacked it, \c{cd} to the directory it's been unpacked into
372 and type \c{./configure}. This shell script will find the best C
373 compiler to use for building NASM and set up \i{Makefiles}
376 Once NASM has auto-configured, you can type \i\c{make} to build the
377 \c{nasm} and \c{ndisasm} binaries, and then \c{make install} to
378 install them in \c{/usr/local/bin} and install the \i{man pages}
379 \i\c{nasm.1} and \i\c{ndisasm.1} in \c{/usr/local/man/man1}.
380 Alternatively, you can give options such as \c{--prefix} to the
381 configure script (see the file \i\c{INSTALL} for more details), or
382 install the programs yourself.
384 NASM also comes with a set of utilities for handling the \c{RDOFF}
385 custom object-file format, which are in the \i\c{rdoff} subdirectory
386 of the NASM archive. You can build these with \c{make rdf} and
387 install them with \c{make rdf_install}, if you want them.
389 If NASM fails to auto-configure, you may still be able to make it
390 compile by using the fall-back Unix makefile \i\c{Makefile.unx}.
391 Copy or rename that file to \c{Makefile} and try typing \c{make}.
392 There is also a Makefile.unx file in the \c{rdoff} subdirectory.
395 \C{running} Running NASM
397 \H{syntax} NASM \i{Command-Line} Syntax
399 To assemble a file, you issue a command of the form
401 \c nasm -f <format> <filename> [-o <output>]
405 \c nasm -f elf myfile.asm
407 will assemble \c{myfile.asm} into an \c{ELF} object file \c{myfile.o}. And
409 \c nasm -f bin myfile.asm -o myfile.com
411 will assemble \c{myfile.asm} into a raw binary file \c{myfile.com}.
413 To produce a listing file, with the hex codes output from NASM
414 displayed on the left of the original sources, use the \c{-l} option
415 to give a listing file name, for example:
417 \c nasm -f coff myfile.asm -l myfile.lst
419 To get further usage instructions from NASM, try typing
423 As \c{-hf}, this will also list the available output file formats, and what they
426 If you use Linux but aren't sure whether your system is \c{a.out}
431 (in the directory in which you put the NASM binary when you
432 installed it). If it says something like
434 \c nasm: ELF 32-bit LSB executable i386 (386 and up) Version 1
436 then your system is \c{ELF}, and you should use the option \c{-f elf}
437 when you want NASM to produce Linux object files. If it says
439 \c nasm: Linux/i386 demand-paged executable (QMAGIC)
441 or something similar, your system is \c{a.out}, and you should use
442 \c{-f aout} instead (Linux \c{a.out} systems have long been obsolete,
443 and are rare these days.)
445 Like Unix compilers and assemblers, NASM is silent unless it
446 goes wrong: you won't see any output at all, unless it gives error
450 \S{opt-o} The \i\c{-o} Option: Specifying the Output File Name
452 NASM will normally choose the name of your output file for you;
453 precisely how it does this is dependent on the object file format.
454 For Microsoft object file formats (\i\c{obj} and \i\c{win32}), it
455 will remove the \c{.asm} \i{extension} (or whatever extension you
456 like to use - NASM doesn't care) from your source file name and
457 substitute \c{.obj}. For Unix object file formats (\i\c{aout},
458 \i\c{coff}, \i\c{elf}, \i\c{macho} and \i\c{as86}) it will substitute \c{.o}. For
459 \i\c{rdf}, it will use \c{.rdf}, and for the \i\c{bin} format it
460 will simply remove the extension, so that \c{myfile.asm} produces
461 the output file \c{myfile}.
463 If the output file already exists, NASM will overwrite it, unless it
464 has the same name as the input file, in which case it will give a
465 warning and use \i\c{nasm.out} as the output file name instead.
467 For situations in which this behaviour is unacceptable, NASM
468 provides the \c{-o} command-line option, which allows you to specify
469 your desired output file name. You invoke \c{-o} by following it
470 with the name you wish for the output file, either with or without
471 an intervening space. For example:
473 \c nasm -f bin program.asm -o program.com
474 \c nasm -f bin driver.asm -odriver.sys
476 Note that this is a small o, and is different from a capital O , which
477 is used to specify the number of optimisation passes required. See \k{opt-On}.
480 \S{opt-f} The \i\c{-f} Option: Specifying the \i{Output File Format}
482 If you do not supply the \c{-f} option to NASM, it will choose an
483 output file format for you itself. In the distribution versions of
484 NASM, the default is always \i\c{bin}; if you've compiled your own
485 copy of NASM, you can redefine \i\c{OF_DEFAULT} at compile time and
486 choose what you want the default to be.
488 Like \c{-o}, the intervening space between \c{-f} and the output
489 file format is optional; so \c{-f elf} and \c{-felf} are both valid.
491 A complete list of the available output file formats can be given by
492 issuing the command \i\c{nasm -hf}.
495 \S{opt-l} The \i\c{-l} Option: Generating a \i{Listing File}
497 If you supply the \c{-l} option to NASM, followed (with the usual
498 optional space) by a file name, NASM will generate a
499 \i{source-listing file} for you, in which addresses and generated
500 code are listed on the left, and the actual source code, with
501 expansions of multi-line macros (except those which specifically
502 request no expansion in source listings: see \k{nolist}) on the
505 \c nasm -f elf myfile.asm -l myfile.lst
507 If a list file is selected, you may turn off listing for a
508 section of your source with \c{[list -]}, and turn it back on
509 with \c{[list +]}, (the default, obviously). There is no "user
510 form" (without the brackets). This can be used to list only
511 sections of interest, avoiding excessively long listings.
514 \S{opt-M} The \i\c{-M} Option: Generate \i{Makefile Dependencies}.
516 This option can be used to generate makefile dependencies on stdout.
517 This can be redirected to a file for further processing. For example:
519 \c NASM -M myfile.asm > myfile.dep
522 \S{opt-F} The \i\c{-F} Option: Selecting a \i{Debug Information Format}
524 This option is used to select the format of the debug information emitted
525 into the output file, to be used by a debugger (or \e{will} be). Use
526 of this switch does \e{not} enable output of the selected debug info format.
527 Use \c{-g}, see \k{opt-g}, to enable output.
529 A complete list of the available debug file formats for an output format
530 can be seen by issuing the command \i\c{nasm -f <format> -y}. (only
531 "borland" in "-f obj", as of 0.98.35, but "watch this space")
534 This should not be confused with the "-f dbg" output format option which
535 is not built into NASM by default. For information on how
536 to enable it when building from the sources, see \k{dbgfmt}
539 \S{opt-g} The \i\c{-g} Option: Enabling \i{Debug Information}.
541 This option can be used to generate debugging information in the specified
542 format. See: \k{opt-F}. Using \c{-g} without \c{-F} results in emitting
543 debug info in the default format, if any, for the selected output format.
544 If no debug information is currently implemented in the selected output
545 format, \c{-g} is \e{silently ignored}.
548 \S{opt-X} The \i\c{-X} Option: Selecting an \i{Error Reporting Format}
550 This option can be used to select an error reporting format for any
551 error messages that might be produced by NASM.
553 Currently, two error reporting formats may be selected. They are
554 the \c{-Xvc} option and the \c{-Xgnu} option. The GNU format is
555 the default and looks like this:
557 \c filename.asm:65: error: specific error message
559 where \c{filename.asm} is the name of the source file in which the
560 error was detected, \c{65} is the source file line number on which
561 the error was detected, \c{error} is the severity of the error (this
562 could be \c{warning}), and \c{specific error message} is a more
563 detailed text message which should help pinpoint the exact problem.
565 The other format, specified by \c{-Xvc} is the style used by Microsoft
566 Visual C++ and some other programs. It looks like this:
568 \c filename.asm(65) : error: specific error message
570 where the only difference is that the line number is in parentheses
571 instead of being delimited by colons.
573 See also the \c{Visual C++} output format, \k{win32fmt}.
575 \S{opt-E} The \i\c{-E} Option: Send Errors to a File
577 Under \I{DOS}\c{MS-DOS} it can be difficult (though there are ways) to
578 redirect the standard-error output of a program to a file. Since
579 NASM usually produces its warning and \i{error messages} on
580 \i\c{stderr}, this can make it hard to capture the errors if (for
581 example) you want to load them into an editor.
583 NASM therefore provides the \c{-E} option, taking a filename argument
584 which causes errors to be sent to the specified files rather than
585 standard error. Therefore you can \I{redirecting errors}redirect
586 the errors into a file by typing
588 \c nasm -E myfile.err -f obj myfile.asm
591 \S{opt-s} The \i\c{-s} Option: Send Errors to \i\c{stdout}
593 The \c{-s} option redirects \i{error messages} to \c{stdout} rather
594 than \c{stderr}, so it can be redirected under \I{DOS}\c{MS-DOS}. To
595 assemble the file \c{myfile.asm} and pipe its output to the \c{more}
596 program, you can type:
598 \c nasm -s -f obj myfile.asm | more
600 See also the \c{-E} option, \k{opt-E}.
603 \S{opt-i} The \i\c{-i}\I\c{-I} Option: Include File Search Directories
605 When NASM sees the \i\c{%include} or \i\c{incbin} directive in
606 a source file (see \k{include} or \k{incbin}),
607 it will search for the given file not only in the
608 current directory, but also in any directories specified on the
609 command line by the use of the \c{-i} option. Therefore you can
610 include files from a \i{macro library}, for example, by typing
612 \c nasm -ic:\macrolib\ -f obj myfile.asm
614 (As usual, a space between \c{-i} and the path name is allowed, and
617 NASM, in the interests of complete source-code portability, does not
618 understand the file naming conventions of the OS it is running on;
619 the string you provide as an argument to the \c{-i} option will be
620 prepended exactly as written to the name of the include file.
621 Therefore the trailing backslash in the above example is necessary.
622 Under Unix, a trailing forward slash is similarly necessary.
624 (You can use this to your advantage, if you're really \i{perverse},
625 by noting that the option \c{-ifoo} will cause \c{%include "bar.i"}
626 to search for the file \c{foobar.i}...)
628 If you want to define a \e{standard} \i{include search path},
629 similar to \c{/usr/include} on Unix systems, you should place one or
630 more \c{-i} directives in the \c{NASMENV} environment variable (see
633 For Makefile compatibility with many C compilers, this option can also
634 be specified as \c{-I}.
637 \S{opt-p} The \i\c{-p}\I\c{-P} Option: \I{pre-including files}Pre-Include a File
639 \I\c{%include}NASM allows you to specify files to be
640 \e{pre-included} into your source file, by the use of the \c{-p}
643 \c nasm myfile.asm -p myinc.inc
645 is equivalent to running \c{nasm myfile.asm} and placing the
646 directive \c{%include "myinc.inc"} at the start of the file.
648 For consistency with the \c{-I}, \c{-D} and \c{-U} options, this
649 option can also be specified as \c{-P}.
652 \S{opt-d} The \i\c{-d}\I\c{-D} Option: \I{pre-defining macros}Pre-Define a Macro
654 \I\c{%define}Just as the \c{-p} option gives an alternative to placing
655 \c{%include} directives at the start of a source file, the \c{-d}
656 option gives an alternative to placing a \c{%define} directive. You
659 \c nasm myfile.asm -dFOO=100
661 as an alternative to placing the directive
665 at the start of the file. You can miss off the macro value, as well:
666 the option \c{-dFOO} is equivalent to coding \c{%define FOO}. This
667 form of the directive may be useful for selecting \i{assembly-time
668 options} which are then tested using \c{%ifdef}, for example
671 For Makefile compatibility with many C compilers, this option can also
672 be specified as \c{-D}.
675 \S{opt-u} The \i\c{-u}\I\c{-U} Option: \I{Undefining macros}Undefine a Macro
677 \I\c{%undef}The \c{-u} option undefines a macro that would otherwise
678 have been pre-defined, either automatically or by a \c{-p} or \c{-d}
679 option specified earlier on the command lines.
681 For example, the following command line:
683 \c nasm myfile.asm -dFOO=100 -uFOO
685 would result in \c{FOO} \e{not} being a predefined macro in the
686 program. This is useful to override options specified at a different
689 For Makefile compatibility with many C compilers, this option can also
690 be specified as \c{-U}.
693 \S{opt-e} The \i\c{-e} Option: Preprocess Only
695 NASM allows the \i{preprocessor} to be run on its own, up to a
696 point. Using the \c{-e} option (which requires no arguments) will
697 cause NASM to preprocess its input file, expand all the macro
698 references, remove all the comments and preprocessor directives, and
699 print the resulting file on standard output (or save it to a file,
700 if the \c{-o} option is also used).
702 This option cannot be applied to programs which require the
703 preprocessor to evaluate \I{preprocessor expressions}\i{expressions}
704 which depend on the values of symbols: so code such as
706 \c %assign tablesize ($-tablestart)
708 will cause an error in \i{preprocess-only mode}.
711 \S{opt-a} The \i\c{-a} Option: Don't Preprocess At All
713 If NASM is being used as the back end to a compiler, it might be
714 desirable to \I{suppressing preprocessing}suppress preprocessing
715 completely and assume the compiler has already done it, to save time
716 and increase compilation speeds. The \c{-a} option, requiring no
717 argument, instructs NASM to replace its powerful \i{preprocessor}
718 with a \i{stub preprocessor} which does nothing.
721 \S{opt-On} The \i\c{-On} Option: Specifying \i{Multipass Optimization}.
723 NASM defaults to being a two pass assembler. This means that if you
724 have a complex source file which needs more than 2 passes to assemble
725 optimally, you have to enable extra passes.
727 Using the \c{-O} option, you can tell NASM to carry out multiple passes.
730 \b \c{-O0} strict two-pass assembly, JMP and Jcc are handled more
731 like v0.98, except that backward JMPs are short, if possible.
732 Immediate operands take their long forms if a short form is
735 \b \c{-O1} strict two-pass assembly, but forward branches are assembled
736 with code guaranteed to reach; may produce larger code than
737 -O0, but will produce successful assembly more often if
738 branch offset sizes are not specified.
739 Additionally, immediate operands which will fit in a signed byte
740 are optimized, unless the long form is specified.
742 \b \c{-On} multi-pass optimization, minimize branch offsets; also will
743 minimize signed immediate bytes, overriding size specification
744 unless the \c{strict} keyword has been used (see \k{strict}).
745 The number specifies the maximum number of passes. The more
746 passes, the better the code, but the slower is the assembly.
748 Note that this is a capital O, and is different from a small o, which
749 is used to specify the output format. See \k{opt-o}.
752 \S{opt-t} The \i\c{-t} option: Enable TASM Compatibility Mode
754 NASM includes a limited form of compatibility with Borland's \i\c{TASM}.
755 When NASM's \c{-t} option is used, the following changes are made:
757 \b local labels may be prefixed with \c{@@} instead of \c{.}
759 \b TASM-style response files beginning with \c{@} may be specified on
760 the command line. This is different from the \c{-@resp} style that NASM
763 \b size override is supported within brackets. In TASM compatible mode,
764 a size override inside square brackets changes the size of the operand,
765 and not the address type of the operand as it does in NASM syntax. E.g.
766 \c{mov eax,[DWORD val]} is valid syntax in TASM compatibility mode.
767 Note that you lose the ability to override the default address type for
770 \b \c{%arg} preprocessor directive is supported which is similar to
771 TASM's \c{ARG} directive.
773 \b \c{%local} preprocessor directive
775 \b \c{%stacksize} preprocessor directive
777 \b unprefixed forms of some directives supported (\c{arg}, \c{elif},
778 \c{else}, \c{endif}, \c{if}, \c{ifdef}, \c{ifdifi}, \c{ifndef},
779 \c{include}, \c{local})
783 For more information on the directives, see the section on TASM
784 Compatiblity preprocessor directives in \k{tasmcompat}.
787 \S{opt-w} The \i\c{-w} Option: Enable or Disable Assembly \i{Warnings}
789 NASM can observe many conditions during the course of assembly which
790 are worth mentioning to the user, but not a sufficiently severe
791 error to justify NASM refusing to generate an output file. These
792 conditions are reported like errors, but come up with the word
793 `warning' before the message. Warnings do not prevent NASM from
794 generating an output file and returning a success status to the
797 Some conditions are even less severe than that: they are only
798 sometimes worth mentioning to the user. Therefore NASM supports the
799 \c{-w} command-line option, which enables or disables certain
800 classes of assembly warning. Such warning classes are described by a
801 name, for example \c{orphan-labels}; you can enable warnings of
802 this class by the command-line option \c{-w+orphan-labels} and
803 disable it by \c{-w-orphan-labels}.
805 The \i{suppressible warning} classes are:
807 \b \i\c{macro-params} covers warnings about \i{multi-line macros}
808 being invoked with the wrong number of parameters. This warning
809 class is enabled by default; see \k{mlmacover} for an example of why
810 you might want to disable it.
812 \b \i\c{macro-selfref} warns if a macro references itself. This
813 warning class is enabled by default.
815 \b \i\c{orphan-labels} covers warnings about source lines which
816 contain no instruction but define a label without a trailing colon.
817 NASM does not warn about this somewhat obscure condition by default;
818 see \k{syntax} for an example of why you might want it to.
820 \b \i\c{number-overflow} covers warnings about numeric constants which
821 don't fit in 32 bits (for example, it's easy to type one too many Fs
822 and produce \c{0x7ffffffff} by mistake). This warning class is
825 \b \i\c{gnu-elf-extensions} warns if 8-bit or 16-bit relocations
826 are used in \c{-f elf} format. The GNU extensions allow this.
827 This warning class is enabled by default.
829 \b In addition, warning classes may be enabled or disabled across
830 sections of source code with \i\c{[warning +warning-name]} or
831 \i\c{[warning -warning-name]}. No "user form" (without the
835 \S{opt-v} The \i\c{-v} Option: Display \i{Version} Info
837 Typing \c{NASM -v} will display the version of NASM which you are using,
838 and the date on which it was compiled. This replaces the deprecated
841 You will need the version number if you report a bug.
843 \S{opt-y} The \i\c{-y} Option: Display Available Debug Info Formats
845 Typing \c{nasm -f <option> -y} will display a list of the available
846 debug info formats for the given output format. The default format
847 is indicated by an asterisk. E.g. \c{nasm -f obj -y} yields \c{* borland}.
848 (as of 0.98.35, the \e{only} debug info format implemented).
851 \S{opt-pfix} The \i\c{--prefix} and \i\c{--postfix} Options.
853 The \c{--prefix} and \c{--postfix} options prepend or append
854 (respectively) the given argument to all \c{global} or
855 \c{extern} variables. E.g. \c{--prefix_} will prepend the
856 underscore to all global and external variables, as C sometimes
857 (but not always) likes it.
860 \S{nasmenv} The \c{NASMENV} \i{Environment} Variable
862 If you define an environment variable called \c{NASMENV}, the program
863 will interpret it as a list of extra command-line options, which are
864 processed before the real command line. You can use this to define
865 standard search directories for include files, by putting \c{-i}
866 options in the \c{NASMENV} variable.
868 The value of the variable is split up at white space, so that the
869 value \c{-s -ic:\\nasmlib} will be treated as two separate options.
870 However, that means that the value \c{-dNAME="my name"} won't do
871 what you might want, because it will be split at the space and the
872 NASM command-line processing will get confused by the two
873 nonsensical words \c{-dNAME="my} and \c{name"}.
875 To get round this, NASM provides a feature whereby, if you begin the
876 \c{NASMENV} environment variable with some character that isn't a minus
877 sign, then NASM will treat this character as the \i{separator
878 character} for options. So setting the \c{NASMENV} variable to the
879 value \c{!-s!-ic:\\nasmlib} is equivalent to setting it to \c{-s
880 -ic:\\nasmlib}, but \c{!-dNAME="my name"} will work.
882 This environment variable was previously called \c{NASM}. This was
883 changed with version 0.98.31.
886 \H{qstart} \i{Quick Start} for \i{MASM} Users
888 If you're used to writing programs with MASM, or with \i{TASM} in
889 MASM-compatible (non-Ideal) mode, or with \i\c{a86}, this section
890 attempts to outline the major differences between MASM's syntax and
891 NASM's. If you're not already used to MASM, it's probably worth
892 skipping this section.
895 \S{qscs} NASM Is \I{case sensitivity}Case-Sensitive
897 One simple difference is that NASM is case-sensitive. It makes a
898 difference whether you call your label \c{foo}, \c{Foo} or \c{FOO}.
899 If you're assembling to \c{DOS} or \c{OS/2} \c{.OBJ} files, you can
900 invoke the \i\c{UPPERCASE} directive (documented in \k{objfmt}) to
901 ensure that all symbols exported to other code modules are forced
902 to be upper case; but even then, \e{within} a single module, NASM
903 will distinguish between labels differing only in case.
906 \S{qsbrackets} NASM Requires \i{Square Brackets} For \i{Memory References}
908 NASM was designed with simplicity of syntax in mind. One of the
909 \i{design goals} of NASM is that it should be possible, as far as is
910 practical, for the user to look at a single line of NASM code
911 and tell what opcode is generated by it. You can't do this in MASM:
912 if you declare, for example,
917 then the two lines of code
922 generate completely different opcodes, despite having
923 identical-looking syntaxes.
925 NASM avoids this undesirable situation by having a much simpler
926 syntax for memory references. The rule is simply that any access to
927 the \e{contents} of a memory location requires square brackets
928 around the address, and any access to the \e{address} of a variable
929 doesn't. So an instruction of the form \c{mov ax,foo} will
930 \e{always} refer to a compile-time constant, whether it's an \c{EQU}
931 or the address of a variable; and to access the \e{contents} of the
932 variable \c{bar}, you must code \c{mov ax,[bar]}.
934 This also means that NASM has no need for MASM's \i\c{OFFSET}
935 keyword, since the MASM code \c{mov ax,offset bar} means exactly the
936 same thing as NASM's \c{mov ax,bar}. If you're trying to get
937 large amounts of MASM code to assemble sensibly under NASM, you
938 can always code \c{%idefine offset} to make the preprocessor treat
939 the \c{OFFSET} keyword as a no-op.
941 This issue is even more confusing in \i\c{a86}, where declaring a
942 label with a trailing colon defines it to be a `label' as opposed to
943 a `variable' and causes \c{a86} to adopt NASM-style semantics; so in
944 \c{a86}, \c{mov ax,var} has different behaviour depending on whether
945 \c{var} was declared as \c{var: dw 0} (a label) or \c{var dw 0} (a
946 word-size variable). NASM is very simple by comparison:
947 \e{everything} is a label.
949 NASM, in the interests of simplicity, also does not support the
950 \i{hybrid syntaxes} supported by MASM and its clones, such as
951 \c{mov ax,table[bx]}, where a memory reference is denoted by one
952 portion outside square brackets and another portion inside. The
953 correct syntax for the above is \c{mov ax,[table+bx]}. Likewise,
954 \c{mov ax,es:[di]} is wrong and \c{mov ax,[es:di]} is right.
957 \S{qstypes} NASM Doesn't Store \i{Variable Types}
959 NASM, by design, chooses not to remember the types of variables you
960 declare. Whereas MASM will remember, on seeing \c{var dw 0}, that
961 you declared \c{var} as a word-size variable, and will then be able
962 to fill in the \i{ambiguity} in the size of the instruction \c{mov
963 var,2}, NASM will deliberately remember nothing about the symbol
964 \c{var} except where it begins, and so you must explicitly code
965 \c{mov word [var],2}.
967 For this reason, NASM doesn't support the \c{LODS}, \c{MOVS},
968 \c{STOS}, \c{SCAS}, \c{CMPS}, \c{INS}, or \c{OUTS} instructions,
969 but only supports the forms such as \c{LODSB}, \c{MOVSW}, and
970 \c{SCASD}, which explicitly specify the size of the components of
971 the strings being manipulated.
974 \S{qsassume} NASM Doesn't \i\c{ASSUME}
976 As part of NASM's drive for simplicity, it also does not support the
977 \c{ASSUME} directive. NASM will not keep track of what values you
978 choose to put in your segment registers, and will never
979 \e{automatically} generate a \i{segment override} prefix.
982 \S{qsmodel} NASM Doesn't Support \i{Memory Models}
984 NASM also does not have any directives to support different 16-bit
985 memory models. The programmer has to keep track of which functions
986 are supposed to be called with a \i{far call} and which with a
987 \i{near call}, and is responsible for putting the correct form of
988 \c{RET} instruction (\c{RETN} or \c{RETF}; NASM accepts \c{RET}
989 itself as an alternate form for \c{RETN}); in addition, the
990 programmer is responsible for coding CALL FAR instructions where
991 necessary when calling \e{external} functions, and must also keep
992 track of which external variable definitions are far and which are
996 \S{qsfpu} \i{Floating-Point} Differences
998 NASM uses different names to refer to floating-point registers from
999 MASM: where MASM would call them \c{ST(0)}, \c{ST(1)} and so on, and
1000 \i\c{a86} would call them simply \c{0}, \c{1} and so on, NASM
1001 chooses to call them \c{st0}, \c{st1} etc.
1003 As of version 0.96, NASM now treats the instructions with
1004 \i{`nowait'} forms in the same way as MASM-compatible assemblers.
1005 The idiosyncratic treatment employed by 0.95 and earlier was based
1006 on a misunderstanding by the authors.
1009 \S{qsother} Other Differences
1011 For historical reasons, NASM uses the keyword \i\c{TWORD} where MASM
1012 and compatible assemblers use \i\c{TBYTE}.
1014 NASM does not declare \i{uninitialized storage} in the same way as
1015 MASM: where a MASM programmer might use \c{stack db 64 dup (?)},
1016 NASM requires \c{stack resb 64}, intended to be read as `reserve 64
1017 bytes'. For a limited amount of compatibility, since NASM treats
1018 \c{?} as a valid character in symbol names, you can code \c{? equ 0}
1019 and then writing \c{dw ?} will at least do something vaguely useful.
1020 \I\c{RESB}\i\c{DUP} is still not a supported syntax, however.
1022 In addition to all of this, macros and directives work completely
1023 differently to MASM. See \k{preproc} and \k{directive} for further
1027 \C{lang} The NASM Language
1029 \H{syntax} Layout of a NASM Source Line
1031 Like most assemblers, each NASM source line contains (unless it
1032 is a macro, a preprocessor directive or an assembler directive: see
1033 \k{preproc} and \k{directive}) some combination of the four fields
1035 \c label: instruction operands ; comment
1037 As usual, most of these fields are optional; the presence or absence
1038 of any combination of a label, an instruction and a comment is allowed.
1039 Of course, the operand field is either required or forbidden by the
1040 presence and nature of the instruction field.
1042 NASM uses backslash (\\) as the line continuation character; if a line
1043 ends with backslash, the next line is considered to be a part of the
1044 backslash-ended line.
1046 NASM places no restrictions on white space within a line: labels may
1047 have white space before them, or instructions may have no space
1048 before them, or anything. The \i{colon} after a label is also
1049 optional. (Note that this means that if you intend to code \c{lodsb}
1050 alone on a line, and type \c{lodab} by accident, then that's still a
1051 valid source line which does nothing but define a label. Running
1052 NASM with the command-line option
1053 \I{orphan-labels}\c{-w+orphan-labels} will cause it to warn you if
1054 you define a label alone on a line without a \i{trailing colon}.)
1056 \i{Valid characters} in labels are letters, numbers, \c{_}, \c{$},
1057 \c{#}, \c{@}, \c{~}, \c{.}, and \c{?}. The only characters which may
1058 be used as the \e{first} character of an identifier are letters,
1059 \c{.} (with special meaning: see \k{locallab}), \c{_} and \c{?}.
1060 An identifier may also be prefixed with a \I{$, prefix}\c{$} to
1061 indicate that it is intended to be read as an identifier and not a
1062 reserved word; thus, if some other module you are linking with
1063 defines a symbol called \c{eax}, you can refer to \c{$eax} in NASM
1064 code to distinguish the symbol from the register. Maximum length of
1065 an identifier is 4095 characters.
1067 The instruction field may contain any machine instruction: Pentium
1068 and P6 instructions, FPU instructions, MMX instructions and even
1069 undocumented instructions are all supported. The instruction may be
1070 prefixed by \c{LOCK}, \c{REP}, \c{REPE}/\c{REPZ} or
1071 \c{REPNE}/\c{REPNZ}, in the usual way. Explicit \I{address-size
1072 prefixes}address-size and \i{operand-size prefixes} \c{A16},
1073 \c{A32}, \c{O16} and \c{O32} are provided - one example of their use
1074 is given in \k{mixsize}. You can also use the name of a \I{segment
1075 override}segment register as an instruction prefix: coding
1076 \c{es mov [bx],ax} is equivalent to coding \c{mov [es:bx],ax}. We
1077 recommend the latter syntax, since it is consistent with other
1078 syntactic features of the language, but for instructions such as
1079 \c{LODSB}, which has no operands and yet can require a segment
1080 override, there is no clean syntactic way to proceed apart from
1083 An instruction is not required to use a prefix: prefixes such as
1084 \c{CS}, \c{A32}, \c{LOCK} or \c{REPE} can appear on a line by
1085 themselves, and NASM will just generate the prefix bytes.
1087 In addition to actual machine instructions, NASM also supports a
1088 number of pseudo-instructions, described in \k{pseudop}.
1090 Instruction \i{operands} may take a number of forms: they can be
1091 registers, described simply by the register name (e.g. \c{ax},
1092 \c{bp}, \c{ebx}, \c{cr0}: NASM does not use the \c{gas}-style
1093 syntax in which register names must be prefixed by a \c{%} sign), or
1094 they can be \i{effective addresses} (see \k{effaddr}), constants
1095 (\k{const}) or expressions (\k{expr}).
1097 For x87 \i{floating-point} instructions, NASM accepts a wide range of
1098 syntaxes: you can use two-operand forms like MASM supports, or you
1099 can use NASM's native single-operand forms in most cases.
1101 \# all forms of each supported instruction are given in
1103 For example, you can code:
1105 \c fadd st1 ; this sets st0 := st0 + st1
1106 \c fadd st0,st1 ; so does this
1108 \c fadd st1,st0 ; this sets st1 := st1 + st0
1109 \c fadd to st1 ; so does this
1111 Almost any x87 floating-point instruction that references memory must
1112 use one of the prefixes \i\c{DWORD}, \i\c{QWORD} or \i\c{TWORD} to
1113 indicate what size of \i{memory operand} it refers to.
1116 \H{pseudop} \i{Pseudo-Instructions}
1118 Pseudo-instructions are things which, though not real x86 machine
1119 instructions, are used in the instruction field anyway because that's
1120 the most convenient place to put them. The current pseudo-instructions
1121 are \i\c{DB}, \i\c{DW}, \i\c{DD}, \i\c{DQ}, \i\c{DT} and \i\c{DO};
1122 their \i{uninitialized} counterparts \i\c{RESB}, \i\c{RESW},
1123 \i\c{RESD}, \i\c{RESQ}, \i\c{REST} and \i\c{RESO}; the \i\c{INCBIN}
1124 command, the \i\c{EQU} command, and the \i\c{TIMES} prefix.
1127 \S{db} \c{DB} and friends: Declaring initialized Data
1129 \i\c{DB}, \i\c{DW}, \i\c{DD}, \i\c{DQ}, \i\c{DT} and \i\c{DO} are
1130 used, much as in MASM, to declare initialized data in the output
1131 file. They can be invoked in a wide range of ways:
1132 \I{floating-point}\I{character constant}\I{string constant}
1134 \c db 0x55 ; just the byte 0x55
1135 \c db 0x55,0x56,0x57 ; three bytes in succession
1136 \c db 'a',0x55 ; character constants are OK
1137 \c db 'hello',13,10,'$' ; so are string constants
1138 \c dw 0x1234 ; 0x34 0x12
1139 \c dw 'a' ; 0x61 0x00 (it's just a number)
1140 \c dw 'ab' ; 0x61 0x62 (character constant)
1141 \c dw 'abc' ; 0x61 0x62 0x63 0x00 (string)
1142 \c dd 0x12345678 ; 0x78 0x56 0x34 0x12
1143 \c dd 1.234567e20 ; floating-point constant
1144 \c dq 0x123456789abcdef0 ; eight byte constant
1145 \c dq 1.234567e20 ; double-precision float
1146 \c dt 1.234567e20 ; extended-precision float
1148 \c{DT} and \c{DO} do not accept \i{numeric constants} as operands.
1149 \c{DB} does not accept \i{floating-point} numbers as operands.
1152 \S{resb} \c{RESB} and friends: Declaring \i{Uninitialized} Data
1154 \i\c{RESB}, \i\c{RESW}, \i\c{RESD}, \i\c{RESQ}, \i\c{REST} and
1155 \i\c{RESO} are designed to be used in the BSS section of a module:
1156 they declare \e{uninitialized} storage space. Each takes a single
1157 operand, which is the number of bytes, words, doublewords or whatever
1158 to reserve. As stated in \k{qsother}, NASM does not support the
1159 MASM/TASM syntax of reserving uninitialized space by writing
1160 \I\c{?}\c{DW ?} or similar things: this is what it does instead. The
1161 operand to a \c{RESB}-type pseudo-instruction is a \i\e{critical
1162 expression}: see \k{crit}.
1166 \c buffer: resb 64 ; reserve 64 bytes
1167 \c wordvar: resw 1 ; reserve a word
1168 \c realarray resq 10 ; array of ten reals
1171 \S{incbin} \i\c{INCBIN}: Including External \i{Binary Files}
1173 \c{INCBIN} is borrowed from the old Amiga assembler \i{DevPac}: it
1174 includes a binary file verbatim into the output file. This can be
1175 handy for (for example) including \i{graphics} and \i{sound} data
1176 directly into a game executable file. It can be called in one of
1179 \c incbin "file.dat" ; include the whole file
1180 \c incbin "file.dat",1024 ; skip the first 1024 bytes
1181 \c incbin "file.dat",1024,512 ; skip the first 1024, and
1182 \c ; actually include at most 512
1185 \S{equ} \i\c{EQU}: Defining Constants
1187 \c{EQU} defines a symbol to a given constant value: when \c{EQU} is
1188 used, the source line must contain a label. The action of \c{EQU} is
1189 to define the given label name to the value of its (only) operand.
1190 This definition is absolute, and cannot change later. So, for
1193 \c message db 'hello, world'
1194 \c msglen equ $-message
1196 defines \c{msglen} to be the constant 12. \c{msglen} may not then be
1197 redefined later. This is not a \i{preprocessor} definition either:
1198 the value of \c{msglen} is evaluated \e{once}, using the value of
1199 \c{$} (see \k{expr} for an explanation of \c{$}) at the point of
1200 definition, rather than being evaluated wherever it is referenced
1201 and using the value of \c{$} at the point of reference. Note that
1202 the operand to an \c{EQU} is also a \i{critical expression}
1206 \S{times} \i\c{TIMES}: \i{Repeating} Instructions or Data
1208 The \c{TIMES} prefix causes the instruction to be assembled multiple
1209 times. This is partly present as NASM's equivalent of the \i\c{DUP}
1210 syntax supported by \i{MASM}-compatible assemblers, in that you can
1213 \c zerobuf: times 64 db 0
1215 or similar things; but \c{TIMES} is more versatile than that. The
1216 argument to \c{TIMES} is not just a numeric constant, but a numeric
1217 \e{expression}, so you can do things like
1219 \c buffer: db 'hello, world'
1220 \c times 64-$+buffer db ' '
1222 which will store exactly enough spaces to make the total length of
1223 \c{buffer} up to 64. Finally, \c{TIMES} can be applied to ordinary
1224 instructions, so you can code trivial \i{unrolled loops} in it:
1228 Note that there is no effective difference between \c{times 100 resb
1229 1} and \c{resb 100}, except that the latter will be assembled about
1230 100 times faster due to the internal structure of the assembler.
1232 The operand to \c{TIMES}, like that of \c{EQU} and those of \c{RESB}
1233 and friends, is a critical expression (\k{crit}).
1235 Note also that \c{TIMES} can't be applied to \i{macros}: the reason
1236 for this is that \c{TIMES} is processed after the macro phase, which
1237 allows the argument to \c{TIMES} to contain expressions such as
1238 \c{64-$+buffer} as above. To repeat more than one line of code, or a
1239 complex macro, use the preprocessor \i\c{%rep} directive.
1242 \H{effaddr} Effective Addresses
1244 An \i{effective address} is any operand to an instruction which
1245 \I{memory reference}references memory. Effective addresses, in NASM,
1246 have a very simple syntax: they consist of an expression evaluating
1247 to the desired address, enclosed in \i{square brackets}. For
1252 \c mov ax,[wordvar+1]
1253 \c mov ax,[es:wordvar+bx]
1255 Anything not conforming to this simple system is not a valid memory
1256 reference in NASM, for example \c{es:wordvar[bx]}.
1258 More complicated effective addresses, such as those involving more
1259 than one register, work in exactly the same way:
1261 \c mov eax,[ebx*2+ecx+offset]
1264 NASM is capable of doing \i{algebra} on these effective addresses,
1265 so that things which don't necessarily \e{look} legal are perfectly
1268 \c mov eax,[ebx*5] ; assembles as [ebx*4+ebx]
1269 \c mov eax,[label1*2-label2] ; ie [label1+(label1-label2)]
1271 Some forms of effective address have more than one assembled form;
1272 in most such cases NASM will generate the smallest form it can. For
1273 example, there are distinct assembled forms for the 32-bit effective
1274 addresses \c{[eax*2+0]} and \c{[eax+eax]}, and NASM will generally
1275 generate the latter on the grounds that the former requires four
1276 bytes to store a zero offset.
1278 NASM has a hinting mechanism which will cause \c{[eax+ebx]} and
1279 \c{[ebx+eax]} to generate different opcodes; this is occasionally
1280 useful because \c{[esi+ebp]} and \c{[ebp+esi]} have different
1281 default segment registers.
1283 However, you can force NASM to generate an effective address in a
1284 particular form by the use of the keywords \c{BYTE}, \c{WORD},
1285 \c{DWORD} and \c{NOSPLIT}. If you need \c{[eax+3]} to be assembled
1286 using a double-word offset field instead of the one byte NASM will
1287 normally generate, you can code \c{[dword eax+3]}. Similarly, you
1288 can force NASM to use a byte offset for a small value which it
1289 hasn't seen on the first pass (see \k{crit} for an example of such a
1290 code fragment) by using \c{[byte eax+offset]}. As special cases,
1291 \c{[byte eax]} will code \c{[eax+0]} with a byte offset of zero, and
1292 \c{[dword eax]} will code it with a double-word offset of zero. The
1293 normal form, \c{[eax]}, will be coded with no offset field.
1295 The form described in the previous paragraph is also useful if you
1296 are trying to access data in a 32-bit segment from within 16 bit code.
1297 For more information on this see the section on mixed-size addressing
1298 (\k{mixaddr}). In particular, if you need to access data with a known
1299 offset that is larger than will fit in a 16-bit value, if you don't
1300 specify that it is a dword offset, nasm will cause the high word of
1301 the offset to be lost.
1303 Similarly, NASM will split \c{[eax*2]} into \c{[eax+eax]} because
1304 that allows the offset field to be absent and space to be saved; in
1305 fact, it will also split \c{[eax*2+offset]} into
1306 \c{[eax+eax+offset]}. You can combat this behaviour by the use of
1307 the \c{NOSPLIT} keyword: \c{[nosplit eax*2]} will force
1308 \c{[eax*2+0]} to be generated literally.
1310 In 64-bit mode, NASM will by default generate absolute addresses. The
1311 \i\c{REL} keyword makes it produce \c{RIP}-relative addresses. Since
1312 this is frequently the normally desired behaviour, see the \c{DEFAULT}
1313 directive (\k{default}). The keyword \i\c{ABS} overrides \i\c{REL}.
1316 \H{const} \i{Constants}
1318 NASM understands four different types of constant: numeric,
1319 character, string and floating-point.
1322 \S{numconst} \i{Numeric Constants}
1324 A numeric constant is simply a number. NASM allows you to specify
1325 numbers in a variety of number bases, in a variety of ways: you can
1326 suffix \c{H}, \c{Q} or \c{O}, and \c{B} for \i{hex}, \i{octal} and \i{binary},
1327 or you can prefix \c{0x} for hex in the style of C, or you can
1328 prefix \c{$} for hex in the style of Borland Pascal. Note, though,
1329 that the \I{$, prefix}\c{$} prefix does double duty as a prefix on
1330 identifiers (see \k{syntax}), so a hex number prefixed with a \c{$}
1331 sign must have a digit after the \c{$} rather than a letter.
1335 \c mov ax,100 ; decimal
1336 \c mov ax,0a2h ; hex
1337 \c mov ax,$0a2 ; hex again: the 0 is required
1338 \c mov ax,0xa2 ; hex yet again
1339 \c mov ax,777q ; octal
1340 \c mov ax,777o ; octal again
1341 \c mov ax,10010011b ; binary
1344 \S{chrconst} \i{Character Constants}
1346 A character constant consists of up to four characters enclosed in
1347 either single or double quotes. The type of quote makes no
1348 difference to NASM, except of course that surrounding the constant
1349 with single quotes allows double quotes to appear within it and vice
1352 A character constant with more than one character will be arranged
1353 with \i{little-endian} order in mind: if you code
1357 then the constant generated is not \c{0x61626364}, but
1358 \c{0x64636261}, so that if you were then to store the value into
1359 memory, it would read \c{abcd} rather than \c{dcba}. This is also
1360 the sense of character constants understood by the Pentium's
1361 \i\c{CPUID} instruction.
1362 \# (see \k{insCPUID})
1365 \S{strconst} String Constants
1367 String constants are only acceptable to some pseudo-instructions,
1368 namely the \I\c{DW}\I\c{DD}\I\c{DQ}\I\c{DT}\i\c{DB} family and
1371 A string constant looks like a character constant, only longer. It
1372 is treated as a concatenation of maximum-size character constants
1373 for the conditions. So the following are equivalent:
1375 \c db 'hello' ; string constant
1376 \c db 'h','e','l','l','o' ; equivalent character constants
1378 And the following are also equivalent:
1380 \c dd 'ninechars' ; doubleword string constant
1381 \c dd 'nine','char','s' ; becomes three doublewords
1382 \c db 'ninechars',0,0,0 ; and really looks like this
1384 Note that when used as an operand to \c{db}, a constant like
1385 \c{'ab'} is treated as a string constant despite being short enough
1386 to be a character constant, because otherwise \c{db 'ab'} would have
1387 the same effect as \c{db 'a'}, which would be silly. Similarly,
1388 three-character or four-character constants are treated as strings
1389 when they are operands to \c{dw}.
1392 \S{fltconst} \I{floating-point, constants}Floating-Point Constants
1394 \i{Floating-point} constants are acceptable only as arguments to
1395 \i\c{DW}, \i\c{DD}, \i\c{DQ}, \i\c{DT}, and \i\c{DO}, or as arguments
1396 to the special operators \i\c{__float16__}, \i\c{__float32__},
1397 \i\c{__float64__}, \i\c{__float80m__}, \i\c{__float80e__},
1398 \i\c{__float128l__}, and \i\c{__float128h__}.
1400 Floating-point constants are expressed in the traditional form:
1401 digits, then a period, then optionally more digits, then optionally an
1402 \c{E} followed by an exponent. The period is mandatory, so that NASM
1403 can distinguish between \c{dd 1}, which declares an integer constant,
1404 and \c{dd 1.0} which declares a floating-point constant. NASM also
1405 support C99-style hexadecimal floating-point: \c{0x}, hexadecimal
1406 digits, period, optionally more hexadeximal digits, then optionally a
1407 \c{P} followed by a \e{binary} (not hexadecimal) exponent in decimal
1412 \c dw -0.5 ; IEEE half precision
1413 \c dd 1.2 ; an easy one
1414 \c dd 0x1p+2 ; 1.0x2^2 = 4.0
1415 \c dq 1.e10 ; 10,000,000,000
1416 \c dq 1.e+10 ; synonymous with 1.e10
1417 \c dq 1.e-10 ; 0.000 000 000 1
1418 \c dt 3.141592653589793238462 ; pi
1419 \c do 1.e+4000 ; IEEE quad precision
1421 The special operators are used to produce floating-point numbers in
1422 other contexts. They produce the binary representation of a specific
1423 floating-point number as an integer, and can use anywhere integer
1424 constants are used in an expression. \c{__float80m__} and
1425 \c{__float80e__} produce the 64-bit mantissa and 16-bit exponent of an
1426 80-bit floating-point number, and \c{__float128l__} and
1427 \c{__float128h__} produce the lower and upper 64-bit half of a 128-bit
1428 floating-point number, respectively.
1432 \c mov rax,__float64__(3.141592653589793238462)
1434 ... would assign the binary representation of pi as a 64-bit floating
1435 point number into \c{RAX}.
1437 NASM cannot do compile-time arithmetic on floating-point constants.
1438 This is because NASM is designed to be portable - although it always
1439 generates code to run on x86 processors, the assembler itself can
1440 run on any system with an ANSI C compiler. Therefore, the assembler
1441 cannot guarantee the presence of a floating-point unit capable of
1442 handling the \i{Intel number formats}, and so for NASM to be able to
1443 do floating arithmetic it would have to include its own complete set
1444 of floating-point routines, which would significantly increase the
1445 size of the assembler for very little benefit.
1447 The special tokens \i\c{__Infinity__}, \i\c{__QNaN__} (or
1448 \i\c{__NaN__}) and \i\c{__SNaN__} can be used to generate
1449 \I{infinity}infinities, quiet \i{NaN}s, and signalling NaNs,
1450 respectively. These are normally used as macros:
1452 \c %define Inf __Infinity__
1453 \c %define NaN __QNaN__
1455 \c dq +1.5, -Inf, NaN ; Double-precision constants
1457 \H{expr} \i{Expressions}
1459 Expressions in NASM are similar in syntax to those in C. Expressions
1460 are evaluated as 64-bit integers which are then adjusted to the
1463 NASM supports two special tokens in expressions, allowing
1464 calculations to involve the current assembly position: the
1465 \I{$, here}\c{$} and \i\c{$$} tokens. \c{$} evaluates to the assembly
1466 position at the beginning of the line containing the expression; so
1467 you can code an \i{infinite loop} using \c{JMP $}. \c{$$} evaluates
1468 to the beginning of the current section; so you can tell how far
1469 into the section you are by using \c{($-$$)}.
1471 The arithmetic \i{operators} provided by NASM are listed here, in
1472 increasing order of \i{precedence}.
1475 \S{expor} \i\c{|}: \i{Bitwise OR} Operator
1477 The \c{|} operator gives a bitwise OR, exactly as performed by the
1478 \c{OR} machine instruction. Bitwise OR is the lowest-priority
1479 arithmetic operator supported by NASM.
1482 \S{expxor} \i\c{^}: \i{Bitwise XOR} Operator
1484 \c{^} provides the bitwise XOR operation.
1487 \S{expand} \i\c{&}: \i{Bitwise AND} Operator
1489 \c{&} provides the bitwise AND operation.
1492 \S{expshift} \i\c{<<} and \i\c{>>}: \i{Bit Shift} Operators
1494 \c{<<} gives a bit-shift to the left, just as it does in C. So \c{5<<3}
1495 evaluates to 5 times 8, or 40. \c{>>} gives a bit-shift to the
1496 right; in NASM, such a shift is \e{always} unsigned, so that
1497 the bits shifted in from the left-hand end are filled with zero
1498 rather than a sign-extension of the previous highest bit.
1501 \S{expplmi} \I{+ opaddition}\c{+} and \I{- opsubtraction}\c{-}:
1502 \i{Addition} and \i{Subtraction} Operators
1504 The \c{+} and \c{-} operators do perfectly ordinary addition and
1508 \S{expmul} \i\c{*}, \i\c{/}, \i\c{//}, \i\c{%} and \i\c{%%}:
1509 \i{Multiplication} and \i{Division}
1511 \c{*} is the multiplication operator. \c{/} and \c{//} are both
1512 division operators: \c{/} is \i{unsigned division} and \c{//} is
1513 \i{signed division}. Similarly, \c{%} and \c{%%} provide \I{unsigned
1514 modulo}\I{modulo operators}unsigned and
1515 \i{signed modulo} operators respectively.
1517 NASM, like ANSI C, provides no guarantees about the sensible
1518 operation of the signed modulo operator.
1520 Since the \c{%} character is used extensively by the macro
1521 \i{preprocessor}, you should ensure that both the signed and unsigned
1522 modulo operators are followed by white space wherever they appear.
1525 \S{expmul} \i{Unary Operators}: \I{+ opunary}\c{+}, \I{- opunary}\c{-},
1526 \i\c{~}, \I{! opunary}\c{!} and \i\c{SEG}
1528 The highest-priority operators in NASM's expression grammar are
1529 those which only apply to one argument. \c{-} negates its operand,
1530 \c{+} does nothing (it's provided for symmetry with \c{-}), \c{~}
1531 computes the \i{one's complement} of its operand, \c{!} is the
1532 \i{logical negation} operator, and \c{SEG} provides the \i{segment address}
1533 of its operand (explained in more detail in \k{segwrt}).
1536 \H{segwrt} \i\c{SEG} and \i\c{WRT}
1538 When writing large 16-bit programs, which must be split into
1539 multiple \i{segments}, it is often necessary to be able to refer to
1540 the \I{segment address}segment part of the address of a symbol. NASM
1541 supports the \c{SEG} operator to perform this function.
1543 The \c{SEG} operator returns the \i\e{preferred} segment base of a
1544 symbol, defined as the segment base relative to which the offset of
1545 the symbol makes sense. So the code
1547 \c mov ax,seg symbol
1551 will load \c{ES:BX} with a valid pointer to the symbol \c{symbol}.
1553 Things can be more complex than this: since 16-bit segments and
1554 \i{groups} may \I{overlapping segments}overlap, you might occasionally
1555 want to refer to some symbol using a different segment base from the
1556 preferred one. NASM lets you do this, by the use of the \c{WRT}
1557 (With Reference To) keyword. So you can do things like
1559 \c mov ax,weird_seg ; weird_seg is a segment base
1561 \c mov bx,symbol wrt weird_seg
1563 to load \c{ES:BX} with a different, but functionally equivalent,
1564 pointer to the symbol \c{symbol}.
1566 NASM supports far (inter-segment) calls and jumps by means of the
1567 syntax \c{call segment:offset}, where \c{segment} and \c{offset}
1568 both represent immediate values. So to call a far procedure, you
1569 could code either of
1571 \c call (seg procedure):procedure
1572 \c call weird_seg:(procedure wrt weird_seg)
1574 (The parentheses are included for clarity, to show the intended
1575 parsing of the above instructions. They are not necessary in
1578 NASM supports the syntax \I\c{CALL FAR}\c{call far procedure} as a
1579 synonym for the first of the above usages. \c{JMP} works identically
1580 to \c{CALL} in these examples.
1582 To declare a \i{far pointer} to a data item in a data segment, you
1585 \c dw symbol, seg symbol
1587 NASM supports no convenient synonym for this, though you can always
1588 invent one using the macro processor.
1591 \H{strict} \i\c{STRICT}: Inhibiting Optimization
1593 When assembling with the optimizer set to level 2 or higher (see
1594 \k{opt-On}), NASM will use size specifiers (\c{BYTE}, \c{WORD},
1595 \c{DWORD}, \c{QWORD}, \c{TWORD} or \c{OWORD}), but will give them the
1596 smallest possible size. The keyword \c{STRICT} can be used to inhibit
1597 optimization and force a particular operand to be emitted in the
1598 specified size. For example, with the optimizer on, and in \c{BITS 16}
1603 is encoded in three bytes \c{66 6A 21}, whereas
1605 \c push strict dword 33
1607 is encoded in six bytes, with a full dword immediate operand \c{66 68
1610 With the optimizer off, the same code (six bytes) is generated whether
1611 the \c{STRICT} keyword was used or not.
1614 \H{crit} \i{Critical Expressions}
1616 A limitation of NASM is that it is a \i{two-pass assembler}; unlike
1617 TASM and others, it will always do exactly two \I{passes}\i{assembly
1618 passes}. Therefore it is unable to cope with source files that are
1619 complex enough to require three or more passes.
1621 The first pass is used to determine the size of all the assembled
1622 code and data, so that the second pass, when generating all the
1623 code, knows all the symbol addresses the code refers to. So one
1624 thing NASM can't handle is code whose size depends on the value of a
1625 symbol declared after the code in question. For example,
1627 \c times (label-$) db 0
1628 \c label: db 'Where am I?'
1630 The argument to \i\c{TIMES} in this case could equally legally
1631 evaluate to anything at all; NASM will reject this example because
1632 it cannot tell the size of the \c{TIMES} line when it first sees it.
1633 It will just as firmly reject the slightly \I{paradox}paradoxical
1636 \c times (label-$+1) db 0
1637 \c label: db 'NOW where am I?'
1639 in which \e{any} value for the \c{TIMES} argument is by definition
1642 NASM rejects these examples by means of a concept called a
1643 \e{critical expression}, which is defined to be an expression whose
1644 value is required to be computable in the first pass, and which must
1645 therefore depend only on symbols defined before it. The argument to
1646 the \c{TIMES} prefix is a critical expression; for the same reason,
1647 the arguments to the \i\c{RESB} family of pseudo-instructions are
1648 also critical expressions.
1650 Critical expressions can crop up in other contexts as well: consider
1654 \c symbol1 equ symbol2
1657 On the first pass, NASM cannot determine the value of \c{symbol1},
1658 because \c{symbol1} is defined to be equal to \c{symbol2} which NASM
1659 hasn't seen yet. On the second pass, therefore, when it encounters
1660 the line \c{mov ax,symbol1}, it is unable to generate the code for
1661 it because it still doesn't know the value of \c{symbol1}. On the
1662 next line, it would see the \i\c{EQU} again and be able to determine
1663 the value of \c{symbol1}, but by then it would be too late.
1665 NASM avoids this problem by defining the right-hand side of an
1666 \c{EQU} statement to be a critical expression, so the definition of
1667 \c{symbol1} would be rejected in the first pass.
1669 There is a related issue involving \i{forward references}: consider
1672 \c mov eax,[ebx+offset]
1675 NASM, on pass one, must calculate the size of the instruction \c{mov
1676 eax,[ebx+offset]} without knowing the value of \c{offset}. It has no
1677 way of knowing that \c{offset} is small enough to fit into a
1678 one-byte offset field and that it could therefore get away with
1679 generating a shorter form of the \i{effective-address} encoding; for
1680 all it knows, in pass one, \c{offset} could be a symbol in the code
1681 segment, and it might need the full four-byte form. So it is forced
1682 to compute the size of the instruction to accommodate a four-byte
1683 address part. In pass two, having made this decision, it is now
1684 forced to honour it and keep the instruction large, so the code
1685 generated in this case is not as small as it could have been. This
1686 problem can be solved by defining \c{offset} before using it, or by
1687 forcing byte size in the effective address by coding \c{[byte
1690 Note that use of the \c{-On} switch (with n>=2) makes some of the above
1691 no longer true (see \k{opt-On}).
1693 \H{locallab} \i{Local Labels}
1695 NASM gives special treatment to symbols beginning with a \i{period}.
1696 A label beginning with a single period is treated as a \e{local}
1697 label, which means that it is associated with the previous non-local
1698 label. So, for example:
1700 \c label1 ; some code
1708 \c label2 ; some code
1716 In the above code fragment, each \c{JNE} instruction jumps to the
1717 line immediately before it, because the two definitions of \c{.loop}
1718 are kept separate by virtue of each being associated with the
1719 previous non-local label.
1721 This form of local label handling is borrowed from the old Amiga
1722 assembler \i{DevPac}; however, NASM goes one step further, in
1723 allowing access to local labels from other parts of the code. This
1724 is achieved by means of \e{defining} a local label in terms of the
1725 previous non-local label: the first definition of \c{.loop} above is
1726 really defining a symbol called \c{label1.loop}, and the second
1727 defines a symbol called \c{label2.loop}. So, if you really needed
1730 \c label3 ; some more code
1735 Sometimes it is useful - in a macro, for instance - to be able to
1736 define a label which can be referenced from anywhere but which
1737 doesn't interfere with the normal local-label mechanism. Such a
1738 label can't be non-local because it would interfere with subsequent
1739 definitions of, and references to, local labels; and it can't be
1740 local because the macro that defined it wouldn't know the label's
1741 full name. NASM therefore introduces a third type of label, which is
1742 probably only useful in macro definitions: if a label begins with
1743 the \I{label prefix}special prefix \i\c{..@}, then it does nothing
1744 to the local label mechanism. So you could code
1746 \c label1: ; a non-local label
1747 \c .local: ; this is really label1.local
1748 \c ..@foo: ; this is a special symbol
1749 \c label2: ; another non-local label
1750 \c .local: ; this is really label2.local
1752 \c jmp ..@foo ; this will jump three lines up
1754 NASM has the capacity to define other special symbols beginning with
1755 a double period: for example, \c{..start} is used to specify the
1756 entry point in the \c{obj} output format (see \k{dotdotstart}).
1759 \C{preproc} The NASM \i{Preprocessor}
1761 NASM contains a powerful \i{macro processor}, which supports
1762 conditional assembly, multi-level file inclusion, two forms of macro
1763 (single-line and multi-line), and a `context stack' mechanism for
1764 extra macro power. Preprocessor directives all begin with a \c{%}
1767 The preprocessor collapses all lines which end with a backslash (\\)
1768 character into a single line. Thus:
1770 \c %define THIS_VERY_LONG_MACRO_NAME_IS_DEFINED_TO \\
1773 will work like a single-line macro without the backslash-newline
1776 \H{slmacro} \i{Single-Line Macros}
1778 \S{define} The Normal Way: \I\c{%idefine}\i\c{%define}
1780 Single-line macros are defined using the \c{%define} preprocessor
1781 directive. The definitions work in a similar way to C; so you can do
1784 \c %define ctrl 0x1F &
1785 \c %define param(a,b) ((a)+(a)*(b))
1787 \c mov byte [param(2,ebx)], ctrl 'D'
1789 which will expand to
1791 \c mov byte [(2)+(2)*(ebx)], 0x1F & 'D'
1793 When the expansion of a single-line macro contains tokens which
1794 invoke another macro, the expansion is performed at invocation time,
1795 not at definition time. Thus the code
1797 \c %define a(x) 1+b(x)
1802 will evaluate in the expected way to \c{mov ax,1+2*8}, even though
1803 the macro \c{b} wasn't defined at the time of definition of \c{a}.
1805 Macros defined with \c{%define} are \i{case sensitive}: after
1806 \c{%define foo bar}, only \c{foo} will expand to \c{bar}: \c{Foo} or
1807 \c{FOO} will not. By using \c{%idefine} instead of \c{%define} (the
1808 `i' stands for `insensitive') you can define all the case variants
1809 of a macro at once, so that \c{%idefine foo bar} would cause
1810 \c{foo}, \c{Foo}, \c{FOO}, \c{fOO} and so on all to expand to
1813 There is a mechanism which detects when a macro call has occurred as
1814 a result of a previous expansion of the same macro, to guard against
1815 \i{circular references} and infinite loops. If this happens, the
1816 preprocessor will only expand the first occurrence of the macro.
1819 \c %define a(x) 1+a(x)
1823 the macro \c{a(3)} will expand once, becoming \c{1+a(3)}, and will
1824 then expand no further. This behaviour can be useful: see \k{32c}
1825 for an example of its use.
1827 You can \I{overloading, single-line macros}overload single-line
1828 macros: if you write
1830 \c %define foo(x) 1+x
1831 \c %define foo(x,y) 1+x*y
1833 the preprocessor will be able to handle both types of macro call,
1834 by counting the parameters you pass; so \c{foo(3)} will become
1835 \c{1+3} whereas \c{foo(ebx,2)} will become \c{1+ebx*2}. However, if
1840 then no other definition of \c{foo} will be accepted: a macro with
1841 no parameters prohibits the definition of the same name as a macro
1842 \e{with} parameters, and vice versa.
1844 This doesn't prevent single-line macros being \e{redefined}: you can
1845 perfectly well define a macro with
1849 and then re-define it later in the same source file with
1853 Then everywhere the macro \c{foo} is invoked, it will be expanded
1854 according to the most recent definition. This is particularly useful
1855 when defining single-line macros with \c{%assign} (see \k{assign}).
1857 You can \i{pre-define} single-line macros using the `-d' option on
1858 the NASM command line: see \k{opt-d}.
1861 \S{xdefine} Enhancing %define: \I\c{%xidefine}\i\c{%xdefine}
1863 To have a reference to an embedded single-line macro resolved at the
1864 time that it is embedded, as opposed to when the calling macro is
1865 expanded, you need a different mechanism to the one offered by
1866 \c{%define}. The solution is to use \c{%xdefine}, or it's
1867 \I{case sensitive}case-insensitive counterpart \c{%xidefine}.
1869 Suppose you have the following code:
1872 \c %define isFalse isTrue
1881 In this case, \c{val1} is equal to 0, and \c{val2} is equal to 1.
1882 This is because, when a single-line macro is defined using
1883 \c{%define}, it is expanded only when it is called. As \c{isFalse}
1884 expands to \c{isTrue}, the expansion will be the current value of
1885 \c{isTrue}. The first time it is called that is 0, and the second
1888 If you wanted \c{isFalse} to expand to the value assigned to the
1889 embedded macro \c{isTrue} at the time that \c{isFalse} was defined,
1890 you need to change the above code to use \c{%xdefine}.
1892 \c %xdefine isTrue 1
1893 \c %xdefine isFalse isTrue
1894 \c %xdefine isTrue 0
1898 \c %xdefine isTrue 1
1902 Now, each time that \c{isFalse} is called, it expands to 1,
1903 as that is what the embedded macro \c{isTrue} expanded to at
1904 the time that \c{isFalse} was defined.
1907 \S{concat%+} Concatenating Single Line Macro Tokens: \i\c{%+}
1909 Individual tokens in single line macros can be concatenated, to produce
1910 longer tokens for later processing. This can be useful if there are
1911 several similar macros that perform similar functions.
1913 Please note that a space is required after \c{%+}, in order to
1914 disambiguate it from the syntax \c{%+1} used in multiline macros.
1916 As an example, consider the following:
1918 \c %define BDASTART 400h ; Start of BIOS data area
1920 \c struc tBIOSDA ; its structure
1926 Now, if we need to access the elements of tBIOSDA in different places,
1929 \c mov ax,BDASTART + tBIOSDA.COM1addr
1930 \c mov bx,BDASTART + tBIOSDA.COM2addr
1932 This will become pretty ugly (and tedious) if used in many places, and
1933 can be reduced in size significantly by using the following macro:
1935 \c ; Macro to access BIOS variables by their names (from tBDA):
1937 \c %define BDA(x) BDASTART + tBIOSDA. %+ x
1939 Now the above code can be written as:
1941 \c mov ax,BDA(COM1addr)
1942 \c mov bx,BDA(COM2addr)
1944 Using this feature, we can simplify references to a lot of macros (and,
1945 in turn, reduce typing errors).
1948 \S{undef} Undefining macros: \i\c{%undef}
1950 Single-line macros can be removed with the \c{%undef} command. For
1951 example, the following sequence:
1958 will expand to the instruction \c{mov eax, foo}, since after
1959 \c{%undef} the macro \c{foo} is no longer defined.
1961 Macros that would otherwise be pre-defined can be undefined on the
1962 command-line using the `-u' option on the NASM command line: see
1966 \S{assign} \i{Preprocessor Variables}: \i\c{%assign}
1968 An alternative way to define single-line macros is by means of the
1969 \c{%assign} command (and its \I{case sensitive}case-insensitive
1970 counterpart \i\c{%iassign}, which differs from \c{%assign} in
1971 exactly the same way that \c{%idefine} differs from \c{%define}).
1973 \c{%assign} is used to define single-line macros which take no
1974 parameters and have a numeric value. This value can be specified in
1975 the form of an expression, and it will be evaluated once, when the
1976 \c{%assign} directive is processed.
1978 Like \c{%define}, macros defined using \c{%assign} can be re-defined
1979 later, so you can do things like
1983 to increment the numeric value of a macro.
1985 \c{%assign} is useful for controlling the termination of \c{%rep}
1986 preprocessor loops: see \k{rep} for an example of this. Another
1987 use for \c{%assign} is given in \k{16c} and \k{32c}.
1989 The expression passed to \c{%assign} is a \i{critical expression}
1990 (see \k{crit}), and must also evaluate to a pure number (rather than
1991 a relocatable reference such as a code or data address, or anything
1992 involving a register).
1995 \H{strlen} \i{String Handling in Macros}: \i\c{%strlen} and \i\c{%substr}
1997 It's often useful to be able to handle strings in macros. NASM
1998 supports two simple string handling macro operators from which
1999 more complex operations can be constructed.
2002 \S{strlen} \i{String Length}: \i\c{%strlen}
2004 The \c{%strlen} macro is like \c{%assign} macro in that it creates
2005 (or redefines) a numeric value to a macro. The difference is that
2006 with \c{%strlen}, the numeric value is the length of a string. An
2007 example of the use of this would be:
2009 \c %strlen charcnt 'my string'
2011 In this example, \c{charcnt} would receive the value 8, just as
2012 if an \c{%assign} had been used. In this example, \c{'my string'}
2013 was a literal string but it could also have been a single-line
2014 macro that expands to a string, as in the following example:
2016 \c %define sometext 'my string'
2017 \c %strlen charcnt sometext
2019 As in the first case, this would result in \c{charcnt} being
2020 assigned the value of 8.
2023 \S{substr} \i{Sub-strings}: \i\c{%substr}
2025 Individual letters in strings can be extracted using \c{%substr}.
2026 An example of its use is probably more useful than the description:
2028 \c %substr mychar 'xyz' 1 ; equivalent to %define mychar 'x'
2029 \c %substr mychar 'xyz' 2 ; equivalent to %define mychar 'y'
2030 \c %substr mychar 'xyz' 3 ; equivalent to %define mychar 'z'
2032 In this example, mychar gets the value of 'y'. As with \c{%strlen}
2033 (see \k{strlen}), the first parameter is the single-line macro to
2034 be created and the second is the string. The third parameter
2035 specifies which character is to be selected. Note that the first
2036 index is 1, not 0 and the last index is equal to the value that
2037 \c{%strlen} would assign given the same string. Index values out
2038 of range result in an empty string.
2041 \H{mlmacro} \i{Multi-Line Macros}: \I\c{%imacro}\i\c{%macro}
2043 Multi-line macros are much more like the type of macro seen in MASM
2044 and TASM: a multi-line macro definition in NASM looks something like
2047 \c %macro prologue 1
2055 This defines a C-like function prologue as a macro: so you would
2056 invoke the macro with a call such as
2058 \c myfunc: prologue 12
2060 which would expand to the three lines of code
2066 The number \c{1} after the macro name in the \c{%macro} line defines
2067 the number of parameters the macro \c{prologue} expects to receive.
2068 The use of \c{%1} inside the macro definition refers to the first
2069 parameter to the macro call. With a macro taking more than one
2070 parameter, subsequent parameters would be referred to as \c{%2},
2073 Multi-line macros, like single-line macros, are \i{case-sensitive},
2074 unless you define them using the alternative directive \c{%imacro}.
2076 If you need to pass a comma as \e{part} of a parameter to a
2077 multi-line macro, you can do that by enclosing the entire parameter
2078 in \I{braces, around macro parameters}braces. So you could code
2087 \c silly 'a', letter_a ; letter_a: db 'a'
2088 \c silly 'ab', string_ab ; string_ab: db 'ab'
2089 \c silly {13,10}, crlf ; crlf: db 13,10
2092 \S{mlmacover} Overloading Multi-Line Macros\I{overloading, multi-line macros}
2094 As with single-line macros, multi-line macros can be overloaded by
2095 defining the same macro name several times with different numbers of
2096 parameters. This time, no exception is made for macros with no
2097 parameters at all. So you could define
2099 \c %macro prologue 0
2106 to define an alternative form of the function prologue which
2107 allocates no local stack space.
2109 Sometimes, however, you might want to `overload' a machine
2110 instruction; for example, you might want to define
2119 so that you could code
2121 \c push ebx ; this line is not a macro call
2122 \c push eax,ecx ; but this one is
2124 Ordinarily, NASM will give a warning for the first of the above two
2125 lines, since \c{push} is now defined to be a macro, and is being
2126 invoked with a number of parameters for which no definition has been
2127 given. The correct code will still be generated, but the assembler
2128 will give a warning. This warning can be disabled by the use of the
2129 \c{-w-macro-params} command-line option (see \k{opt-w}).
2132 \S{maclocal} \i{Macro-Local Labels}
2134 NASM allows you to define labels within a multi-line macro
2135 definition in such a way as to make them local to the macro call: so
2136 calling the same macro multiple times will use a different label
2137 each time. You do this by prefixing \i\c{%%} to the label name. So
2138 you can invent an instruction which executes a \c{RET} if the \c{Z}
2139 flag is set by doing this:
2149 You can call this macro as many times as you want, and every time
2150 you call it NASM will make up a different `real' name to substitute
2151 for the label \c{%%skip}. The names NASM invents are of the form
2152 \c{..@2345.skip}, where the number 2345 changes with every macro
2153 call. The \i\c{..@} prefix prevents macro-local labels from
2154 interfering with the local label mechanism, as described in
2155 \k{locallab}. You should avoid defining your own labels in this form
2156 (the \c{..@} prefix, then a number, then another period) in case
2157 they interfere with macro-local labels.
2160 \S{mlmacgre} \i{Greedy Macro Parameters}
2162 Occasionally it is useful to define a macro which lumps its entire
2163 command line into one parameter definition, possibly after
2164 extracting one or two smaller parameters from the front. An example
2165 might be a macro to write a text string to a file in MS-DOS, where
2166 you might want to be able to write
2168 \c writefile [filehandle],"hello, world",13,10
2170 NASM allows you to define the last parameter of a macro to be
2171 \e{greedy}, meaning that if you invoke the macro with more
2172 parameters than it expects, all the spare parameters get lumped into
2173 the last defined one along with the separating commas. So if you
2176 \c %macro writefile 2+
2182 \c mov cx,%%endstr-%%str
2189 then the example call to \c{writefile} above will work as expected:
2190 the text before the first comma, \c{[filehandle]}, is used as the
2191 first macro parameter and expanded when \c{%1} is referred to, and
2192 all the subsequent text is lumped into \c{%2} and placed after the
2195 The greedy nature of the macro is indicated to NASM by the use of
2196 the \I{+ modifier}\c{+} sign after the parameter count on the
2199 If you define a greedy macro, you are effectively telling NASM how
2200 it should expand the macro given \e{any} number of parameters from
2201 the actual number specified up to infinity; in this case, for
2202 example, NASM now knows what to do when it sees a call to
2203 \c{writefile} with 2, 3, 4 or more parameters. NASM will take this
2204 into account when overloading macros, and will not allow you to
2205 define another form of \c{writefile} taking 4 parameters (for
2208 Of course, the above macro could have been implemented as a
2209 non-greedy macro, in which case the call to it would have had to
2212 \c writefile [filehandle], {"hello, world",13,10}
2214 NASM provides both mechanisms for putting \i{commas in macro
2215 parameters}, and you choose which one you prefer for each macro
2218 See \k{sectmac} for a better way to write the above macro.
2221 \S{mlmacdef} \i{Default Macro Parameters}
2223 NASM also allows you to define a multi-line macro with a \e{range}
2224 of allowable parameter counts. If you do this, you can specify
2225 defaults for \i{omitted parameters}. So, for example:
2227 \c %macro die 0-1 "Painful program death has occurred."
2235 This macro (which makes use of the \c{writefile} macro defined in
2236 \k{mlmacgre}) can be called with an explicit error message, which it
2237 will display on the error output stream before exiting, or it can be
2238 called with no parameters, in which case it will use the default
2239 error message supplied in the macro definition.
2241 In general, you supply a minimum and maximum number of parameters
2242 for a macro of this type; the minimum number of parameters are then
2243 required in the macro call, and then you provide defaults for the
2244 optional ones. So if a macro definition began with the line
2246 \c %macro foobar 1-3 eax,[ebx+2]
2248 then it could be called with between one and three parameters, and
2249 \c{%1} would always be taken from the macro call. \c{%2}, if not
2250 specified by the macro call, would default to \c{eax}, and \c{%3} if
2251 not specified would default to \c{[ebx+2]}.
2253 You may omit parameter defaults from the macro definition, in which
2254 case the parameter default is taken to be blank. This can be useful
2255 for macros which can take a variable number of parameters, since the
2256 \i\c{%0} token (see \k{percent0}) allows you to determine how many
2257 parameters were really passed to the macro call.
2259 This defaulting mechanism can be combined with the greedy-parameter
2260 mechanism; so the \c{die} macro above could be made more powerful,
2261 and more useful, by changing the first line of the definition to
2263 \c %macro die 0-1+ "Painful program death has occurred.",13,10
2265 The maximum parameter count can be infinite, denoted by \c{*}. In
2266 this case, of course, it is impossible to provide a \e{full} set of
2267 default parameters. Examples of this usage are shown in \k{rotate}.
2270 \S{percent0} \i\c{%0}: \I{counting macro parameters}Macro Parameter Counter
2272 For a macro which can take a variable number of parameters, the
2273 parameter reference \c{%0} will return a numeric constant giving the
2274 number of parameters passed to the macro. This can be used as an
2275 argument to \c{%rep} (see \k{rep}) in order to iterate through all
2276 the parameters of a macro. Examples are given in \k{rotate}.
2279 \S{rotate} \i\c{%rotate}: \i{Rotating Macro Parameters}
2281 Unix shell programmers will be familiar with the \I{shift
2282 command}\c{shift} shell command, which allows the arguments passed
2283 to a shell script (referenced as \c{$1}, \c{$2} and so on) to be
2284 moved left by one place, so that the argument previously referenced
2285 as \c{$2} becomes available as \c{$1}, and the argument previously
2286 referenced as \c{$1} is no longer available at all.
2288 NASM provides a similar mechanism, in the form of \c{%rotate}. As
2289 its name suggests, it differs from the Unix \c{shift} in that no
2290 parameters are lost: parameters rotated off the left end of the
2291 argument list reappear on the right, and vice versa.
2293 \c{%rotate} is invoked with a single numeric argument (which may be
2294 an expression). The macro parameters are rotated to the left by that
2295 many places. If the argument to \c{%rotate} is negative, the macro
2296 parameters are rotated to the right.
2298 \I{iterating over macro parameters}So a pair of macros to save and
2299 restore a set of registers might work as follows:
2301 \c %macro multipush 1-*
2310 This macro invokes the \c{PUSH} instruction on each of its arguments
2311 in turn, from left to right. It begins by pushing its first
2312 argument, \c{%1}, then invokes \c{%rotate} to move all the arguments
2313 one place to the left, so that the original second argument is now
2314 available as \c{%1}. Repeating this procedure as many times as there
2315 were arguments (achieved by supplying \c{%0} as the argument to
2316 \c{%rep}) causes each argument in turn to be pushed.
2318 Note also the use of \c{*} as the maximum parameter count,
2319 indicating that there is no upper limit on the number of parameters
2320 you may supply to the \i\c{multipush} macro.
2322 It would be convenient, when using this macro, to have a \c{POP}
2323 equivalent, which \e{didn't} require the arguments to be given in
2324 reverse order. Ideally, you would write the \c{multipush} macro
2325 call, then cut-and-paste the line to where the pop needed to be
2326 done, and change the name of the called macro to \c{multipop}, and
2327 the macro would take care of popping the registers in the opposite
2328 order from the one in which they were pushed.
2330 This can be done by the following definition:
2332 \c %macro multipop 1-*
2341 This macro begins by rotating its arguments one place to the
2342 \e{right}, so that the original \e{last} argument appears as \c{%1}.
2343 This is then popped, and the arguments are rotated right again, so
2344 the second-to-last argument becomes \c{%1}. Thus the arguments are
2345 iterated through in reverse order.
2348 \S{concat} \i{Concatenating Macro Parameters}
2350 NASM can concatenate macro parameters on to other text surrounding
2351 them. This allows you to declare a family of symbols, for example,
2352 in a macro definition. If, for example, you wanted to generate a
2353 table of key codes along with offsets into the table, you could code
2356 \c %macro keytab_entry 2
2358 \c keypos%1 equ $-keytab
2364 \c keytab_entry F1,128+1
2365 \c keytab_entry F2,128+2
2366 \c keytab_entry Return,13
2368 which would expand to
2371 \c keyposF1 equ $-keytab
2373 \c keyposF2 equ $-keytab
2375 \c keyposReturn equ $-keytab
2378 You can just as easily concatenate text on to the other end of a
2379 macro parameter, by writing \c{%1foo}.
2381 If you need to append a \e{digit} to a macro parameter, for example
2382 defining labels \c{foo1} and \c{foo2} when passed the parameter
2383 \c{foo}, you can't code \c{%11} because that would be taken as the
2384 eleventh macro parameter. Instead, you must code
2385 \I{braces, after % sign}\c{%\{1\}1}, which will separate the first
2386 \c{1} (giving the number of the macro parameter) from the second
2387 (literal text to be concatenated to the parameter).
2389 This concatenation can also be applied to other preprocessor in-line
2390 objects, such as macro-local labels (\k{maclocal}) and context-local
2391 labels (\k{ctxlocal}). In all cases, ambiguities in syntax can be
2392 resolved by enclosing everything after the \c{%} sign and before the
2393 literal text in braces: so \c{%\{%foo\}bar} concatenates the text
2394 \c{bar} to the end of the real name of the macro-local label
2395 \c{%%foo}. (This is unnecessary, since the form NASM uses for the
2396 real names of macro-local labels means that the two usages
2397 \c{%\{%foo\}bar} and \c{%%foobar} would both expand to the same
2398 thing anyway; nevertheless, the capability is there.)
2401 \S{mlmaccc} \i{Condition Codes as Macro Parameters}
2403 NASM can give special treatment to a macro parameter which contains
2404 a condition code. For a start, you can refer to the macro parameter
2405 \c{%1} by means of the alternative syntax \i\c{%+1}, which informs
2406 NASM that this macro parameter is supposed to contain a condition
2407 code, and will cause the preprocessor to report an error message if
2408 the macro is called with a parameter which is \e{not} a valid
2411 Far more usefully, though, you can refer to the macro parameter by
2412 means of \i\c{%-1}, which NASM will expand as the \e{inverse}
2413 condition code. So the \c{retz} macro defined in \k{maclocal} can be
2414 replaced by a general \i{conditional-return macro} like this:
2424 This macro can now be invoked using calls like \c{retc ne}, which
2425 will cause the conditional-jump instruction in the macro expansion
2426 to come out as \c{JE}, or \c{retc po} which will make the jump a
2429 The \c{%+1} macro-parameter reference is quite happy to interpret
2430 the arguments \c{CXZ} and \c{ECXZ} as valid condition codes;
2431 however, \c{%-1} will report an error if passed either of these,
2432 because no inverse condition code exists.
2435 \S{nolist} \i{Disabling Listing Expansion}\I\c{.nolist}
2437 When NASM is generating a listing file from your program, it will
2438 generally expand multi-line macros by means of writing the macro
2439 call and then listing each line of the expansion. This allows you to
2440 see which instructions in the macro expansion are generating what
2441 code; however, for some macros this clutters the listing up
2444 NASM therefore provides the \c{.nolist} qualifier, which you can
2445 include in a macro definition to inhibit the expansion of the macro
2446 in the listing file. The \c{.nolist} qualifier comes directly after
2447 the number of parameters, like this:
2449 \c %macro foo 1.nolist
2453 \c %macro bar 1-5+.nolist a,b,c,d,e,f,g,h
2455 \H{condasm} \i{Conditional Assembly}\I\c{%if}
2457 Similarly to the C preprocessor, NASM allows sections of a source
2458 file to be assembled only if certain conditions are met. The general
2459 syntax of this feature looks like this:
2462 \c ; some code which only appears if <condition> is met
2463 \c %elif<condition2>
2464 \c ; only appears if <condition> is not met but <condition2> is
2466 \c ; this appears if neither <condition> nor <condition2> was met
2469 The \i\c{%else} clause is optional, as is the \i\c{%elif} clause.
2470 You can have more than one \c{%elif} clause as well.
2473 \S{ifdef} \i\c{%ifdef}: Testing Single-Line Macro Existence\I{testing,
2474 single-line macro existence}
2476 Beginning a conditional-assembly block with the line \c{%ifdef
2477 MACRO} will assemble the subsequent code if, and only if, a
2478 single-line macro called \c{MACRO} is defined. If not, then the
2479 \c{%elif} and \c{%else} blocks (if any) will be processed instead.
2481 For example, when debugging a program, you might want to write code
2484 \c ; perform some function
2486 \c writefile 2,"Function performed successfully",13,10
2488 \c ; go and do something else
2490 Then you could use the command-line option \c{-dDEBUG} to create a
2491 version of the program which produced debugging messages, and remove
2492 the option to generate the final release version of the program.
2494 You can test for a macro \e{not} being defined by using
2495 \i\c{%ifndef} instead of \c{%ifdef}. You can also test for macro
2496 definitions in \c{%elif} blocks by using \i\c{%elifdef} and
2500 \S{ifmacro} \i\c{ifmacro}: Testing Multi-Line Macro
2501 Existence\I{testing, multi-line macro existence}
2503 The \c{%ifmacro} directive operates in the same way as the \c{%ifdef}
2504 directive, except that it checks for the existence of a multi-line macro.
2506 For example, you may be working with a large project and not have control
2507 over the macros in a library. You may want to create a macro with one
2508 name if it doesn't already exist, and another name if one with that name
2511 The \c{%ifmacro} is considered true if defining a macro with the given name
2512 and number of arguments would cause a definitions conflict. For example:
2514 \c %ifmacro MyMacro 1-3
2516 \c %error "MyMacro 1-3" causes a conflict with an existing macro.
2520 \c %macro MyMacro 1-3
2522 \c ; insert code to define the macro
2528 This will create the macro "MyMacro 1-3" if no macro already exists which
2529 would conflict with it, and emits a warning if there would be a definition
2532 You can test for the macro not existing by using the \i\c{%ifnmacro} instead
2533 of \c{%ifmacro}. Additional tests can be performed in \c{%elif} blocks by using
2534 \i\c{%elifmacro} and \i\c{%elifnmacro}.
2537 \S{ifctx} \i\c{%ifctx}: Testing the Context Stack\I{testing, context
2540 The conditional-assembly construct \c{%ifctx ctxname} will cause the
2541 subsequent code to be assembled if and only if the top context on
2542 the preprocessor's context stack has the name \c{ctxname}. As with
2543 \c{%ifdef}, the inverse and \c{%elif} forms \i\c{%ifnctx},
2544 \i\c{%elifctx} and \i\c{%elifnctx} are also supported.
2546 For more details of the context stack, see \k{ctxstack}. For a
2547 sample use of \c{%ifctx}, see \k{blockif}.
2550 \S{if} \i\c{%if}: Testing Arbitrary Numeric Expressions\I{testing,
2551 arbitrary numeric expressions}
2553 The conditional-assembly construct \c{%if expr} will cause the
2554 subsequent code to be assembled if and only if the value of the
2555 numeric expression \c{expr} is non-zero. An example of the use of
2556 this feature is in deciding when to break out of a \c{%rep}
2557 preprocessor loop: see \k{rep} for a detailed example.
2559 The expression given to \c{%if}, and its counterpart \i\c{%elif}, is
2560 a critical expression (see \k{crit}).
2562 \c{%if} extends the normal NASM expression syntax, by providing a
2563 set of \i{relational operators} which are not normally available in
2564 expressions. The operators \i\c{=}, \i\c{<}, \i\c{>}, \i\c{<=},
2565 \i\c{>=} and \i\c{<>} test equality, less-than, greater-than,
2566 less-or-equal, greater-or-equal and not-equal respectively. The
2567 C-like forms \i\c{==} and \i\c{!=} are supported as alternative
2568 forms of \c{=} and \c{<>}. In addition, low-priority logical
2569 operators \i\c{&&}, \i\c{^^} and \i\c{||} are provided, supplying
2570 \i{logical AND}, \i{logical XOR} and \i{logical OR}. These work like
2571 the C logical operators (although C has no logical XOR), in that
2572 they always return either 0 or 1, and treat any non-zero input as 1
2573 (so that \c{^^}, for example, returns 1 if exactly one of its inputs
2574 is zero, and 0 otherwise). The relational operators also return 1
2575 for true and 0 for false.
2577 Like most other \c{%if} constructs, \c{%if} has a counterpart
2578 \i\c{%elif}, and negative forms \i\c{%ifn} and \i\c{%elifn}.
2580 \S{ifidn} \i\c{%ifidn} and \i\c{%ifidni}: Testing Exact Text
2581 Identity\I{testing, exact text identity}
2583 The construct \c{%ifidn text1,text2} will cause the subsequent code
2584 to be assembled if and only if \c{text1} and \c{text2}, after
2585 expanding single-line macros, are identical pieces of text.
2586 Differences in white space are not counted.
2588 \c{%ifidni} is similar to \c{%ifidn}, but is \i{case-insensitive}.
2590 For example, the following macro pushes a register or number on the
2591 stack, and allows you to treat \c{IP} as a real register:
2593 \c %macro pushparam 1
2604 Like most other \c{%if} constructs, \c{%ifidn} has a counterpart
2605 \i\c{%elifidn}, and negative forms \i\c{%ifnidn} and \i\c{%elifnidn}.
2606 Similarly, \c{%ifidni} has counterparts \i\c{%elifidni},
2607 \i\c{%ifnidni} and \i\c{%elifnidni}.
2610 \S{iftyp} \i\c{%ifid}, \i\c{%ifnum}, \i\c{%ifstr}: Testing Token
2611 Types\I{testing, token types}
2613 Some macros will want to perform different tasks depending on
2614 whether they are passed a number, a string, or an identifier. For
2615 example, a string output macro might want to be able to cope with
2616 being passed either a string constant or a pointer to an existing
2619 The conditional assembly construct \c{%ifid}, taking one parameter
2620 (which may be blank), assembles the subsequent code if and only if
2621 the first token in the parameter exists and is an identifier.
2622 \c{%ifnum} works similarly, but tests for the token being a numeric
2623 constant; \c{%ifstr} tests for it being a string.
2625 For example, the \c{writefile} macro defined in \k{mlmacgre} can be
2626 extended to take advantage of \c{%ifstr} in the following fashion:
2628 \c %macro writefile 2-3+
2637 \c %%endstr: mov dx,%%str
2638 \c mov cx,%%endstr-%%str
2649 Then the \c{writefile} macro can cope with being called in either of
2650 the following two ways:
2652 \c writefile [file], strpointer, length
2653 \c writefile [file], "hello", 13, 10
2655 In the first, \c{strpointer} is used as the address of an
2656 already-declared string, and \c{length} is used as its length; in
2657 the second, a string is given to the macro, which therefore declares
2658 it itself and works out the address and length for itself.
2660 Note the use of \c{%if} inside the \c{%ifstr}: this is to detect
2661 whether the macro was passed two arguments (so the string would be a
2662 single string constant, and \c{db %2} would be adequate) or more (in
2663 which case, all but the first two would be lumped together into
2664 \c{%3}, and \c{db %2,%3} would be required).
2666 \I\c{%ifnid}\I\c{%elifid}\I\c{%elifnid}\I\c{%ifnnum}\I\c{%elifnum}
2667 \I\c{%elifnnum}\I\c{%ifnstr}\I\c{%elifstr}\I\c{%elifnstr}
2668 The usual \c{%elifXXX}, \c{%ifnXXX} and \c{%elifnXXX} versions exist
2669 for each of \c{%ifid}, \c{%ifnum} and \c{%ifstr}.
2672 \S{pperror} \i\c{%error}: Reporting \i{User-Defined Errors}
2674 The preprocessor directive \c{%error} will cause NASM to report an
2675 error if it occurs in assembled code. So if other users are going to
2676 try to assemble your source files, you can ensure that they define
2677 the right macros by means of code like this:
2679 \c %ifdef SOME_MACRO
2681 \c %elifdef SOME_OTHER_MACRO
2682 \c ; do some different setup
2684 \c %error Neither SOME_MACRO nor SOME_OTHER_MACRO was defined.
2687 Then any user who fails to understand the way your code is supposed
2688 to be assembled will be quickly warned of their mistake, rather than
2689 having to wait until the program crashes on being run and then not
2690 knowing what went wrong.
2693 \H{rep} \i{Preprocessor Loops}\I{repeating code}: \i\c{%rep}
2695 NASM's \c{TIMES} prefix, though useful, cannot be used to invoke a
2696 multi-line macro multiple times, because it is processed by NASM
2697 after macros have already been expanded. Therefore NASM provides
2698 another form of loop, this time at the preprocessor level: \c{%rep}.
2700 The directives \c{%rep} and \i\c{%endrep} (\c{%rep} takes a numeric
2701 argument, which can be an expression; \c{%endrep} takes no
2702 arguments) can be used to enclose a chunk of code, which is then
2703 replicated as many times as specified by the preprocessor:
2707 \c inc word [table+2*i]
2711 This will generate a sequence of 64 \c{INC} instructions,
2712 incrementing every word of memory from \c{[table]} to
2715 For more complex termination conditions, or to break out of a repeat
2716 loop part way along, you can use the \i\c{%exitrep} directive to
2717 terminate the loop, like this:
2732 \c fib_number equ ($-fibonacci)/2
2734 This produces a list of all the Fibonacci numbers that will fit in
2735 16 bits. Note that a maximum repeat count must still be given to
2736 \c{%rep}. This is to prevent the possibility of NASM getting into an
2737 infinite loop in the preprocessor, which (on multitasking or
2738 multi-user systems) would typically cause all the system memory to
2739 be gradually used up and other applications to start crashing.
2742 \H{include} \i{Including Other Files}
2744 Using, once again, a very similar syntax to the C preprocessor,
2745 NASM's preprocessor lets you include other source files into your
2746 code. This is done by the use of the \i\c{%include} directive:
2748 \c %include "macros.mac"
2750 will include the contents of the file \c{macros.mac} into the source
2751 file containing the \c{%include} directive.
2753 Include files are \I{searching for include files}searched for in the
2754 current directory (the directory you're in when you run NASM, as
2755 opposed to the location of the NASM executable or the location of
2756 the source file), plus any directories specified on the NASM command
2757 line using the \c{-i} option.
2759 The standard C idiom for preventing a file being included more than
2760 once is just as applicable in NASM: if the file \c{macros.mac} has
2763 \c %ifndef MACROS_MAC
2764 \c %define MACROS_MAC
2765 \c ; now define some macros
2768 then including the file more than once will not cause errors,
2769 because the second time the file is included nothing will happen
2770 because the macro \c{MACROS_MAC} will already be defined.
2772 You can force a file to be included even if there is no \c{%include}
2773 directive that explicitly includes it, by using the \i\c{-p} option
2774 on the NASM command line (see \k{opt-p}).
2777 \H{ctxstack} The \i{Context Stack}
2779 Having labels that are local to a macro definition is sometimes not
2780 quite powerful enough: sometimes you want to be able to share labels
2781 between several macro calls. An example might be a \c{REPEAT} ...
2782 \c{UNTIL} loop, in which the expansion of the \c{REPEAT} macro
2783 would need to be able to refer to a label which the \c{UNTIL} macro
2784 had defined. However, for such a macro you would also want to be
2785 able to nest these loops.
2787 NASM provides this level of power by means of a \e{context stack}.
2788 The preprocessor maintains a stack of \e{contexts}, each of which is
2789 characterized by a name. You add a new context to the stack using
2790 the \i\c{%push} directive, and remove one using \i\c{%pop}. You can
2791 define labels that are local to a particular context on the stack.
2794 \S{pushpop} \i\c{%push} and \i\c{%pop}: \I{creating
2795 contexts}\I{removing contexts}Creating and Removing Contexts
2797 The \c{%push} directive is used to create a new context and place it
2798 on the top of the context stack. \c{%push} requires one argument,
2799 which is the name of the context. For example:
2803 This pushes a new context called \c{foobar} on the stack. You can
2804 have several contexts on the stack with the same name: they can
2805 still be distinguished.
2807 The directive \c{%pop}, requiring no arguments, removes the top
2808 context from the context stack and destroys it, along with any
2809 labels associated with it.
2812 \S{ctxlocal} \i{Context-Local Labels}
2814 Just as the usage \c{%%foo} defines a label which is local to the
2815 particular macro call in which it is used, the usage \I{%$}\c{%$foo}
2816 is used to define a label which is local to the context on the top
2817 of the context stack. So the \c{REPEAT} and \c{UNTIL} example given
2818 above could be implemented by means of:
2834 and invoked by means of, for example,
2842 which would scan every fourth byte of a string in search of the byte
2845 If you need to define, or access, labels local to the context
2846 \e{below} the top one on the stack, you can use \I{%$$}\c{%$$foo}, or
2847 \c{%$$$foo} for the context below that, and so on.
2850 \S{ctxdefine} \i{Context-Local Single-Line Macros}
2852 NASM also allows you to define single-line macros which are local to
2853 a particular context, in just the same way:
2855 \c %define %$localmac 3
2857 will define the single-line macro \c{%$localmac} to be local to the
2858 top context on the stack. Of course, after a subsequent \c{%push},
2859 it can then still be accessed by the name \c{%$$localmac}.
2862 \S{ctxrepl} \i\c{%repl}: \I{renaming contexts}Renaming a Context
2864 If you need to change the name of the top context on the stack (in
2865 order, for example, to have it respond differently to \c{%ifctx}),
2866 you can execute a \c{%pop} followed by a \c{%push}; but this will
2867 have the side effect of destroying all context-local labels and
2868 macros associated with the context that was just popped.
2870 NASM provides the directive \c{%repl}, which \e{replaces} a context
2871 with a different name, without touching the associated macros and
2872 labels. So you could replace the destructive code
2877 with the non-destructive version \c{%repl newname}.
2880 \S{blockif} Example Use of the \i{Context Stack}: \i{Block IFs}
2882 This example makes use of almost all the context-stack features,
2883 including the conditional-assembly construct \i\c{%ifctx}, to
2884 implement a block IF statement as a set of macros.
2900 \c %error "expected `if' before `else'"
2914 \c %error "expected `if' or `else' before `endif'"
2919 This code is more robust than the \c{REPEAT} and \c{UNTIL} macros
2920 given in \k{ctxlocal}, because it uses conditional assembly to check
2921 that the macros are issued in the right order (for example, not
2922 calling \c{endif} before \c{if}) and issues a \c{%error} if they're
2925 In addition, the \c{endif} macro has to be able to cope with the two
2926 distinct cases of either directly following an \c{if}, or following
2927 an \c{else}. It achieves this, again, by using conditional assembly
2928 to do different things depending on whether the context on top of
2929 the stack is \c{if} or \c{else}.
2931 The \c{else} macro has to preserve the context on the stack, in
2932 order to have the \c{%$ifnot} referred to by the \c{if} macro be the
2933 same as the one defined by the \c{endif} macro, but has to change
2934 the context's name so that \c{endif} will know there was an
2935 intervening \c{else}. It does this by the use of \c{%repl}.
2937 A sample usage of these macros might look like:
2959 The block-\c{IF} macros handle nesting quite happily, by means of
2960 pushing another context, describing the inner \c{if}, on top of the
2961 one describing the outer \c{if}; thus \c{else} and \c{endif} always
2962 refer to the last unmatched \c{if} or \c{else}.
2965 \H{stdmac} \i{Standard Macros}
2967 NASM defines a set of standard macros, which are already defined
2968 when it starts to process any source file. If you really need a
2969 program to be assembled with no pre-defined macros, you can use the
2970 \i\c{%clear} directive to empty the preprocessor of everything but
2971 context-local preprocessor variables and single-line macros.
2973 Most \i{user-level assembler directives} (see \k{directive}) are
2974 implemented as macros which invoke primitive directives; these are
2975 described in \k{directive}. The rest of the standard macro set is
2979 \S{stdmacver} \i\c{__NASM_MAJOR__}, \i\c{__NASM_MINOR__},
2980 \i\c{__NASM_SUBMINOR__} and \i\c{___NASM_PATCHLEVEL__}: \i{NASM Version}
2982 The single-line macros \c{__NASM_MAJOR__}, \c{__NASM_MINOR__},
2983 \c{__NASM_SUBMINOR__} and \c{___NASM_PATCHLEVEL__} expand to the
2984 major, minor, subminor and patch level parts of the \i{version
2985 number of NASM} being used. So, under NASM 0.98.32p1 for
2986 example, \c{__NASM_MAJOR__} would be defined to be 0, \c{__NASM_MINOR__}
2987 would be defined as 98, \c{__NASM_SUBMINOR__} would be defined to 32,
2988 and \c{___NASM_PATCHLEVEL__} would be defined as 1.
2991 \S{stdmacverid} \i\c{__NASM_VERSION_ID__}: \i{NASM Version ID}
2993 The single-line macro \c{__NASM_VERSION_ID__} expands to a dword integer
2994 representing the full version number of the version of nasm being used.
2995 The value is the equivalent to \c{__NASM_MAJOR__}, \c{__NASM_MINOR__},
2996 \c{__NASM_SUBMINOR__} and \c{___NASM_PATCHLEVEL__} concatenated to
2997 produce a single doubleword. Hence, for 0.98.32p1, the returned number
2998 would be equivalent to:
3006 Note that the above lines are generate exactly the same code, the second
3007 line is used just to give an indication of the order that the separate
3008 values will be present in memory.
3011 \S{stdmacverstr} \i\c{__NASM_VER__}: \i{NASM Version string}
3013 The single-line macro \c{__NASM_VER__} expands to a string which defines
3014 the version number of nasm being used. So, under NASM 0.98.32 for example,
3023 \S{fileline} \i\c{__FILE__} and \i\c{__LINE__}: File Name and Line Number
3025 Like the C preprocessor, NASM allows the user to find out the file
3026 name and line number containing the current instruction. The macro
3027 \c{__FILE__} expands to a string constant giving the name of the
3028 current input file (which may change through the course of assembly
3029 if \c{%include} directives are used), and \c{__LINE__} expands to a
3030 numeric constant giving the current line number in the input file.
3032 These macros could be used, for example, to communicate debugging
3033 information to a macro, since invoking \c{__LINE__} inside a macro
3034 definition (either single-line or multi-line) will return the line
3035 number of the macro \e{call}, rather than \e{definition}. So to
3036 determine where in a piece of code a crash is occurring, for
3037 example, one could write a routine \c{stillhere}, which is passed a
3038 line number in \c{EAX} and outputs something like `line 155: still
3039 here'. You could then write a macro
3041 \c %macro notdeadyet 0
3050 and then pepper your code with calls to \c{notdeadyet} until you
3051 find the crash point.
3053 \S{bitsm} \i\c{__BITS__}: Current BITS Mode
3055 The \c{__BITS__} standard macro is updated every time that the BITS mode is
3056 set using the \c{BITS XX} or \c{[BITS XX]} directive, where XX is a valid mode
3057 number of 16, 32 or 64. \c{__BITS__} receives the specified mode number and
3058 makes it globally available. This can be very useful for those who utilize
3059 mode-dependent macros.
3062 \S{struc} \i\c{STRUC} and \i\c{ENDSTRUC}: \i{Declaring Structure} Data Types
3064 The core of NASM contains no intrinsic means of defining data
3065 structures; instead, the preprocessor is sufficiently powerful that
3066 data structures can be implemented as a set of macros. The macros
3067 \c{STRUC} and \c{ENDSTRUC} are used to define a structure data type.
3069 \c{STRUC} takes one parameter, which is the name of the data type.
3070 This name is defined as a symbol with the value zero, and also has
3071 the suffix \c{_size} appended to it and is then defined as an
3072 \c{EQU} giving the size of the structure. Once \c{STRUC} has been
3073 issued, you are defining the structure, and should define fields
3074 using the \c{RESB} family of pseudo-instructions, and then invoke
3075 \c{ENDSTRUC} to finish the definition.
3077 For example, to define a structure called \c{mytype} containing a
3078 longword, a word, a byte and a string of bytes, you might code
3089 The above code defines six symbols: \c{mt_long} as 0 (the offset
3090 from the beginning of a \c{mytype} structure to the longword field),
3091 \c{mt_word} as 4, \c{mt_byte} as 6, \c{mt_str} as 7, \c{mytype_size}
3092 as 39, and \c{mytype} itself as zero.
3094 The reason why the structure type name is defined at zero is a side
3095 effect of allowing structures to work with the local label
3096 mechanism: if your structure members tend to have the same names in
3097 more than one structure, you can define the above structure like this:
3108 This defines the offsets to the structure fields as \c{mytype.long},
3109 \c{mytype.word}, \c{mytype.byte} and \c{mytype.str}.
3111 NASM, since it has no \e{intrinsic} structure support, does not
3112 support any form of period notation to refer to the elements of a
3113 structure once you have one (except the above local-label notation),
3114 so code such as \c{mov ax,[mystruc.mt_word]} is not valid.
3115 \c{mt_word} is a constant just like any other constant, so the
3116 correct syntax is \c{mov ax,[mystruc+mt_word]} or \c{mov
3117 ax,[mystruc+mytype.word]}.
3120 \S{istruc} \i\c{ISTRUC}, \i\c{AT} and \i\c{IEND}: Declaring
3121 \i{Instances of Structures}
3123 Having defined a structure type, the next thing you typically want
3124 to do is to declare instances of that structure in your data
3125 segment. NASM provides an easy way to do this in the \c{ISTRUC}
3126 mechanism. To declare a structure of type \c{mytype} in a program,
3127 you code something like this:
3132 \c at mt_long, dd 123456
3133 \c at mt_word, dw 1024
3134 \c at mt_byte, db 'x'
3135 \c at mt_str, db 'hello, world', 13, 10, 0
3139 The function of the \c{AT} macro is to make use of the \c{TIMES}
3140 prefix to advance the assembly position to the correct point for the
3141 specified structure field, and then to declare the specified data.
3142 Therefore the structure fields must be declared in the same order as
3143 they were specified in the structure definition.
3145 If the data to go in a structure field requires more than one source
3146 line to specify, the remaining source lines can easily come after
3147 the \c{AT} line. For example:
3149 \c at mt_str, db 123,134,145,156,167,178,189
3152 Depending on personal taste, you can also omit the code part of the
3153 \c{AT} line completely, and start the structure field on the next
3157 \c db 'hello, world'
3161 \S{align} \i\c{ALIGN} and \i\c{ALIGNB}: Data Alignment
3163 The \c{ALIGN} and \c{ALIGNB} macros provides a convenient way to
3164 align code or data on a word, longword, paragraph or other boundary.
3165 (Some assemblers call this directive \i\c{EVEN}.) The syntax of the
3166 \c{ALIGN} and \c{ALIGNB} macros is
3168 \c align 4 ; align on 4-byte boundary
3169 \c align 16 ; align on 16-byte boundary
3170 \c align 8,db 0 ; pad with 0s rather than NOPs
3171 \c align 4,resb 1 ; align to 4 in the BSS
3172 \c alignb 4 ; equivalent to previous line
3174 Both macros require their first argument to be a power of two; they
3175 both compute the number of additional bytes required to bring the
3176 length of the current section up to a multiple of that power of two,
3177 and then apply the \c{TIMES} prefix to their second argument to
3178 perform the alignment.
3180 If the second argument is not specified, the default for \c{ALIGN}
3181 is \c{NOP}, and the default for \c{ALIGNB} is \c{RESB 1}. So if the
3182 second argument is specified, the two macros are equivalent.
3183 Normally, you can just use \c{ALIGN} in code and data sections and
3184 \c{ALIGNB} in BSS sections, and never need the second argument
3185 except for special purposes.
3187 \c{ALIGN} and \c{ALIGNB}, being simple macros, perform no error
3188 checking: they cannot warn you if their first argument fails to be a
3189 power of two, or if their second argument generates more than one
3190 byte of code. In each of these cases they will silently do the wrong
3193 \c{ALIGNB} (or \c{ALIGN} with a second argument of \c{RESB 1}) can
3194 be used within structure definitions:
3211 This will ensure that the structure members are sensibly aligned
3212 relative to the base of the structure.
3214 A final caveat: \c{ALIGN} and \c{ALIGNB} work relative to the
3215 beginning of the \e{section}, not the beginning of the address space
3216 in the final executable. Aligning to a 16-byte boundary when the
3217 section you're in is only guaranteed to be aligned to a 4-byte
3218 boundary, for example, is a waste of effort. Again, NASM does not
3219 check that the section's alignment characteristics are sensible for
3220 the use of \c{ALIGN} or \c{ALIGNB}.
3223 \H{tasmcompat} \i{TASM Compatible Preprocessor Directives}
3225 The following preprocessor directives may only be used when TASM
3226 compatibility is turned on using the \c{-t} command line switch
3227 (This switch is described in \k{opt-t}.)
3229 \b\c{%arg} (see \k{arg})
3231 \b\c{%stacksize} (see \k{stacksize})
3233 \b\c{%local} (see \k{local})
3236 \S{arg} \i\c{%arg} Directive
3238 The \c{%arg} directive is used to simplify the handling of
3239 parameters passed on the stack. Stack based parameter passing
3240 is used by many high level languages, including C, C++ and Pascal.
3242 While NASM comes with macros which attempt to duplicate this
3243 functionality (see \k{16cmacro}), the syntax is not particularly
3244 convenient to use and is not TASM compatible. Here is an example
3245 which shows the use of \c{%arg} without any external macros:
3249 \c %push mycontext ; save the current context
3250 \c %stacksize large ; tell NASM to use bp
3251 \c %arg i:word, j_ptr:word
3258 \c %pop ; restore original context
3260 This is similar to the procedure defined in \k{16cmacro} and adds
3261 the value in i to the value pointed to by j_ptr and returns the
3262 sum in the ax register. See \k{pushpop} for an explanation of
3263 \c{push} and \c{pop} and the use of context stacks.
3266 \S{stacksize} \i\c{%stacksize} Directive
3268 The \c{%stacksize} directive is used in conjunction with the
3269 \c{%arg} (see \k{arg}) and the \c{%local} (see \k{local}) directives.
3270 It tells NASM the default size to use for subsequent \c{%arg} and
3271 \c{%local} directives. The \c{%stacksize} directive takes one
3272 required argument which is one of \c{flat}, \c{large} or \c{small}.
3276 This form causes NASM to use stack-based parameter addressing
3277 relative to \c{ebp} and it assumes that a near form of call was used
3278 to get to this label (i.e. that \c{eip} is on the stack).
3282 This form uses \c{bp} to do stack-based parameter addressing and
3283 assumes that a far form of call was used to get to this address
3284 (i.e. that \c{ip} and \c{cs} are on the stack).
3288 This form also uses \c{bp} to address stack parameters, but it is
3289 different from \c{large} because it also assumes that the old value
3290 of bp is pushed onto the stack (i.e. it expects an \c{ENTER}
3291 instruction). In other words, it expects that \c{bp}, \c{ip} and
3292 \c{cs} are on the top of the stack, underneath any local space which
3293 may have been allocated by \c{ENTER}. This form is probably most
3294 useful when used in combination with the \c{%local} directive
3298 \S{local} \i\c{%local} Directive
3300 The \c{%local} directive is used to simplify the use of local
3301 temporary stack variables allocated in a stack frame. Automatic
3302 local variables in C are an example of this kind of variable. The
3303 \c{%local} directive is most useful when used with the \c{%stacksize}
3304 (see \k{stacksize} and is also compatible with the \c{%arg} directive
3305 (see \k{arg}). It allows simplified reference to variables on the
3306 stack which have been allocated typically by using the \c{ENTER}
3308 \# (see \k{insENTER} for a description of that instruction).
3309 An example of its use is the following:
3313 \c %push mycontext ; save the current context
3314 \c %stacksize small ; tell NASM to use bp
3315 \c %assign %$localsize 0 ; see text for explanation
3316 \c %local old_ax:word, old_dx:word
3318 \c enter %$localsize,0 ; see text for explanation
3319 \c mov [old_ax],ax ; swap ax & bx
3320 \c mov [old_dx],dx ; and swap dx & cx
3325 \c leave ; restore old bp
3328 \c %pop ; restore original context
3330 The \c{%$localsize} variable is used internally by the
3331 \c{%local} directive and \e{must} be defined within the
3332 current context before the \c{%local} directive may be used.
3333 Failure to do so will result in one expression syntax error for
3334 each \c{%local} variable declared. It then may be used in
3335 the construction of an appropriately sized ENTER instruction
3336 as shown in the example.
3338 \H{otherpreproc} \i{Other Preprocessor Directives}
3340 NASM also has preprocessor directives which allow access to
3341 information from external sources. Currently they include:
3343 The following preprocessor directive is supported to allow NASM to
3344 correctly handle output of the cpp C language preprocessor.
3346 \b\c{%line} enables NAsM to correctly handle the output of the cpp
3347 C language preprocessor (see \k{line}).
3349 \b\c{%!} enables NASM to read in the value of an environment variable,
3350 which can then be used in your program (see \k{getenv}).
3352 \S{line} \i\c{%line} Directive
3354 The \c{%line} directive is used to notify NASM that the input line
3355 corresponds to a specific line number in another file. Typically
3356 this other file would be an original source file, with the current
3357 NASM input being the output of a pre-processor. The \c{%line}
3358 directive allows NASM to output messages which indicate the line
3359 number of the original source file, instead of the file that is being
3362 This preprocessor directive is not generally of use to programmers,
3363 by may be of interest to preprocessor authors. The usage of the
3364 \c{%line} preprocessor directive is as follows:
3366 \c %line nnn[+mmm] [filename]
3368 In this directive, \c{nnn} indentifies the line of the original source
3369 file which this line corresponds to. \c{mmm} is an optional parameter
3370 which specifies a line increment value; each line of the input file
3371 read in is considered to correspond to \c{mmm} lines of the original
3372 source file. Finally, \c{filename} is an optional parameter which
3373 specifies the file name of the original source file.
3375 After reading a \c{%line} preprocessor directive, NASM will report
3376 all file name and line numbers relative to the values specified
3380 \S{getenv} \i\c{%!}\c{<env>}: Read an environment variable.
3382 The \c{%!<env>} directive makes it possible to read the value of an
3383 environment variable at assembly time. This could, for example, be used
3384 to store the contents of an environment variable into a string, which
3385 could be used at some other point in your code.
3387 For example, suppose that you have an environment variable \c{FOO}, and
3388 you want the contents of \c{FOO} to be embedded in your program. You
3389 could do that as follows:
3391 \c %define FOO %!FOO
3394 \c tmpstr db quote FOO quote
3396 At the time of writing, this will generate an "unterminated string"
3397 warning at the time of defining "quote", and it will add a space
3398 before and after the string that is read in. I was unable to find
3399 a simple workaround (although a workaround can be created using a
3400 multi-line macro), so I believe that you will need to either learn how
3401 to create more complex macros, or allow for the extra spaces if you
3402 make use of this feature in that way.
3405 \C{directive} \i{Assembler Directives}
3407 NASM, though it attempts to avoid the bureaucracy of assemblers like
3408 MASM and TASM, is nevertheless forced to support a \e{few}
3409 directives. These are described in this chapter.
3411 NASM's directives come in two types: \I{user-level
3412 directives}\e{user-level} directives and \I{primitive
3413 directives}\e{primitive} directives. Typically, each directive has a
3414 user-level form and a primitive form. In almost all cases, we
3415 recommend that users use the user-level forms of the directives,
3416 which are implemented as macros which call the primitive forms.
3418 Primitive directives are enclosed in square brackets; user-level
3421 In addition to the universal directives described in this chapter,
3422 each object file format can optionally supply extra directives in
3423 order to control particular features of that file format. These
3424 \I{format-specific directives}\e{format-specific} directives are
3425 documented along with the formats that implement them, in \k{outfmt}.
3428 \H{bits} \i\c{BITS}: Specifying Target \i{Processor Mode}
3430 The \c{BITS} directive specifies whether NASM should generate code
3431 \I{16-bit mode, versus 32-bit mode}designed to run on a processor
3432 operating in 16-bit mode, 32-bit mode or 64-bit mode. The syntax is
3433 \c{BITS XX}, where XX is 16, 32 or 64.
3435 In most cases, you should not need to use \c{BITS} explicitly. The
3436 \c{aout}, \c{coff}, \c{elf}, \c{macho}, \c{win32} and \c{win64}
3437 object formats, which are designed for use in 32-bit or 64-bit
3438 operating systems, all cause NASM to select 32-bit or 64-bit mode,
3439 respectively, by default. The \c{obj} object format allows you
3440 to specify each segment you define as either \c{USE16} or \c{USE32},
3441 and NASM will set its operating mode accordingly, so the use of the
3442 \c{BITS} directive is once again unnecessary.
3444 The most likely reason for using the \c{BITS} directive is to write
3445 32-bit or 64-bit code in a flat binary file; this is because the \c{bin}
3446 output format defaults to 16-bit mode in anticipation of it being
3447 used most frequently to write DOS \c{.COM} programs, DOS \c{.SYS}
3448 device drivers and boot loader software.
3450 You do \e{not} need to specify \c{BITS 32} merely in order to use
3451 32-bit instructions in a 16-bit DOS program; if you do, the
3452 assembler will generate incorrect code because it will be writing
3453 code targeted at a 32-bit platform, to be run on a 16-bit one.
3455 When NASM is in \c{BITS 16} mode, instructions which use 32-bit
3456 data are prefixed with an 0x66 byte, and those referring to 32-bit
3457 addresses have an 0x67 prefix. In \c{BITS 32} mode, the reverse is
3458 true: 32-bit instructions require no prefixes, whereas instructions
3459 using 16-bit data need an 0x66 and those working on 16-bit addresses
3462 When NASM is in \c{BITS 64} mode, most instructions operate the same
3463 as they do for \c{BITS 32} mode. However, there are 8 more general and
3464 SSE registers, and 16-bit addressing is no longer supported.
3466 The default address size is 64 bits; 32-bit addressing can be selected
3467 with the 0x67 prefix. The default operand size is still 32 bits,
3468 however, and the 0x66 prefix selects 16-bit operand size. The \c{REX}
3469 prefix is used both to select 64-bit operand size, and to access the
3470 new registers. NASM automatically inserts REX prefixes when
3473 When the \c{REX} prefix is used, the processor does not know how to
3474 address the AH, BH, CH or DH (high 8-bit legacy) registers. Instead,
3475 it is possible to access the the low 8-bits of the SP, BP SI and DI
3476 registers as SPL, BPL, SIL and DIL, respectively; but only when the
3479 The \c{BITS} directive has an exactly equivalent primitive form,
3480 \c{[BITS 16]}, \c{[BITS 32]} and \c{[BITS 64]}. The user-level form is
3481 a macro which has no function other than to call the primitive form.
3483 Note that the space is neccessary, e.g. \c{BITS32} will \e{not} work!
3485 \S{USE16 & USE32} \i\c{USE16} & \i\c{USE32}: Aliases for BITS
3487 The `\c{USE16}' and `\c{USE32}' directives can be used in place of
3488 `\c{BITS 16}' and `\c{BITS 32}', for compatibility with other assemblers.
3491 \H{default} \i\c{DEFAULT}: Change the assembler defaults
3493 The \c{DEFAULT} directive changes the assembler defaults. Normally,
3494 NASM defaults to a mode where the programmer is expected to explicitly
3495 specify most features directly. However, this is occationally
3496 obnoxious, as the explicit form is pretty much the only one one wishes
3499 Currently, the only \c{DEFAULT} that is settable is whether or not
3500 registerless instructions in 64-bit mode are \c{RIP}-relative or not.
3501 By default, they are absolute unless overridden with the \i\c{REL}
3502 specifier (see \k{effaddr}). However, if \c{DEFAULT REL} is
3503 specified, \c{REL} is default, unless overridden with the \c{ABS}
3504 specifier, \e{except when used with an FS or GS segment override}.
3506 The special handling of \c{FS} and \c{GS} overrides are due to the
3507 fact that these registers are generally used as thread pointers or
3508 other special functions in 64-bit mode, and generating
3509 \c{RIP}-relative addresses would be extremely confusing.
3511 \c{DEFAULT REL} is disabled with \c{DEFAULT ABS}.
3513 \H{section} \i\c{SECTION} or \i\c{SEGMENT}: Changing and \i{Defining
3516 \I{changing sections}\I{switching between sections}The \c{SECTION}
3517 directive (\c{SEGMENT} is an exactly equivalent synonym) changes
3518 which section of the output file the code you write will be
3519 assembled into. In some object file formats, the number and names of
3520 sections are fixed; in others, the user may make up as many as they
3521 wish. Hence \c{SECTION} may sometimes give an error message, or may
3522 define a new section, if you try to switch to a section that does
3525 The Unix object formats, and the \c{bin} object format (but see
3526 \k{multisec}, all support
3527 the \i{standardized section names} \c{.text}, \c{.data} and \c{.bss}
3528 for the code, data and uninitialized-data sections. The \c{obj}
3529 format, by contrast, does not recognize these section names as being
3530 special, and indeed will strip off the leading period of any section
3534 \S{sectmac} The \i\c{__SECT__} Macro
3536 The \c{SECTION} directive is unusual in that its user-level form
3537 functions differently from its primitive form. The primitive form,
3538 \c{[SECTION xyz]}, simply switches the current target section to the
3539 one given. The user-level form, \c{SECTION xyz}, however, first
3540 defines the single-line macro \c{__SECT__} to be the primitive
3541 \c{[SECTION]} directive which it is about to issue, and then issues
3542 it. So the user-level directive
3546 expands to the two lines
3548 \c %define __SECT__ [SECTION .text]
3551 Users may find it useful to make use of this in their own macros.
3552 For example, the \c{writefile} macro defined in \k{mlmacgre} can be
3553 usefully rewritten in the following more sophisticated form:
3555 \c %macro writefile 2+
3565 \c mov cx,%%endstr-%%str
3572 This form of the macro, once passed a string to output, first
3573 switches temporarily to the data section of the file, using the
3574 primitive form of the \c{SECTION} directive so as not to modify
3575 \c{__SECT__}. It then declares its string in the data section, and
3576 then invokes \c{__SECT__} to switch back to \e{whichever} section
3577 the user was previously working in. It thus avoids the need, in the
3578 previous version of the macro, to include a \c{JMP} instruction to
3579 jump over the data, and also does not fail if, in a complicated
3580 \c{OBJ} format module, the user could potentially be assembling the
3581 code in any of several separate code sections.
3584 \H{absolute} \i\c{ABSOLUTE}: Defining Absolute Labels
3586 The \c{ABSOLUTE} directive can be thought of as an alternative form
3587 of \c{SECTION}: it causes the subsequent code to be directed at no
3588 physical section, but at the hypothetical section starting at the
3589 given absolute address. The only instructions you can use in this
3590 mode are the \c{RESB} family.
3592 \c{ABSOLUTE} is used as follows:
3600 This example describes a section of the PC BIOS data area, at
3601 segment address 0x40: the above code defines \c{kbuf_chr} to be
3602 0x1A, \c{kbuf_free} to be 0x1C, and \c{kbuf} to be 0x1E.
3604 The user-level form of \c{ABSOLUTE}, like that of \c{SECTION},
3605 redefines the \i\c{__SECT__} macro when it is invoked.
3607 \i\c{STRUC} and \i\c{ENDSTRUC} are defined as macros which use
3608 \c{ABSOLUTE} (and also \c{__SECT__}).
3610 \c{ABSOLUTE} doesn't have to take an absolute constant as an
3611 argument: it can take an expression (actually, a \i{critical
3612 expression}: see \k{crit}) and it can be a value in a segment. For
3613 example, a TSR can re-use its setup code as run-time BSS like this:
3615 \c org 100h ; it's a .COM program
3617 \c jmp setup ; setup code comes last
3619 \c ; the resident part of the TSR goes here
3621 \c ; now write the code that installs the TSR here
3625 \c runtimevar1 resw 1
3626 \c runtimevar2 resd 20
3630 This defines some variables `on top of' the setup code, so that
3631 after the setup has finished running, the space it took up can be
3632 re-used as data storage for the running TSR. The symbol `tsr_end'
3633 can be used to calculate the total size of the part of the TSR that
3634 needs to be made resident.
3637 \H{extern} \i\c{EXTERN}: \i{Importing Symbols} from Other Modules
3639 \c{EXTERN} is similar to the MASM directive \c{EXTRN} and the C
3640 keyword \c{extern}: it is used to declare a symbol which is not
3641 defined anywhere in the module being assembled, but is assumed to be
3642 defined in some other module and needs to be referred to by this
3643 one. Not every object-file format can support external variables:
3644 the \c{bin} format cannot.
3646 The \c{EXTERN} directive takes as many arguments as you like. Each
3647 argument is the name of a symbol:
3650 \c extern _sscanf,_fscanf
3652 Some object-file formats provide extra features to the \c{EXTERN}
3653 directive. In all cases, the extra features are used by suffixing a
3654 colon to the symbol name followed by object-format specific text.
3655 For example, the \c{obj} format allows you to declare that the
3656 default segment base of an external should be the group \c{dgroup}
3657 by means of the directive
3659 \c extern _variable:wrt dgroup
3661 The primitive form of \c{EXTERN} differs from the user-level form
3662 only in that it can take only one argument at a time: the support
3663 for multiple arguments is implemented at the preprocessor level.
3665 You can declare the same variable as \c{EXTERN} more than once: NASM
3666 will quietly ignore the second and later redeclarations. You can't
3667 declare a variable as \c{EXTERN} as well as something else, though.
3670 \H{global} \i\c{GLOBAL}: \i{Exporting Symbols} to Other Modules
3672 \c{GLOBAL} is the other end of \c{EXTERN}: if one module declares a
3673 symbol as \c{EXTERN} and refers to it, then in order to prevent
3674 linker errors, some other module must actually \e{define} the
3675 symbol and declare it as \c{GLOBAL}. Some assemblers use the name
3676 \i\c{PUBLIC} for this purpose.
3678 The \c{GLOBAL} directive applying to a symbol must appear \e{before}
3679 the definition of the symbol.
3681 \c{GLOBAL} uses the same syntax as \c{EXTERN}, except that it must
3682 refer to symbols which \e{are} defined in the same module as the
3683 \c{GLOBAL} directive. For example:
3689 \c{GLOBAL}, like \c{EXTERN}, allows object formats to define private
3690 extensions by means of a colon. The \c{elf} object format, for
3691 example, lets you specify whether global data items are functions or
3694 \c global hashlookup:function, hashtable:data
3696 Like \c{EXTERN}, the primitive form of \c{GLOBAL} differs from the
3697 user-level form only in that it can take only one argument at a
3701 \H{common} \i\c{COMMON}: Defining Common Data Areas
3703 The \c{COMMON} directive is used to declare \i\e{common variables}.
3704 A common variable is much like a global variable declared in the
3705 uninitialized data section, so that
3709 is similar in function to
3716 The difference is that if more than one module defines the same
3717 common variable, then at link time those variables will be
3718 \e{merged}, and references to \c{intvar} in all modules will point
3719 at the same piece of memory.
3721 Like \c{GLOBAL} and \c{EXTERN}, \c{COMMON} supports object-format
3722 specific extensions. For example, the \c{obj} format allows common
3723 variables to be NEAR or FAR, and the \c{elf} format allows you to
3724 specify the alignment requirements of a common variable:
3726 \c common commvar 4:near ; works in OBJ
3727 \c common intarray 100:4 ; works in ELF: 4 byte aligned
3729 Once again, like \c{EXTERN} and \c{GLOBAL}, the primitive form of
3730 \c{COMMON} differs from the user-level form only in that it can take
3731 only one argument at a time.
3734 \H{CPU} \i\c{CPU}: Defining CPU Dependencies
3736 The \i\c{CPU} directive restricts assembly to those instructions which
3737 are available on the specified CPU.
3741 \b\c{CPU 8086} Assemble only 8086 instruction set
3743 \b\c{CPU 186} Assemble instructions up to the 80186 instruction set
3745 \b\c{CPU 286} Assemble instructions up to the 286 instruction set
3747 \b\c{CPU 386} Assemble instructions up to the 386 instruction set
3749 \b\c{CPU 486} 486 instruction set
3751 \b\c{CPU 586} Pentium instruction set
3753 \b\c{CPU PENTIUM} Same as 586
3755 \b\c{CPU 686} P6 instruction set
3757 \b\c{CPU PPRO} Same as 686
3759 \b\c{CPU P2} Same as 686
3761 \b\c{CPU P3} Pentium III (Katmai) instruction sets
3763 \b\c{CPU KATMAI} Same as P3
3765 \b\c{CPU P4} Pentium 4 (Willamette) instruction set
3767 \b\c{CPU WILLAMETTE} Same as P4
3769 \b\c{CPU PRESCOTT} Prescott instruction set
3771 \b\c{CPU X64} x86-64 (x64/AMD64/EM64T) instruction set
3773 \b\c{CPU IA64} IA64 CPU (in x86 mode) instruction set
3775 All options are case insensitive. All instructions will be selected
3776 only if they apply to the selected CPU or lower. By default, all
3777 instructions are available.
3780 \C{outfmt} \i{Output Formats}
3782 NASM is a portable assembler, designed to be able to compile on any
3783 ANSI C-supporting platform and produce output to run on a variety of
3784 Intel x86 operating systems. For this reason, it has a large number
3785 of available output formats, selected using the \i\c{-f} option on
3786 the NASM \i{command line}. Each of these formats, along with its
3787 extensions to the base NASM syntax, is detailed in this chapter.
3789 As stated in \k{opt-o}, NASM chooses a \i{default name} for your
3790 output file based on the input file name and the chosen output
3791 format. This will be generated by removing the \i{extension}
3792 (\c{.asm}, \c{.s}, or whatever you like to use) from the input file
3793 name, and substituting an extension defined by the output format.
3794 The extensions are given with each format below.
3797 \H{binfmt} \i\c{bin}: \i{Flat-Form Binary}\I{pure binary} Output
3799 The \c{bin} format does not produce object files: it generates
3800 nothing in the output file except the code you wrote. Such `pure
3801 binary' files are used by \i{MS-DOS}: \i\c{.COM} executables and
3802 \i\c{.SYS} device drivers are pure binary files. Pure binary output
3803 is also useful for \i{operating system} and \i{boot loader}
3806 The \c{bin} format supports \i{multiple section names}. For details of
3807 how nasm handles sections in the \c{bin} format, see \k{multisec}.
3809 Using the \c{bin} format puts NASM by default into 16-bit mode (see
3810 \k{bits}). In order to use \c{bin} to write 32-bit or 64-bit code,
3811 such as an OS kernel, you need to explicitly issue the \I\c{BITS}\c{BITS 32}
3812 or \I\c{BITS}\c{BITS 64} directive.
3814 \c{bin} has no default output file name extension: instead, it
3815 leaves your file name as it is once the original extension has been
3816 removed. Thus, the default is for NASM to assemble \c{binprog.asm}
3817 into a binary file called \c{binprog}.
3820 \S{org} \i\c{ORG}: Binary File \i{Program Origin}
3822 The \c{bin} format provides an additional directive to the list
3823 given in \k{directive}: \c{ORG}. The function of the \c{ORG}
3824 directive is to specify the origin address which NASM will assume
3825 the program begins at when it is loaded into memory.
3827 For example, the following code will generate the longword
3834 Unlike the \c{ORG} directive provided by MASM-compatible assemblers,
3835 which allows you to jump around in the object file and overwrite
3836 code you have already generated, NASM's \c{ORG} does exactly what
3837 the directive says: \e{origin}. Its sole function is to specify one
3838 offset which is added to all internal address references within the
3839 section; it does not permit any of the trickery that MASM's version
3840 does. See \k{proborg} for further comments.
3843 \S{binseg} \c{bin} Extensions to the \c{SECTION}
3844 Directive\I{SECTION, bin extensions to}
3846 The \c{bin} output format extends the \c{SECTION} (or \c{SEGMENT})
3847 directive to allow you to specify the alignment requirements of
3848 segments. This is done by appending the \i\c{ALIGN} qualifier to the
3849 end of the section-definition line. For example,
3851 \c section .data align=16
3853 switches to the section \c{.data} and also specifies that it must be
3854 aligned on a 16-byte boundary.
3856 The parameter to \c{ALIGN} specifies how many low bits of the
3857 section start address must be forced to zero. The alignment value
3858 given may be any power of two.\I{section alignment, in
3859 bin}\I{segment alignment, in bin}\I{alignment, in bin sections}
3862 \S{multisec} \i\c{Multisection}\I{bin, multisection} support for the BIN format.
3864 The \c{bin} format allows the use of multiple sections, of arbitrary names,
3865 besides the "known" \c{.text}, \c{.data}, and \c{.bss} names.
3867 \b Sections may be designated \i\c{progbits} or \i\c{nobits}. Default
3868 is \c{progbits} (except \c{.bss}, which defaults to \c{nobits},
3871 \b Sections can be aligned at a specified boundary following the previous
3872 section with \c{align=}, or at an arbitrary byte-granular position with
3875 \b Sections can be given a virtual start address, which will be used
3876 for the calculation of all memory references within that section
3879 \b Sections can be ordered using \i\c{follows=}\c{<section>} or
3880 \i\c{vfollows=}\c{<section>} as an alternative to specifying an explicit
3883 \b Arguments to \c{org}, \c{start}, \c{vstart}, and \c{align=} are
3884 critical expressions. See \k{crit}. E.g. \c{align=(1 << ALIGN_SHIFT)}
3885 - \c{ALIGN_SHIFT} must be defined before it is used here.
3887 \b Any code which comes before an explicit \c{SECTION} directive
3888 is directed by default into the \c{.text} section.
3890 \b If an \c{ORG} statement is not given, \c{ORG 0} is used
3893 \b The \c{.bss} section will be placed after the last \c{progbits}
3894 section, unless \c{start=}, \c{vstart=}, \c{follows=}, or \c{vfollows=}
3897 \b All sections are aligned on dword boundaries, unless a different
3898 alignment has been specified.
3900 \b Sections may not overlap.
3902 \b Nasm creates the \c{section.<secname>.start} for each section,
3903 which may be used in your code.
3905 \S{map}\i{Map files}
3907 Map files can be generated in \c{-f bin} format by means of the \c{[map]}
3908 option. Map types of \c{all} (default), \c{brief}, \c{sections}, \c{segments},
3909 or \c{symbols} may be specified. Output may be directed to \c{stdout}
3910 (default), \c{stderr}, or a specified file. E.g.
3911 \c{[map symbols myfile.map]}. No "user form" exists, the square
3912 brackets must be used.
3915 \H{objfmt} \i\c{obj}: \i{Microsoft OMF}\I{OMF} Object Files
3917 The \c{obj} file format (NASM calls it \c{obj} rather than \c{omf}
3918 for historical reasons) is the one produced by \i{MASM} and
3919 \i{TASM}, which is typically fed to 16-bit DOS linkers to produce
3920 \i\c{.EXE} files. It is also the format used by \i{OS/2}.
3922 \c{obj} provides a default output file-name extension of \c{.obj}.
3924 \c{obj} is not exclusively a 16-bit format, though: NASM has full
3925 support for the 32-bit extensions to the format. In particular,
3926 32-bit \c{obj} format files are used by \i{Borland's Win32
3927 compilers}, instead of using Microsoft's newer \i\c{win32} object
3930 The \c{obj} format does not define any special segment names: you
3931 can call your segments anything you like. Typical names for segments
3932 in \c{obj} format files are \c{CODE}, \c{DATA} and \c{BSS}.
3934 If your source file contains code before specifying an explicit
3935 \c{SEGMENT} directive, then NASM will invent its own segment called
3936 \i\c{__NASMDEFSEG} for you.
3938 When you define a segment in an \c{obj} file, NASM defines the
3939 segment name as a symbol as well, so that you can access the segment
3940 address of the segment. So, for example:
3949 \c mov ax,data ; get segment address of data
3950 \c mov ds,ax ; and move it into DS
3951 \c inc word [dvar] ; now this reference will work
3954 The \c{obj} format also enables the use of the \i\c{SEG} and
3955 \i\c{WRT} operators, so that you can write code which does things
3960 \c mov ax,seg foo ; get preferred segment of foo
3962 \c mov ax,data ; a different segment
3964 \c mov ax,[ds:foo] ; this accesses `foo'
3965 \c mov [es:foo wrt data],bx ; so does this
3968 \S{objseg} \c{obj} Extensions to the \c{SEGMENT}
3969 Directive\I{SEGMENT, obj extensions to}
3971 The \c{obj} output format extends the \c{SEGMENT} (or \c{SECTION})
3972 directive to allow you to specify various properties of the segment
3973 you are defining. This is done by appending extra qualifiers to the
3974 end of the segment-definition line. For example,
3976 \c segment code private align=16
3978 defines the segment \c{code}, but also declares it to be a private
3979 segment, and requires that the portion of it described in this code
3980 module must be aligned on a 16-byte boundary.
3982 The available qualifiers are:
3984 \b \i\c{PRIVATE}, \i\c{PUBLIC}, \i\c{COMMON} and \i\c{STACK} specify
3985 the combination characteristics of the segment. \c{PRIVATE} segments
3986 do not get combined with any others by the linker; \c{PUBLIC} and
3987 \c{STACK} segments get concatenated together at link time; and
3988 \c{COMMON} segments all get overlaid on top of each other rather
3989 than stuck end-to-end.
3991 \b \i\c{ALIGN} is used, as shown above, to specify how many low bits
3992 of the segment start address must be forced to zero. The alignment
3993 value given may be any power of two from 1 to 4096; in reality, the
3994 only values supported are 1, 2, 4, 16, 256 and 4096, so if 8 is
3995 specified it will be rounded up to 16, and 32, 64 and 128 will all
3996 be rounded up to 256, and so on. Note that alignment to 4096-byte
3997 boundaries is a \i{PharLap} extension to the format and may not be
3998 supported by all linkers.\I{section alignment, in OBJ}\I{segment
3999 alignment, in OBJ}\I{alignment, in OBJ sections}
4001 \b \i\c{CLASS} can be used to specify the segment class; this feature
4002 indicates to the linker that segments of the same class should be
4003 placed near each other in the output file. The class name can be any
4004 word, e.g. \c{CLASS=CODE}.
4006 \b \i\c{OVERLAY}, like \c{CLASS}, is specified with an arbitrary word
4007 as an argument, and provides overlay information to an
4008 overlay-capable linker.
4010 \b Segments can be declared as \i\c{USE16} or \i\c{USE32}, which has
4011 the effect of recording the choice in the object file and also
4012 ensuring that NASM's default assembly mode when assembling in that
4013 segment is 16-bit or 32-bit respectively.
4015 \b When writing \i{OS/2} object files, you should declare 32-bit
4016 segments as \i\c{FLAT}, which causes the default segment base for
4017 anything in the segment to be the special group \c{FLAT}, and also
4018 defines the group if it is not already defined.
4020 \b The \c{obj} file format also allows segments to be declared as
4021 having a pre-defined absolute segment address, although no linkers
4022 are currently known to make sensible use of this feature;
4023 nevertheless, NASM allows you to declare a segment such as
4024 \c{SEGMENT SCREEN ABSOLUTE=0xB800} if you need to. The \i\c{ABSOLUTE}
4025 and \c{ALIGN} keywords are mutually exclusive.
4027 NASM's default segment attributes are \c{PUBLIC}, \c{ALIGN=1}, no
4028 class, no overlay, and \c{USE16}.
4031 \S{group} \i\c{GROUP}: Defining Groups of Segments\I{segments, groups of}
4033 The \c{obj} format also allows segments to be grouped, so that a
4034 single segment register can be used to refer to all the segments in
4035 a group. NASM therefore supplies the \c{GROUP} directive, whereby
4044 \c ; some uninitialized data
4046 \c group dgroup data bss
4048 which will define a group called \c{dgroup} to contain the segments
4049 \c{data} and \c{bss}. Like \c{SEGMENT}, \c{GROUP} causes the group
4050 name to be defined as a symbol, so that you can refer to a variable
4051 \c{var} in the \c{data} segment as \c{var wrt data} or as \c{var wrt
4052 dgroup}, depending on which segment value is currently in your
4055 If you just refer to \c{var}, however, and \c{var} is declared in a
4056 segment which is part of a group, then NASM will default to giving
4057 you the offset of \c{var} from the beginning of the \e{group}, not
4058 the \e{segment}. Therefore \c{SEG var}, also, will return the group
4059 base rather than the segment base.
4061 NASM will allow a segment to be part of more than one group, but
4062 will generate a warning if you do this. Variables declared in a
4063 segment which is part of more than one group will default to being
4064 relative to the first group that was defined to contain the segment.
4066 A group does not have to contain any segments; you can still make
4067 \c{WRT} references to a group which does not contain the variable
4068 you are referring to. OS/2, for example, defines the special group
4069 \c{FLAT} with no segments in it.
4072 \S{uppercase} \i\c{UPPERCASE}: Disabling Case Sensitivity in Output
4074 Although NASM itself is \i{case sensitive}, some OMF linkers are
4075 not; therefore it can be useful for NASM to output single-case
4076 object files. The \c{UPPERCASE} format-specific directive causes all
4077 segment, group and symbol names that are written to the object file
4078 to be forced to upper case just before being written. Within a
4079 source file, NASM is still case-sensitive; but the object file can
4080 be written entirely in upper case if desired.
4082 \c{UPPERCASE} is used alone on a line; it requires no parameters.
4085 \S{import} \i\c{IMPORT}: Importing DLL Symbols\I{DLL symbols,
4086 importing}\I{symbols, importing from DLLs}
4088 The \c{IMPORT} format-specific directive defines a symbol to be
4089 imported from a DLL, for use if you are writing a DLL's \i{import
4090 library} in NASM. You still need to declare the symbol as \c{EXTERN}
4091 as well as using the \c{IMPORT} directive.
4093 The \c{IMPORT} directive takes two required parameters, separated by
4094 white space, which are (respectively) the name of the symbol you
4095 wish to import and the name of the library you wish to import it
4098 \c import WSAStartup wsock32.dll
4100 A third optional parameter gives the name by which the symbol is
4101 known in the library you are importing it from, in case this is not
4102 the same as the name you wish the symbol to be known by to your code
4103 once you have imported it. For example:
4105 \c import asyncsel wsock32.dll WSAAsyncSelect
4108 \S{export} \i\c{EXPORT}: Exporting DLL Symbols\I{DLL symbols,
4109 exporting}\I{symbols, exporting from DLLs}
4111 The \c{EXPORT} format-specific directive defines a global symbol to
4112 be exported as a DLL symbol, for use if you are writing a DLL in
4113 NASM. You still need to declare the symbol as \c{GLOBAL} as well as
4114 using the \c{EXPORT} directive.
4116 \c{EXPORT} takes one required parameter, which is the name of the
4117 symbol you wish to export, as it was defined in your source file. An
4118 optional second parameter (separated by white space from the first)
4119 gives the \e{external} name of the symbol: the name by which you
4120 wish the symbol to be known to programs using the DLL. If this name
4121 is the same as the internal name, you may leave the second parameter
4124 Further parameters can be given to define attributes of the exported
4125 symbol. These parameters, like the second, are separated by white
4126 space. If further parameters are given, the external name must also
4127 be specified, even if it is the same as the internal name. The
4128 available attributes are:
4130 \b \c{resident} indicates that the exported name is to be kept
4131 resident by the system loader. This is an optimisation for
4132 frequently used symbols imported by name.
4134 \b \c{nodata} indicates that the exported symbol is a function which
4135 does not make use of any initialized data.
4137 \b \c{parm=NNN}, where \c{NNN} is an integer, sets the number of
4138 parameter words for the case in which the symbol is a call gate
4139 between 32-bit and 16-bit segments.
4141 \b An attribute which is just a number indicates that the symbol
4142 should be exported with an identifying number (ordinal), and gives
4148 \c export myfunc TheRealMoreFormalLookingFunctionName
4149 \c export myfunc myfunc 1234 ; export by ordinal
4150 \c export myfunc myfunc resident parm=23 nodata
4153 \S{dotdotstart} \i\c{..start}: Defining the \i{Program Entry
4156 \c{OMF} linkers require exactly one of the object files being linked to
4157 define the program entry point, where execution will begin when the
4158 program is run. If the object file that defines the entry point is
4159 assembled using NASM, you specify the entry point by declaring the
4160 special symbol \c{..start} at the point where you wish execution to
4164 \S{objextern} \c{obj} Extensions to the \c{EXTERN}
4165 Directive\I{EXTERN, obj extensions to}
4167 If you declare an external symbol with the directive
4171 then references such as \c{mov ax,foo} will give you the offset of
4172 \c{foo} from its preferred segment base (as specified in whichever
4173 module \c{foo} is actually defined in). So to access the contents of
4174 \c{foo} you will usually need to do something like
4176 \c mov ax,seg foo ; get preferred segment base
4177 \c mov es,ax ; move it into ES
4178 \c mov ax,[es:foo] ; and use offset `foo' from it
4180 This is a little unwieldy, particularly if you know that an external
4181 is going to be accessible from a given segment or group, say
4182 \c{dgroup}. So if \c{DS} already contained \c{dgroup}, you could
4185 \c mov ax,[foo wrt dgroup]
4187 However, having to type this every time you want to access \c{foo}
4188 can be a pain; so NASM allows you to declare \c{foo} in the
4191 \c extern foo:wrt dgroup
4193 This form causes NASM to pretend that the preferred segment base of
4194 \c{foo} is in fact \c{dgroup}; so the expression \c{seg foo} will
4195 now return \c{dgroup}, and the expression \c{foo} is equivalent to
4198 This \I{default-WRT mechanism}default-\c{WRT} mechanism can be used
4199 to make externals appear to be relative to any group or segment in
4200 your program. It can also be applied to common variables: see
4204 \S{objcommon} \c{obj} Extensions to the \c{COMMON}
4205 Directive\I{COMMON, obj extensions to}
4207 The \c{obj} format allows common variables to be either near\I{near
4208 common variables} or far\I{far common variables}; NASM allows you to
4209 specify which your variables should be by the use of the syntax
4211 \c common nearvar 2:near ; `nearvar' is a near common
4212 \c common farvar 10:far ; and `farvar' is far
4214 Far common variables may be greater in size than 64Kb, and so the
4215 OMF specification says that they are declared as a number of
4216 \e{elements} of a given size. So a 10-byte far common variable could
4217 be declared as ten one-byte elements, five two-byte elements, two
4218 five-byte elements or one ten-byte element.
4220 Some \c{OMF} linkers require the \I{element size, in common
4221 variables}\I{common variables, element size}element size, as well as
4222 the variable size, to match when resolving common variables declared
4223 in more than one module. Therefore NASM must allow you to specify
4224 the element size on your far common variables. This is done by the
4227 \c common c_5by2 10:far 5 ; two five-byte elements
4228 \c common c_2by5 10:far 2 ; five two-byte elements
4230 If no element size is specified, the default is 1. Also, the \c{FAR}
4231 keyword is not required when an element size is specified, since
4232 only far commons may have element sizes at all. So the above
4233 declarations could equivalently be
4235 \c common c_5by2 10:5 ; two five-byte elements
4236 \c common c_2by5 10:2 ; five two-byte elements
4238 In addition to these extensions, the \c{COMMON} directive in \c{obj}
4239 also supports default-\c{WRT} specification like \c{EXTERN} does
4240 (explained in \k{objextern}). So you can also declare things like
4242 \c common foo 10:wrt dgroup
4243 \c common bar 16:far 2:wrt data
4244 \c common baz 24:wrt data:6
4247 \H{win32fmt} \i\c{win32}: Microsoft Win32 Object Files
4249 The \c{win32} output format generates Microsoft Win32 object files,
4250 suitable for passing to Microsoft linkers such as \i{Visual C++}.
4251 Note that Borland Win32 compilers do not use this format, but use
4252 \c{obj} instead (see \k{objfmt}).
4254 \c{win32} provides a default output file-name extension of \c{.obj}.
4256 Note that although Microsoft say that Win32 object files follow the
4257 \c{COFF} (Common Object File Format) standard, the object files produced
4258 by Microsoft Win32 compilers are not compatible with COFF linkers
4259 such as DJGPP's, and vice versa. This is due to a difference of
4260 opinion over the precise semantics of PC-relative relocations. To
4261 produce COFF files suitable for DJGPP, use NASM's \c{coff} output
4262 format; conversely, the \c{coff} format does not produce object
4263 files that Win32 linkers can generate correct output from.
4266 \S{win32sect} \c{win32} Extensions to the \c{SECTION}
4267 Directive\I{SECTION, win32 extensions to}
4269 Like the \c{obj} format, \c{win32} allows you to specify additional
4270 information on the \c{SECTION} directive line, to control the type
4271 and properties of sections you declare. Section types and properties
4272 are generated automatically by NASM for the \i{standard section names}
4273 \c{.text}, \c{.data} and \c{.bss}, but may still be overridden by
4276 The available qualifiers are:
4278 \b \c{code}, or equivalently \c{text}, defines the section to be a
4279 code section. This marks the section as readable and executable, but
4280 not writable, and also indicates to the linker that the type of the
4283 \b \c{data} and \c{bss} define the section to be a data section,
4284 analogously to \c{code}. Data sections are marked as readable and
4285 writable, but not executable. \c{data} declares an initialized data
4286 section, whereas \c{bss} declares an uninitialized data section.
4288 \b \c{rdata} declares an initialized data section that is readable
4289 but not writable. Microsoft compilers use this section to place
4292 \b \c{info} defines the section to be an \i{informational section},
4293 which is not included in the executable file by the linker, but may
4294 (for example) pass information \e{to} the linker. For example,
4295 declaring an \c{info}-type section called \i\c{.drectve} causes the
4296 linker to interpret the contents of the section as command-line
4299 \b \c{align=}, used with a trailing number as in \c{obj}, gives the
4300 \I{section alignment, in win32}\I{alignment, in win32
4301 sections}alignment requirements of the section. The maximum you may
4302 specify is 64: the Win32 object file format contains no means to
4303 request a greater section alignment than this. If alignment is not
4304 explicitly specified, the defaults are 16-byte alignment for code
4305 sections, 8-byte alignment for rdata sections and 4-byte alignment
4306 for data (and BSS) sections.
4307 Informational sections get a default alignment of 1 byte (no
4308 alignment), though the value does not matter.
4310 The defaults assumed by NASM if you do not specify the above
4313 \c section .text code align=16
4314 \c section .data data align=4
4315 \c section .rdata rdata align=8
4316 \c section .bss bss align=4
4318 Any other section name is treated by default like \c{.text}.
4321 \H{win64fmt} \i\c{win64}: Microsoft Win64 Object Files
4323 The \c{win64} output format generates Microsoft Win64 object files,
4324 which is nearly 100% indentical to the \c{win32} object format (\k{win32fmt})
4325 with the exception that it is meant to target 64-bit code and the x86-64
4326 platform altogether. This object file is used exactly the same as the \c{win32}
4327 object format (\k{win32fmt}), in NASM, with regard to this exception.
4330 \H{cofffmt} \i\c{coff}: \i{Common Object File Format}
4332 The \c{coff} output type produces \c{COFF} object files suitable for
4333 linking with the \i{DJGPP} linker.
4335 \c{coff} provides a default output file-name extension of \c{.o}.
4337 The \c{coff} format supports the same extensions to the \c{SECTION}
4338 directive as \c{win32} does, except that the \c{align} qualifier and
4339 the \c{info} section type are not supported.
4341 \H{machofmt} \i\c{macho}: \i{Mach Object File Format}
4343 The \c{macho} output type produces \c{Mach-O} object files suitable for
4344 linking with the \i{Mac OSX} linker.
4346 \c{macho} provides a default output file-name extension of \c{.o}.
4348 \H{elffmt} \i\c{elf, elf32, and elf64}: \I{ELF}\I{linux, elf}\i{Executable and Linkable
4349 Format} Object Files
4351 The \c{elf32} and \c{elf64} output formats generate \c{ELF32 and ELF64} (Executable and Linkable Format) object files, as used by Linux as well as \i{Unix System V},
4352 including \i{Solaris x86}, \i{UnixWare} and \i{SCO Unix}. \c{elf}
4353 provides a default output file-name extension of \c{.o}. \c{elf} is a synonym for \c{elf32}.
4356 \S{elfsect} \c{elf} Extensions to the \c{SECTION}
4357 Directive\I{SECTION, elf extensions to}
4359 Like the \c{obj} format, \c{elf} allows you to specify additional
4360 information on the \c{SECTION} directive line, to control the type
4361 and properties of sections you declare. Section types and properties
4362 are generated automatically by NASM for the \i{standard section
4363 names} \i\c{.text}, \i\c{.data} and \i\c{.bss}, but may still be
4364 overridden by these qualifiers.
4366 The available qualifiers are:
4368 \b \i\c{alloc} defines the section to be one which is loaded into
4369 memory when the program is run. \i\c{noalloc} defines it to be one
4370 which is not, such as an informational or comment section.
4372 \b \i\c{exec} defines the section to be one which should have execute
4373 permission when the program is run. \i\c{noexec} defines it as one
4376 \b \i\c{write} defines the section to be one which should be writable
4377 when the program is run. \i\c{nowrite} defines it as one which should
4380 \b \i\c{progbits} defines the section to be one with explicit contents
4381 stored in the object file: an ordinary code or data section, for
4382 example, \i\c{nobits} defines the section to be one with no explicit
4383 contents given, such as a BSS section.
4385 \b \c{align=}, used with a trailing number as in \c{obj}, gives the
4386 \I{section alignment, in elf}\I{alignment, in elf sections}alignment
4387 requirements of the section.
4389 The defaults assumed by NASM if you do not specify the above
4392 \c section .text progbits alloc exec nowrite align=16
4393 \c section .rodata progbits alloc noexec nowrite align=4
4394 \c section .data progbits alloc noexec write align=4
4395 \c section .bss nobits alloc noexec write align=4
4396 \c section other progbits alloc noexec nowrite align=1
4398 (Any section name other than \c{.text}, \c{.rodata}, \c{.data} and
4399 \c{.bss} is treated by default like \c{other} in the above code.)
4402 \S{elfwrt} \i{Position-Independent Code}\I{PIC}: \c{elf} Special
4403 Symbols and \i\c{WRT}
4405 The \c{ELF} specification contains enough features to allow
4406 position-independent code (PIC) to be written, which makes \i{ELF
4407 shared libraries} very flexible. However, it also means NASM has to
4408 be able to generate a variety of strange relocation types in ELF
4409 object files, if it is to be an assembler which can write PIC.
4411 Since \c{ELF} does not support segment-base references, the \c{WRT}
4412 operator is not used for its normal purpose; therefore NASM's
4413 \c{elf} output format makes use of \c{WRT} for a different purpose,
4414 namely the PIC-specific \I{relocations, PIC-specific}relocation
4417 \c{elf} defines five special symbols which you can use as the
4418 right-hand side of the \c{WRT} operator to obtain PIC relocation
4419 types. They are \i\c{..gotpc}, \i\c{..gotoff}, \i\c{..got},
4420 \i\c{..plt} and \i\c{..sym}. Their functions are summarized here:
4422 \b Referring to the symbol marking the global offset table base
4423 using \c{wrt ..gotpc} will end up giving the distance from the
4424 beginning of the current section to the global offset table.
4425 (\i\c{_GLOBAL_OFFSET_TABLE_} is the standard symbol name used to
4426 refer to the \i{GOT}.) So you would then need to add \i\c{$$} to the
4427 result to get the real address of the GOT.
4429 \b Referring to a location in one of your own sections using \c{wrt
4430 ..gotoff} will give the distance from the beginning of the GOT to
4431 the specified location, so that adding on the address of the GOT
4432 would give the real address of the location you wanted.
4434 \b Referring to an external or global symbol using \c{wrt ..got}
4435 causes the linker to build an entry \e{in} the GOT containing the
4436 address of the symbol, and the reference gives the distance from the
4437 beginning of the GOT to the entry; so you can add on the address of
4438 the GOT, load from the resulting address, and end up with the
4439 address of the symbol.
4441 \b Referring to a procedure name using \c{wrt ..plt} causes the
4442 linker to build a \i{procedure linkage table} entry for the symbol,
4443 and the reference gives the address of the \i{PLT} entry. You can
4444 only use this in contexts which would generate a PC-relative
4445 relocation normally (i.e. as the destination for \c{CALL} or
4446 \c{JMP}), since ELF contains no relocation type to refer to PLT
4449 \b Referring to a symbol name using \c{wrt ..sym} causes NASM to
4450 write an ordinary relocation, but instead of making the relocation
4451 relative to the start of the section and then adding on the offset
4452 to the symbol, it will write a relocation record aimed directly at
4453 the symbol in question. The distinction is a necessary one due to a
4454 peculiarity of the dynamic linker.
4456 A fuller explanation of how to use these relocation types to write
4457 shared libraries entirely in NASM is given in \k{picdll}.
4460 \S{elfglob} \c{elf} Extensions to the \c{GLOBAL} Directive\I{GLOBAL,
4461 elf extensions to}\I{GLOBAL, aoutb extensions to}
4463 \c{ELF} object files can contain more information about a global symbol
4464 than just its address: they can contain the \I{symbol sizes,
4465 specifying}\I{size, of symbols}size of the symbol and its \I{symbol
4466 types, specifying}\I{type, of symbols}type as well. These are not
4467 merely debugger conveniences, but are actually necessary when the
4468 program being written is a \i{shared library}. NASM therefore
4469 supports some extensions to the \c{GLOBAL} directive, allowing you
4470 to specify these features.
4472 You can specify whether a global variable is a function or a data
4473 object by suffixing the name with a colon and the word
4474 \i\c{function} or \i\c{data}. (\i\c{object} is a synonym for
4475 \c{data}.) For example:
4477 \c global hashlookup:function, hashtable:data
4479 exports the global symbol \c{hashlookup} as a function and
4480 \c{hashtable} as a data object.
4482 Optionally, you can control the ELF visibility of the symbol. Just
4483 add one of the visibility keywords: \i\c{default}, \i\c{internal},
4484 \i\c{hidden}, or \i\c{protected}. The default is \i\c{default} of
4485 course. For example, to make \c{hashlookup} hidden:
4487 \c global hashlookup:function hidden
4489 You can also specify the size of the data associated with the
4490 symbol, as a numeric expression (which may involve labels, and even
4491 forward references) after the type specifier. Like this:
4493 \c global hashtable:data (hashtable.end - hashtable)
4496 \c db this,that,theother ; some data here
4499 This makes NASM automatically calculate the length of the table and
4500 place that information into the \c{ELF} symbol table.
4502 Declaring the type and size of global symbols is necessary when
4503 writing shared library code. For more information, see
4507 \S{elfcomm} \c{elf} Extensions to the \c{COMMON} Directive
4508 \I{COMMON, elf extensions to}
4510 \c{ELF} also allows you to specify alignment requirements \I{common
4511 variables, alignment in elf}\I{alignment, of elf common variables}on
4512 common variables. This is done by putting a number (which must be a
4513 power of two) after the name and size of the common variable,
4514 separated (as usual) by a colon. For example, an array of
4515 doublewords would benefit from 4-byte alignment:
4517 \c common dwordarray 128:4
4519 This declares the total size of the array to be 128 bytes, and
4520 requires that it be aligned on a 4-byte boundary.
4523 \S{elf16} 16-bit code and ELF
4524 \I{ELF, 16-bit code and}
4526 The \c{ELF32} specification doesn't provide relocations for 8- and
4527 16-bit values, but the GNU \c{ld} linker adds these as an extension.
4528 NASM can generate GNU-compatible relocations, to allow 16-bit code to
4529 be linked as ELF using GNU \c{ld}. If NASM is used with the
4530 \c{-w+gnu-elf-extensions} option, a warning is issued when one of
4531 these relocations is generated.
4533 \H{aoutfmt} \i\c{aout}: Linux \I{a.out, Linux version}\I{linux, a.out}\c{a.out} Object Files
4535 The \c{aout} format generates \c{a.out} object files, in the form used
4536 by early Linux systems (current Linux systems use ELF, see
4537 \k{elffmt}.) These differ from other \c{a.out} object files in that
4538 the magic number in the first four bytes of the file is
4539 different; also, some implementations of \c{a.out}, for example
4540 NetBSD's, support position-independent code, which Linux's
4541 implementation does not.
4543 \c{a.out} provides a default output file-name extension of \c{.o}.
4545 \c{a.out} is a very simple object format. It supports no special
4546 directives, no special symbols, no use of \c{SEG} or \c{WRT}, and no
4547 extensions to any standard directives. It supports only the three
4548 \i{standard section names} \i\c{.text}, \i\c{.data} and \i\c{.bss}.
4551 \H{aoutfmt} \i\c{aoutb}: \i{NetBSD}/\i{FreeBSD}/\i{OpenBSD}
4552 \I{a.out, BSD version}\c{a.out} Object Files
4554 The \c{aoutb} format generates \c{a.out} object files, in the form
4555 used by the various free \c{BSD Unix} clones, \c{NetBSD}, \c{FreeBSD}
4556 and \c{OpenBSD}. For simple object files, this object format is exactly
4557 the same as \c{aout} except for the magic number in the first four bytes
4558 of the file. However, the \c{aoutb} format supports
4559 \I{PIC}\i{position-independent code} in the same way as the \c{elf}
4560 format, so you can use it to write \c{BSD} \i{shared libraries}.
4562 \c{aoutb} provides a default output file-name extension of \c{.o}.
4564 \c{aoutb} supports no special directives, no special symbols, and
4565 only the three \i{standard section names} \i\c{.text}, \i\c{.data}
4566 and \i\c{.bss}. However, it also supports the same use of \i\c{WRT} as
4567 \c{elf} does, to provide position-independent code relocation types.
4568 See \k{elfwrt} for full documentation of this feature.
4570 \c{aoutb} also supports the same extensions to the \c{GLOBAL}
4571 directive as \c{elf} does: see \k{elfglob} for documentation of
4575 \H{as86fmt} \c{as86}: \i{Minix}/Linux\I{linux, as86} \i\c{as86} Object Files
4577 The Minix/Linux 16-bit assembler \c{as86} has its own non-standard
4578 object file format. Although its companion linker \i\c{ld86} produces
4579 something close to ordinary \c{a.out} binaries as output, the object
4580 file format used to communicate between \c{as86} and \c{ld86} is not
4583 NASM supports this format, just in case it is useful, as \c{as86}.
4584 \c{as86} provides a default output file-name extension of \c{.o}.
4586 \c{as86} is a very simple object format (from the NASM user's point
4587 of view). It supports no special directives, no special symbols, no
4588 use of \c{SEG} or \c{WRT}, and no extensions to any standard
4589 directives. It supports only the three \i{standard section names}
4590 \i\c{.text}, \i\c{.data} and \i\c{.bss}.
4593 \H{rdffmt} \I{RDOFF}\i\c{rdf}: \i{Relocatable Dynamic Object File
4596 The \c{rdf} output format produces \c{RDOFF} object files. \c{RDOFF}
4597 (Relocatable Dynamic Object File Format) is a home-grown object-file
4598 format, designed alongside NASM itself and reflecting in its file
4599 format the internal structure of the assembler.
4601 \c{RDOFF} is not used by any well-known operating systems. Those
4602 writing their own systems, however, may well wish to use \c{RDOFF}
4603 as their object format, on the grounds that it is designed primarily
4604 for simplicity and contains very little file-header bureaucracy.
4606 The Unix NASM archive, and the DOS archive which includes sources,
4607 both contain an \I{rdoff subdirectory}\c{rdoff} subdirectory holding
4608 a set of RDOFF utilities: an RDF linker, an \c{RDF} static-library
4609 manager, an RDF file dump utility, and a program which will load and
4610 execute an RDF executable under Linux.
4612 \c{rdf} supports only the \i{standard section names} \i\c{.text},
4613 \i\c{.data} and \i\c{.bss}.
4616 \S{rdflib} Requiring a Library: The \i\c{LIBRARY} Directive
4618 \c{RDOFF} contains a mechanism for an object file to demand a given
4619 library to be linked to the module, either at load time or run time.
4620 This is done by the \c{LIBRARY} directive, which takes one argument
4621 which is the name of the module:
4623 \c library mylib.rdl
4626 \S{rdfmod} Specifying a Module Name: The \i\c{MODULE} Directive
4628 Special \c{RDOFF} header record is used to store the name of the module.
4629 It can be used, for example, by run-time loader to perform dynamic
4630 linking. \c{MODULE} directive takes one argument which is the name
4635 Note that when you statically link modules and tell linker to strip
4636 the symbols from output file, all module names will be stripped too.
4637 To avoid it, you should start module names with \I{$, prefix}\c{$}, like:
4639 \c module $kernel.core
4642 \S{rdfglob} \c{rdf} Extensions to the \c{GLOBAL} directive\I{GLOBAL,
4645 \c{RDOFF} global symbols can contain additional information needed by
4646 the static linker. You can mark a global symbol as exported, thus
4647 telling the linker do not strip it from target executable or library
4648 file. Like in \c{ELF}, you can also specify whether an exported symbol
4649 is a procedure (function) or data object.
4651 Suffixing the name with a colon and the word \i\c{export} you make the
4654 \c global sys_open:export
4656 To specify that exported symbol is a procedure (function), you add the
4657 word \i\c{proc} or \i\c{function} after declaration:
4659 \c global sys_open:export proc
4661 Similarly, to specify exported data object, add the word \i\c{data}
4662 or \i\c{object} to the directive:
4664 \c global kernel_ticks:export data
4667 \S{rdfimpt} \c{rdf} Extensions to the \c{EXTERN} directive\I{EXTERN,
4670 By default the \c{EXTERN} directive in \c{RDOFF} declares a "pure external"
4671 symbol (i.e. the static linker will complain if such a symbol is not resolved).
4672 To declare an "imported" symbol, which must be resolved later during a dynamic
4673 linking phase, \c{RDOFF} offers an additional \c{import} modifier. As in
4674 \c{GLOBAL}, you can also specify whether an imported symbol is a procedure
4675 (function) or data object. For example:
4678 \c extern _open:import
4679 \c extern _printf:import proc
4680 \c extern _errno:import data
4682 Here the directive \c{LIBRARY} is also included, which gives the dynamic linker
4683 a hint as to where to find requested symbols.
4686 \H{dbgfmt} \i\c{dbg}: Debugging Format
4688 The \c{dbg} output format is not built into NASM in the default
4689 configuration. If you are building your own NASM executable from the
4690 sources, you can define \i\c{OF_DBG} in \c{outform.h} or on the
4691 compiler command line, and obtain the \c{dbg} output format.
4693 The \c{dbg} format does not output an object file as such; instead,
4694 it outputs a text file which contains a complete list of all the
4695 transactions between the main body of NASM and the output-format
4696 back end module. It is primarily intended to aid people who want to
4697 write their own output drivers, so that they can get a clearer idea
4698 of the various requests the main program makes of the output driver,
4699 and in what order they happen.
4701 For simple files, one can easily use the \c{dbg} format like this:
4703 \c nasm -f dbg filename.asm
4705 which will generate a diagnostic file called \c{filename.dbg}.
4706 However, this will not work well on files which were designed for a
4707 different object format, because each object format defines its own
4708 macros (usually user-level forms of directives), and those macros
4709 will not be defined in the \c{dbg} format. Therefore it can be
4710 useful to run NASM twice, in order to do the preprocessing with the
4711 native object format selected:
4713 \c nasm -e -f rdf -o rdfprog.i rdfprog.asm
4714 \c nasm -a -f dbg rdfprog.i
4716 This preprocesses \c{rdfprog.asm} into \c{rdfprog.i}, keeping the
4717 \c{rdf} object format selected in order to make sure RDF special
4718 directives are converted into primitive form correctly. Then the
4719 preprocessed source is fed through the \c{dbg} format to generate
4720 the final diagnostic output.
4722 This workaround will still typically not work for programs intended
4723 for \c{obj} format, because the \c{obj} \c{SEGMENT} and \c{GROUP}
4724 directives have side effects of defining the segment and group names
4725 as symbols; \c{dbg} will not do this, so the program will not
4726 assemble. You will have to work around that by defining the symbols
4727 yourself (using \c{EXTERN}, for example) if you really need to get a
4728 \c{dbg} trace of an \c{obj}-specific source file.
4730 \c{dbg} accepts any section name and any directives at all, and logs
4731 them all to its output file.
4734 \C{16bit} Writing 16-bit Code (DOS, Windows 3/3.1)
4736 This chapter attempts to cover some of the common issues encountered
4737 when writing 16-bit code to run under \c{MS-DOS} or \c{Windows 3.x}. It
4738 covers how to link programs to produce \c{.EXE} or \c{.COM} files,
4739 how to write \c{.SYS} device drivers, and how to interface assembly
4740 language code with 16-bit C compilers and with Borland Pascal.
4743 \H{exefiles} Producing \i\c{.EXE} Files
4745 Any large program written under DOS needs to be built as a \c{.EXE}
4746 file: only \c{.EXE} files have the necessary internal structure
4747 required to span more than one 64K segment. \i{Windows} programs,
4748 also, have to be built as \c{.EXE} files, since Windows does not
4749 support the \c{.COM} format.
4751 In general, you generate \c{.EXE} files by using the \c{obj} output
4752 format to produce one or more \i\c{.OBJ} files, and then linking
4753 them together using a linker. However, NASM also supports the direct
4754 generation of simple DOS \c{.EXE} files using the \c{bin} output
4755 format (by using \c{DB} and \c{DW} to construct the \c{.EXE} file
4756 header), and a macro package is supplied to do this. Thanks to
4757 Yann Guidon for contributing the code for this.
4759 NASM may also support \c{.EXE} natively as another output format in
4763 \S{objexe} Using the \c{obj} Format To Generate \c{.EXE} Files
4765 This section describes the usual method of generating \c{.EXE} files
4766 by linking \c{.OBJ} files together.
4768 Most 16-bit programming language packages come with a suitable
4769 linker; if you have none of these, there is a free linker called
4770 \i{VAL}\I{linker, free}, available in \c{LZH} archive format from
4771 \W{ftp://x2ftp.oulu.fi/pub/msdos/programming/lang/}\i\c{x2ftp.oulu.fi}.
4772 An LZH archiver can be found at
4773 \W{ftp://ftp.simtel.net/pub/simtelnet/msdos/arcers}\i\c{ftp.simtel.net}.
4774 There is another `free' linker (though this one doesn't come with
4775 sources) called \i{FREELINK}, available from
4776 \W{http://www.pcorner.com/tpc/old/3-101.html}\i\c{www.pcorner.com}.
4777 A third, \i\c{djlink}, written by DJ Delorie, is available at
4778 \W{http://www.delorie.com/djgpp/16bit/djlink/}\i\c{www.delorie.com}.
4779 A fourth linker, \i\c{ALINK}, written by Anthony A.J. Williams, is
4780 available at \W{http://alink.sourceforge.net}\i\c{alink.sourceforge.net}.
4782 When linking several \c{.OBJ} files into a \c{.EXE} file, you should
4783 ensure that exactly one of them has a start point defined (using the
4784 \I{program entry point}\i\c{..start} special symbol defined by the
4785 \c{obj} format: see \k{dotdotstart}). If no module defines a start
4786 point, the linker will not know what value to give the entry-point
4787 field in the output file header; if more than one defines a start
4788 point, the linker will not know \e{which} value to use.
4790 An example of a NASM source file which can be assembled to a
4791 \c{.OBJ} file and linked on its own to a \c{.EXE} is given here. It
4792 demonstrates the basic principles of defining a stack, initialising
4793 the segment registers, and declaring a start point. This file is
4794 also provided in the \I{test subdirectory}\c{test} subdirectory of
4795 the NASM archives, under the name \c{objexe.asm}.
4806 This initial piece of code sets up \c{DS} to point to the data
4807 segment, and initializes \c{SS} and \c{SP} to point to the top of
4808 the provided stack. Notice that interrupts are implicitly disabled
4809 for one instruction after a move into \c{SS}, precisely for this
4810 situation, so that there's no chance of an interrupt occurring
4811 between the loads of \c{SS} and \c{SP} and not having a stack to
4814 Note also that the special symbol \c{..start} is defined at the
4815 beginning of this code, which means that will be the entry point
4816 into the resulting executable file.
4822 The above is the main program: load \c{DS:DX} with a pointer to the
4823 greeting message (\c{hello} is implicitly relative to the segment
4824 \c{data}, which was loaded into \c{DS} in the setup code, so the
4825 full pointer is valid), and call the DOS print-string function.
4830 This terminates the program using another DOS system call.
4834 \c hello: db 'hello, world', 13, 10, '$'
4836 The data segment contains the string we want to display.
4838 \c segment stack stack
4842 The above code declares a stack segment containing 64 bytes of
4843 uninitialized stack space, and points \c{stacktop} at the top of it.
4844 The directive \c{segment stack stack} defines a segment \e{called}
4845 \c{stack}, and also of \e{type} \c{STACK}. The latter is not
4846 necessary to the correct running of the program, but linkers are
4847 likely to issue warnings or errors if your program has no segment of
4850 The above file, when assembled into a \c{.OBJ} file, will link on
4851 its own to a valid \c{.EXE} file, which when run will print `hello,
4852 world' and then exit.
4855 \S{binexe} Using the \c{bin} Format To Generate \c{.EXE} Files
4857 The \c{.EXE} file format is simple enough that it's possible to
4858 build a \c{.EXE} file by writing a pure-binary program and sticking
4859 a 32-byte header on the front. This header is simple enough that it
4860 can be generated using \c{DB} and \c{DW} commands by NASM itself, so
4861 that you can use the \c{bin} output format to directly generate
4864 Included in the NASM archives, in the \I{misc subdirectory}\c{misc}
4865 subdirectory, is a file \i\c{exebin.mac} of macros. It defines three
4866 macros: \i\c{EXE_begin}, \i\c{EXE_stack} and \i\c{EXE_end}.
4868 To produce a \c{.EXE} file using this method, you should start by
4869 using \c{%include} to load the \c{exebin.mac} macro package into
4870 your source file. You should then issue the \c{EXE_begin} macro call
4871 (which takes no arguments) to generate the file header data. Then
4872 write code as normal for the \c{bin} format - you can use all three
4873 standard sections \c{.text}, \c{.data} and \c{.bss}. At the end of
4874 the file you should call the \c{EXE_end} macro (again, no arguments),
4875 which defines some symbols to mark section sizes, and these symbols
4876 are referred to in the header code generated by \c{EXE_begin}.
4878 In this model, the code you end up writing starts at \c{0x100}, just
4879 like a \c{.COM} file - in fact, if you strip off the 32-byte header
4880 from the resulting \c{.EXE} file, you will have a valid \c{.COM}
4881 program. All the segment bases are the same, so you are limited to a
4882 64K program, again just like a \c{.COM} file. Note that an \c{ORG}
4883 directive is issued by the \c{EXE_begin} macro, so you should not
4884 explicitly issue one of your own.
4886 You can't directly refer to your segment base value, unfortunately,
4887 since this would require a relocation in the header, and things
4888 would get a lot more complicated. So you should get your segment
4889 base by copying it out of \c{CS} instead.
4891 On entry to your \c{.EXE} file, \c{SS:SP} are already set up to
4892 point to the top of a 2Kb stack. You can adjust the default stack
4893 size of 2Kb by calling the \c{EXE_stack} macro. For example, to
4894 change the stack size of your program to 64 bytes, you would call
4897 A sample program which generates a \c{.EXE} file in this way is
4898 given in the \c{test} subdirectory of the NASM archive, as
4902 \H{comfiles} Producing \i\c{.COM} Files
4904 While large DOS programs must be written as \c{.EXE} files, small
4905 ones are often better written as \c{.COM} files. \c{.COM} files are
4906 pure binary, and therefore most easily produced using the \c{bin}
4910 \S{combinfmt} Using the \c{bin} Format To Generate \c{.COM} Files
4912 \c{.COM} files expect to be loaded at offset \c{100h} into their
4913 segment (though the segment may change). Execution then begins at
4914 \I\c{ORG}\c{100h}, i.e. right at the start of the program. So to
4915 write a \c{.COM} program, you would create a source file looking
4923 \c ; put your code here
4927 \c ; put data items here
4931 \c ; put uninitialized data here
4933 The \c{bin} format puts the \c{.text} section first in the file, so
4934 you can declare data or BSS items before beginning to write code if
4935 you want to and the code will still end up at the front of the file
4938 The BSS (uninitialized data) section does not take up space in the
4939 \c{.COM} file itself: instead, addresses of BSS items are resolved
4940 to point at space beyond the end of the file, on the grounds that
4941 this will be free memory when the program is run. Therefore you
4942 should not rely on your BSS being initialized to all zeros when you
4945 To assemble the above program, you should use a command line like
4947 \c nasm myprog.asm -fbin -o myprog.com
4949 The \c{bin} format would produce a file called \c{myprog} if no
4950 explicit output file name were specified, so you have to override it
4951 and give the desired file name.
4954 \S{comobjfmt} Using the \c{obj} Format To Generate \c{.COM} Files
4956 If you are writing a \c{.COM} program as more than one module, you
4957 may wish to assemble several \c{.OBJ} files and link them together
4958 into a \c{.COM} program. You can do this, provided you have a linker
4959 capable of outputting \c{.COM} files directly (\i{TLINK} does this),
4960 or alternatively a converter program such as \i\c{EXE2BIN} to
4961 transform the \c{.EXE} file output from the linker into a \c{.COM}
4964 If you do this, you need to take care of several things:
4966 \b The first object file containing code should start its code
4967 segment with a line like \c{RESB 100h}. This is to ensure that the
4968 code begins at offset \c{100h} relative to the beginning of the code
4969 segment, so that the linker or converter program does not have to
4970 adjust address references within the file when generating the
4971 \c{.COM} file. Other assemblers use an \i\c{ORG} directive for this
4972 purpose, but \c{ORG} in NASM is a format-specific directive to the
4973 \c{bin} output format, and does not mean the same thing as it does
4974 in MASM-compatible assemblers.
4976 \b You don't need to define a stack segment.
4978 \b All your segments should be in the same group, so that every time
4979 your code or data references a symbol offset, all offsets are
4980 relative to the same segment base. This is because, when a \c{.COM}
4981 file is loaded, all the segment registers contain the same value.
4984 \H{sysfiles} Producing \i\c{.SYS} Files
4986 \i{MS-DOS device drivers} - \c{.SYS} files - are pure binary files,
4987 similar to \c{.COM} files, except that they start at origin zero
4988 rather than \c{100h}. Therefore, if you are writing a device driver
4989 using the \c{bin} format, you do not need the \c{ORG} directive,
4990 since the default origin for \c{bin} is zero. Similarly, if you are
4991 using \c{obj}, you do not need the \c{RESB 100h} at the start of
4994 \c{.SYS} files start with a header structure, containing pointers to
4995 the various routines inside the driver which do the work. This
4996 structure should be defined at the start of the code segment, even
4997 though it is not actually code.
4999 For more information on the format of \c{.SYS} files, and the data
5000 which has to go in the header structure, a list of books is given in
5001 the Frequently Asked Questions list for the newsgroup
5002 \W{news:comp.os.msdos.programmer}\i\c{comp.os.msdos.programmer}.
5005 \H{16c} Interfacing to 16-bit C Programs
5007 This section covers the basics of writing assembly routines that
5008 call, or are called from, C programs. To do this, you would
5009 typically write an assembly module as a \c{.OBJ} file, and link it
5010 with your C modules to produce a \i{mixed-language program}.
5013 \S{16cunder} External Symbol Names
5015 \I{C symbol names}\I{underscore, in C symbols}C compilers have the
5016 convention that the names of all global symbols (functions or data)
5017 they define are formed by prefixing an underscore to the name as it
5018 appears in the C program. So, for example, the function a C
5019 programmer thinks of as \c{printf} appears to an assembly language
5020 programmer as \c{_printf}. This means that in your assembly
5021 programs, you can define symbols without a leading underscore, and
5022 not have to worry about name clashes with C symbols.
5024 If you find the underscores inconvenient, you can define macros to
5025 replace the \c{GLOBAL} and \c{EXTERN} directives as follows:
5041 (These forms of the macros only take one argument at a time; a
5042 \c{%rep} construct could solve this.)
5044 If you then declare an external like this:
5048 then the macro will expand it as
5051 \c %define printf _printf
5053 Thereafter, you can reference \c{printf} as if it was a symbol, and
5054 the preprocessor will put the leading underscore on where necessary.
5056 The \c{cglobal} macro works similarly. You must use \c{cglobal}
5057 before defining the symbol in question, but you would have had to do
5058 that anyway if you used \c{GLOBAL}.
5060 Also see \k{opt-pfix}.
5062 \S{16cmodels} \i{Memory Models}
5064 NASM contains no mechanism to support the various C memory models
5065 directly; you have to keep track yourself of which one you are
5066 writing for. This means you have to keep track of the following
5069 \b In models using a single code segment (tiny, small and compact),
5070 functions are near. This means that function pointers, when stored
5071 in data segments or pushed on the stack as function arguments, are
5072 16 bits long and contain only an offset field (the \c{CS} register
5073 never changes its value, and always gives the segment part of the
5074 full function address), and that functions are called using ordinary
5075 near \c{CALL} instructions and return using \c{RETN} (which, in
5076 NASM, is synonymous with \c{RET} anyway). This means both that you
5077 should write your own routines to return with \c{RETN}, and that you
5078 should call external C routines with near \c{CALL} instructions.
5080 \b In models using more than one code segment (medium, large and
5081 huge), functions are far. This means that function pointers are 32
5082 bits long (consisting of a 16-bit offset followed by a 16-bit
5083 segment), and that functions are called using \c{CALL FAR} (or
5084 \c{CALL seg:offset}) and return using \c{RETF}. Again, you should
5085 therefore write your own routines to return with \c{RETF} and use
5086 \c{CALL FAR} to call external routines.
5088 \b In models using a single data segment (tiny, small and medium),
5089 data pointers are 16 bits long, containing only an offset field (the
5090 \c{DS} register doesn't change its value, and always gives the
5091 segment part of the full data item address).
5093 \b In models using more than one data segment (compact, large and
5094 huge), data pointers are 32 bits long, consisting of a 16-bit offset
5095 followed by a 16-bit segment. You should still be careful not to
5096 modify \c{DS} in your routines without restoring it afterwards, but
5097 \c{ES} is free for you to use to access the contents of 32-bit data
5098 pointers you are passed.
5100 \b The huge memory model allows single data items to exceed 64K in
5101 size. In all other memory models, you can access the whole of a data
5102 item just by doing arithmetic on the offset field of the pointer you
5103 are given, whether a segment field is present or not; in huge model,
5104 you have to be more careful of your pointer arithmetic.
5106 \b In most memory models, there is a \e{default} data segment, whose
5107 segment address is kept in \c{DS} throughout the program. This data
5108 segment is typically the same segment as the stack, kept in \c{SS},
5109 so that functions' local variables (which are stored on the stack)
5110 and global data items can both be accessed easily without changing
5111 \c{DS}. Particularly large data items are typically stored in other
5112 segments. However, some memory models (though not the standard
5113 ones, usually) allow the assumption that \c{SS} and \c{DS} hold the
5114 same value to be removed. Be careful about functions' local
5115 variables in this latter case.
5117 In models with a single code segment, the segment is called
5118 \i\c{_TEXT}, so your code segment must also go by this name in order
5119 to be linked into the same place as the main code segment. In models
5120 with a single data segment, or with a default data segment, it is
5124 \S{16cfunc} Function Definitions and Function Calls
5126 \I{functions, C calling convention}The \i{C calling convention} in
5127 16-bit programs is as follows. In the following description, the
5128 words \e{caller} and \e{callee} are used to denote the function
5129 doing the calling and the function which gets called.
5131 \b The caller pushes the function's parameters on the stack, one
5132 after another, in reverse order (right to left, so that the first
5133 argument specified to the function is pushed last).
5135 \b The caller then executes a \c{CALL} instruction to pass control
5136 to the callee. This \c{CALL} is either near or far depending on the
5139 \b The callee receives control, and typically (although this is not
5140 actually necessary, in functions which do not need to access their
5141 parameters) starts by saving the value of \c{SP} in \c{BP} so as to
5142 be able to use \c{BP} as a base pointer to find its parameters on
5143 the stack. However, the caller was probably doing this too, so part
5144 of the calling convention states that \c{BP} must be preserved by
5145 any C function. Hence the callee, if it is going to set up \c{BP} as
5146 a \i\e{frame pointer}, must push the previous value first.
5148 \b The callee may then access its parameters relative to \c{BP}.
5149 The word at \c{[BP]} holds the previous value of \c{BP} as it was
5150 pushed; the next word, at \c{[BP+2]}, holds the offset part of the
5151 return address, pushed implicitly by \c{CALL}. In a small-model
5152 (near) function, the parameters start after that, at \c{[BP+4]}; in
5153 a large-model (far) function, the segment part of the return address
5154 lives at \c{[BP+4]}, and the parameters begin at \c{[BP+6]}. The
5155 leftmost parameter of the function, since it was pushed last, is
5156 accessible at this offset from \c{BP}; the others follow, at
5157 successively greater offsets. Thus, in a function such as \c{printf}
5158 which takes a variable number of parameters, the pushing of the
5159 parameters in reverse order means that the function knows where to
5160 find its first parameter, which tells it the number and type of the
5163 \b The callee may also wish to decrease \c{SP} further, so as to
5164 allocate space on the stack for local variables, which will then be
5165 accessible at negative offsets from \c{BP}.
5167 \b The callee, if it wishes to return a value to the caller, should
5168 leave the value in \c{AL}, \c{AX} or \c{DX:AX} depending on the size
5169 of the value. Floating-point results are sometimes (depending on the
5170 compiler) returned in \c{ST0}.
5172 \b Once the callee has finished processing, it restores \c{SP} from
5173 \c{BP} if it had allocated local stack space, then pops the previous
5174 value of \c{BP}, and returns via \c{RETN} or \c{RETF} depending on
5177 \b When the caller regains control from the callee, the function
5178 parameters are still on the stack, so it typically adds an immediate
5179 constant to \c{SP} to remove them (instead of executing a number of
5180 slow \c{POP} instructions). Thus, if a function is accidentally
5181 called with the wrong number of parameters due to a prototype
5182 mismatch, the stack will still be returned to a sensible state since
5183 the caller, which \e{knows} how many parameters it pushed, does the
5186 It is instructive to compare this calling convention with that for
5187 Pascal programs (described in \k{16bpfunc}). Pascal has a simpler
5188 convention, since no functions have variable numbers of parameters.
5189 Therefore the callee knows how many parameters it should have been
5190 passed, and is able to deallocate them from the stack itself by
5191 passing an immediate argument to the \c{RET} or \c{RETF}
5192 instruction, so the caller does not have to do it. Also, the
5193 parameters are pushed in left-to-right order, not right-to-left,
5194 which means that a compiler can give better guarantees about
5195 sequence points without performance suffering.
5197 Thus, you would define a function in C style in the following way.
5198 The following example is for small model:
5205 \c sub sp,0x40 ; 64 bytes of local stack space
5206 \c mov bx,[bp+4] ; first parameter to function
5210 \c mov sp,bp ; undo "sub sp,0x40" above
5214 For a large-model function, you would replace \c{RET} by \c{RETF},
5215 and look for the first parameter at \c{[BP+6]} instead of
5216 \c{[BP+4]}. Of course, if one of the parameters is a pointer, then
5217 the offsets of \e{subsequent} parameters will change depending on
5218 the memory model as well: far pointers take up four bytes on the
5219 stack when passed as a parameter, whereas near pointers take up two.
5221 At the other end of the process, to call a C function from your
5222 assembly code, you would do something like this:
5226 \c ; and then, further down...
5228 \c push word [myint] ; one of my integer variables
5229 \c push word mystring ; pointer into my data segment
5231 \c add sp,byte 4 ; `byte' saves space
5233 \c ; then those data items...
5238 \c mystring db 'This number -> %d <- should be 1234',10,0
5240 This piece of code is the small-model assembly equivalent of the C
5243 \c int myint = 1234;
5244 \c printf("This number -> %d <- should be 1234\n", myint);
5246 In large model, the function-call code might look more like this. In
5247 this example, it is assumed that \c{DS} already holds the segment
5248 base of the segment \c{_DATA}. If not, you would have to initialize
5251 \c push word [myint]
5252 \c push word seg mystring ; Now push the segment, and...
5253 \c push word mystring ; ... offset of "mystring"
5257 The integer value still takes up one word on the stack, since large
5258 model does not affect the size of the \c{int} data type. The first
5259 argument (pushed last) to \c{printf}, however, is a data pointer,
5260 and therefore has to contain a segment and offset part. The segment
5261 should be stored second in memory, and therefore must be pushed
5262 first. (Of course, \c{PUSH DS} would have been a shorter instruction
5263 than \c{PUSH WORD SEG mystring}, if \c{DS} was set up as the above
5264 example assumed.) Then the actual call becomes a far call, since
5265 functions expect far calls in large model; and \c{SP} has to be
5266 increased by 6 rather than 4 afterwards to make up for the extra
5270 \S{16cdata} Accessing Data Items
5272 To get at the contents of C variables, or to declare variables which
5273 C can access, you need only declare the names as \c{GLOBAL} or
5274 \c{EXTERN}. (Again, the names require leading underscores, as stated
5275 in \k{16cunder}.) Thus, a C variable declared as \c{int i} can be
5276 accessed from assembler as
5282 And to declare your own integer variable which C programs can access
5283 as \c{extern int j}, you do this (making sure you are assembling in
5284 the \c{_DATA} segment, if necessary):
5290 To access a C array, you need to know the size of the components of
5291 the array. For example, \c{int} variables are two bytes long, so if
5292 a C program declares an array as \c{int a[10]}, you can access
5293 \c{a[3]} by coding \c{mov ax,[_a+6]}. (The byte offset 6 is obtained
5294 by multiplying the desired array index, 3, by the size of the array
5295 element, 2.) The sizes of the C base types in 16-bit compilers are:
5296 1 for \c{char}, 2 for \c{short} and \c{int}, 4 for \c{long} and
5297 \c{float}, and 8 for \c{double}.
5299 To access a C \i{data structure}, you need to know the offset from
5300 the base of the structure to the field you are interested in. You
5301 can either do this by converting the C structure definition into a
5302 NASM structure definition (using \i\c{STRUC}), or by calculating the
5303 one offset and using just that.
5305 To do either of these, you should read your C compiler's manual to
5306 find out how it organizes data structures. NASM gives no special
5307 alignment to structure members in its own \c{STRUC} macro, so you
5308 have to specify alignment yourself if the C compiler generates it.
5309 Typically, you might find that a structure like
5316 might be four bytes long rather than three, since the \c{int} field
5317 would be aligned to a two-byte boundary. However, this sort of
5318 feature tends to be a configurable option in the C compiler, either
5319 using command-line options or \c{#pragma} lines, so you have to find
5320 out how your own compiler does it.
5323 \S{16cmacro} \i\c{c16.mac}: Helper Macros for the 16-bit C Interface
5325 Included in the NASM archives, in the \I{misc subdirectory}\c{misc}
5326 directory, is a file \c{c16.mac} of macros. It defines three macros:
5327 \i\c{proc}, \i\c{arg} and \i\c{endproc}. These are intended to be
5328 used for C-style procedure definitions, and they automate a lot of
5329 the work involved in keeping track of the calling convention.
5331 (An alternative, TASM compatible form of \c{arg} is also now built
5332 into NASM's preprocessor. See \k{tasmcompat} for details.)
5334 An example of an assembly function using the macro set is given
5341 \c mov ax,[bp + %$i]
5342 \c mov bx,[bp + %$j]
5347 This defines \c{_nearproc} to be a procedure taking two arguments,
5348 the first (\c{i}) an integer and the second (\c{j}) a pointer to an
5349 integer. It returns \c{i + *j}.
5351 Note that the \c{arg} macro has an \c{EQU} as the first line of its
5352 expansion, and since the label before the macro call gets prepended
5353 to the first line of the expanded macro, the \c{EQU} works, defining
5354 \c{%$i} to be an offset from \c{BP}. A context-local variable is
5355 used, local to the context pushed by the \c{proc} macro and popped
5356 by the \c{endproc} macro, so that the same argument name can be used
5357 in later procedures. Of course, you don't \e{have} to do that.
5359 The macro set produces code for near functions (tiny, small and
5360 compact-model code) by default. You can have it generate far
5361 functions (medium, large and huge-model code) by means of coding
5362 \I\c{FARCODE}\c{%define FARCODE}. This changes the kind of return
5363 instruction generated by \c{endproc}, and also changes the starting
5364 point for the argument offsets. The macro set contains no intrinsic
5365 dependency on whether data pointers are far or not.
5367 \c{arg} can take an optional parameter, giving the size of the
5368 argument. If no size is given, 2 is assumed, since it is likely that
5369 many function parameters will be of type \c{int}.
5371 The large-model equivalent of the above function would look like this:
5379 \c mov ax,[bp + %$i]
5380 \c mov bx,[bp + %$j]
5381 \c mov es,[bp + %$j + 2]
5386 This makes use of the argument to the \c{arg} macro to define a
5387 parameter of size 4, because \c{j} is now a far pointer. When we
5388 load from \c{j}, we must load a segment and an offset.
5391 \H{16bp} Interfacing to \i{Borland Pascal} Programs
5393 Interfacing to Borland Pascal programs is similar in concept to
5394 interfacing to 16-bit C programs. The differences are:
5396 \b The leading underscore required for interfacing to C programs is
5397 not required for Pascal.
5399 \b The memory model is always large: functions are far, data
5400 pointers are far, and no data item can be more than 64K long.
5401 (Actually, some functions are near, but only those functions that
5402 are local to a Pascal unit and never called from outside it. All
5403 assembly functions that Pascal calls, and all Pascal functions that
5404 assembly routines are able to call, are far.) However, all static
5405 data declared in a Pascal program goes into the default data
5406 segment, which is the one whose segment address will be in \c{DS}
5407 when control is passed to your assembly code. The only things that
5408 do not live in the default data segment are local variables (they
5409 live in the stack segment) and dynamically allocated variables. All
5410 data \e{pointers}, however, are far.
5412 \b The function calling convention is different - described below.
5414 \b Some data types, such as strings, are stored differently.
5416 \b There are restrictions on the segment names you are allowed to
5417 use - Borland Pascal will ignore code or data declared in a segment
5418 it doesn't like the name of. The restrictions are described below.
5421 \S{16bpfunc} The Pascal Calling Convention
5423 \I{functions, Pascal calling convention}\I{Pascal calling
5424 convention}The 16-bit Pascal calling convention is as follows. In
5425 the following description, the words \e{caller} and \e{callee} are
5426 used to denote the function doing the calling and the function which
5429 \b The caller pushes the function's parameters on the stack, one
5430 after another, in normal order (left to right, so that the first
5431 argument specified to the function is pushed first).
5433 \b The caller then executes a far \c{CALL} instruction to pass
5434 control to the callee.
5436 \b The callee receives control, and typically (although this is not
5437 actually necessary, in functions which do not need to access their
5438 parameters) starts by saving the value of \c{SP} in \c{BP} so as to
5439 be able to use \c{BP} as a base pointer to find its parameters on
5440 the stack. However, the caller was probably doing this too, so part
5441 of the calling convention states that \c{BP} must be preserved by
5442 any function. Hence the callee, if it is going to set up \c{BP} as a
5443 \i{frame pointer}, must push the previous value first.
5445 \b The callee may then access its parameters relative to \c{BP}.
5446 The word at \c{[BP]} holds the previous value of \c{BP} as it was
5447 pushed. The next word, at \c{[BP+2]}, holds the offset part of the
5448 return address, and the next one at \c{[BP+4]} the segment part. The
5449 parameters begin at \c{[BP+6]}. The rightmost parameter of the
5450 function, since it was pushed last, is accessible at this offset
5451 from \c{BP}; the others follow, at successively greater offsets.
5453 \b The callee may also wish to decrease \c{SP} further, so as to
5454 allocate space on the stack for local variables, which will then be
5455 accessible at negative offsets from \c{BP}.
5457 \b The callee, if it wishes to return a value to the caller, should
5458 leave the value in \c{AL}, \c{AX} or \c{DX:AX} depending on the size
5459 of the value. Floating-point results are returned in \c{ST0}.
5460 Results of type \c{Real} (Borland's own custom floating-point data
5461 type, not handled directly by the FPU) are returned in \c{DX:BX:AX}.
5462 To return a result of type \c{String}, the caller pushes a pointer
5463 to a temporary string before pushing the parameters, and the callee
5464 places the returned string value at that location. The pointer is
5465 not a parameter, and should not be removed from the stack by the
5466 \c{RETF} instruction.
5468 \b Once the callee has finished processing, it restores \c{SP} from
5469 \c{BP} if it had allocated local stack space, then pops the previous
5470 value of \c{BP}, and returns via \c{RETF}. It uses the form of
5471 \c{RETF} with an immediate parameter, giving the number of bytes
5472 taken up by the parameters on the stack. This causes the parameters
5473 to be removed from the stack as a side effect of the return
5476 \b When the caller regains control from the callee, the function
5477 parameters have already been removed from the stack, so it needs to
5480 Thus, you would define a function in Pascal style, taking two
5481 \c{Integer}-type parameters, in the following way:
5487 \c sub sp,0x40 ; 64 bytes of local stack space
5488 \c mov bx,[bp+8] ; first parameter to function
5489 \c mov bx,[bp+6] ; second parameter to function
5493 \c mov sp,bp ; undo "sub sp,0x40" above
5495 \c retf 4 ; total size of params is 4
5497 At the other end of the process, to call a Pascal function from your
5498 assembly code, you would do something like this:
5502 \c ; and then, further down...
5504 \c push word seg mystring ; Now push the segment, and...
5505 \c push word mystring ; ... offset of "mystring"
5506 \c push word [myint] ; one of my variables
5507 \c call far SomeFunc
5509 This is equivalent to the Pascal code
5511 \c procedure SomeFunc(String: PChar; Int: Integer);
5512 \c SomeFunc(@mystring, myint);
5515 \S{16bpseg} Borland Pascal \I{segment names, Borland Pascal}Segment
5518 Since Borland Pascal's internal unit file format is completely
5519 different from \c{OBJ}, it only makes a very sketchy job of actually
5520 reading and understanding the various information contained in a
5521 real \c{OBJ} file when it links that in. Therefore an object file
5522 intended to be linked to a Pascal program must obey a number of
5525 \b Procedures and functions must be in a segment whose name is
5526 either \c{CODE}, \c{CSEG}, or something ending in \c{_TEXT}.
5528 \b initialized data must be in a segment whose name is either
5529 \c{CONST} or something ending in \c{_DATA}.
5531 \b Uninitialized data must be in a segment whose name is either
5532 \c{DATA}, \c{DSEG}, or something ending in \c{_BSS}.
5534 \b Any other segments in the object file are completely ignored.
5535 \c{GROUP} directives and segment attributes are also ignored.
5538 \S{16bpmacro} Using \i\c{c16.mac} With Pascal Programs
5540 The \c{c16.mac} macro package, described in \k{16cmacro}, can also
5541 be used to simplify writing functions to be called from Pascal
5542 programs, if you code \I\c{PASCAL}\c{%define PASCAL}. This
5543 definition ensures that functions are far (it implies
5544 \i\c{FARCODE}), and also causes procedure return instructions to be
5545 generated with an operand.
5547 Defining \c{PASCAL} does not change the code which calculates the
5548 argument offsets; you must declare your function's arguments in
5549 reverse order. For example:
5557 \c mov ax,[bp + %$i]
5558 \c mov bx,[bp + %$j]
5559 \c mov es,[bp + %$j + 2]
5564 This defines the same routine, conceptually, as the example in
5565 \k{16cmacro}: it defines a function taking two arguments, an integer
5566 and a pointer to an integer, which returns the sum of the integer
5567 and the contents of the pointer. The only difference between this
5568 code and the large-model C version is that \c{PASCAL} is defined
5569 instead of \c{FARCODE}, and that the arguments are declared in
5573 \C{32bit} Writing 32-bit Code (Unix, Win32, DJGPP)
5575 This chapter attempts to cover some of the common issues involved
5576 when writing 32-bit code, to run under \i{Win32} or Unix, or to be
5577 linked with C code generated by a Unix-style C compiler such as
5578 \i{DJGPP}. It covers how to write assembly code to interface with
5579 32-bit C routines, and how to write position-independent code for
5582 Almost all 32-bit code, and in particular all code running under
5583 \c{Win32}, \c{DJGPP} or any of the PC Unix variants, runs in \I{flat
5584 memory model}\e{flat} memory model. This means that the segment registers
5585 and paging have already been set up to give you the same 32-bit 4Gb
5586 address space no matter what segment you work relative to, and that
5587 you should ignore all segment registers completely. When writing
5588 flat-model application code, you never need to use a segment
5589 override or modify any segment register, and the code-section
5590 addresses you pass to \c{CALL} and \c{JMP} live in the same address
5591 space as the data-section addresses you access your variables by and
5592 the stack-section addresses you access local variables and procedure
5593 parameters by. Every address is 32 bits long and contains only an
5597 \H{32c} Interfacing to 32-bit C Programs
5599 A lot of the discussion in \k{16c}, about interfacing to 16-bit C
5600 programs, still applies when working in 32 bits. The absence of
5601 memory models or segmentation worries simplifies things a lot.
5604 \S{32cunder} External Symbol Names
5606 Most 32-bit C compilers share the convention used by 16-bit
5607 compilers, that the names of all global symbols (functions or data)
5608 they define are formed by prefixing an underscore to the name as it
5609 appears in the C program. However, not all of them do: the \c{ELF}
5610 specification states that C symbols do \e{not} have a leading
5611 underscore on their assembly-language names.
5613 The older Linux \c{a.out} C compiler, all \c{Win32} compilers,
5614 \c{DJGPP}, and \c{NetBSD} and \c{FreeBSD}, all use the leading
5615 underscore; for these compilers, the macros \c{cextern} and
5616 \c{cglobal}, as given in \k{16cunder}, will still work. For \c{ELF},
5617 though, the leading underscore should not be used.
5619 See also \k{opt-pfix}.
5621 \S{32cfunc} Function Definitions and Function Calls
5623 \I{functions, C calling convention}The \i{C calling convention}The C
5624 calling convention in 32-bit programs is as follows. In the
5625 following description, the words \e{caller} and \e{callee} are used
5626 to denote the function doing the calling and the function which gets
5629 \b The caller pushes the function's parameters on the stack, one
5630 after another, in reverse order (right to left, so that the first
5631 argument specified to the function is pushed last).
5633 \b The caller then executes a near \c{CALL} instruction to pass
5634 control to the callee.
5636 \b The callee receives control, and typically (although this is not
5637 actually necessary, in functions which do not need to access their
5638 parameters) starts by saving the value of \c{ESP} in \c{EBP} so as
5639 to be able to use \c{EBP} as a base pointer to find its parameters
5640 on the stack. However, the caller was probably doing this too, so
5641 part of the calling convention states that \c{EBP} must be preserved
5642 by any C function. Hence the callee, if it is going to set up
5643 \c{EBP} as a \i{frame pointer}, must push the previous value first.
5645 \b The callee may then access its parameters relative to \c{EBP}.
5646 The doubleword at \c{[EBP]} holds the previous value of \c{EBP} as
5647 it was pushed; the next doubleword, at \c{[EBP+4]}, holds the return
5648 address, pushed implicitly by \c{CALL}. The parameters start after
5649 that, at \c{[EBP+8]}. The leftmost parameter of the function, since
5650 it was pushed last, is accessible at this offset from \c{EBP}; the
5651 others follow, at successively greater offsets. Thus, in a function
5652 such as \c{printf} which takes a variable number of parameters, the
5653 pushing of the parameters in reverse order means that the function
5654 knows where to find its first parameter, which tells it the number
5655 and type of the remaining ones.
5657 \b The callee may also wish to decrease \c{ESP} further, so as to
5658 allocate space on the stack for local variables, which will then be
5659 accessible at negative offsets from \c{EBP}.
5661 \b The callee, if it wishes to return a value to the caller, should
5662 leave the value in \c{AL}, \c{AX} or \c{EAX} depending on the size
5663 of the value. Floating-point results are typically returned in
5666 \b Once the callee has finished processing, it restores \c{ESP} from
5667 \c{EBP} if it had allocated local stack space, then pops the previous
5668 value of \c{EBP}, and returns via \c{RET} (equivalently, \c{RETN}).
5670 \b When the caller regains control from the callee, the function
5671 parameters are still on the stack, so it typically adds an immediate
5672 constant to \c{ESP} to remove them (instead of executing a number of
5673 slow \c{POP} instructions). Thus, if a function is accidentally
5674 called with the wrong number of parameters due to a prototype
5675 mismatch, the stack will still be returned to a sensible state since
5676 the caller, which \e{knows} how many parameters it pushed, does the
5679 There is an alternative calling convention used by Win32 programs
5680 for Windows API calls, and also for functions called \e{by} the
5681 Windows API such as window procedures: they follow what Microsoft
5682 calls the \c{__stdcall} convention. This is slightly closer to the
5683 Pascal convention, in that the callee clears the stack by passing a
5684 parameter to the \c{RET} instruction. However, the parameters are
5685 still pushed in right-to-left order.
5687 Thus, you would define a function in C style in the following way:
5694 \c sub esp,0x40 ; 64 bytes of local stack space
5695 \c mov ebx,[ebp+8] ; first parameter to function
5699 \c leave ; mov esp,ebp / pop ebp
5702 At the other end of the process, to call a C function from your
5703 assembly code, you would do something like this:
5707 \c ; and then, further down...
5709 \c push dword [myint] ; one of my integer variables
5710 \c push dword mystring ; pointer into my data segment
5712 \c add esp,byte 8 ; `byte' saves space
5714 \c ; then those data items...
5719 \c mystring db 'This number -> %d <- should be 1234',10,0
5721 This piece of code is the assembly equivalent of the C code
5723 \c int myint = 1234;
5724 \c printf("This number -> %d <- should be 1234\n", myint);
5727 \S{32cdata} Accessing Data Items
5729 To get at the contents of C variables, or to declare variables which
5730 C can access, you need only declare the names as \c{GLOBAL} or
5731 \c{EXTERN}. (Again, the names require leading underscores, as stated
5732 in \k{32cunder}.) Thus, a C variable declared as \c{int i} can be
5733 accessed from assembler as
5738 And to declare your own integer variable which C programs can access
5739 as \c{extern int j}, you do this (making sure you are assembling in
5740 the \c{_DATA} segment, if necessary):
5745 To access a C array, you need to know the size of the components of
5746 the array. For example, \c{int} variables are four bytes long, so if
5747 a C program declares an array as \c{int a[10]}, you can access
5748 \c{a[3]} by coding \c{mov ax,[_a+12]}. (The byte offset 12 is obtained
5749 by multiplying the desired array index, 3, by the size of the array
5750 element, 4.) The sizes of the C base types in 32-bit compilers are:
5751 1 for \c{char}, 2 for \c{short}, 4 for \c{int}, \c{long} and
5752 \c{float}, and 8 for \c{double}. Pointers, being 32-bit addresses,
5753 are also 4 bytes long.
5755 To access a C \i{data structure}, you need to know the offset from
5756 the base of the structure to the field you are interested in. You
5757 can either do this by converting the C structure definition into a
5758 NASM structure definition (using \c{STRUC}), or by calculating the
5759 one offset and using just that.
5761 To do either of these, you should read your C compiler's manual to
5762 find out how it organizes data structures. NASM gives no special
5763 alignment to structure members in its own \i\c{STRUC} macro, so you
5764 have to specify alignment yourself if the C compiler generates it.
5765 Typically, you might find that a structure like
5772 might be eight bytes long rather than five, since the \c{int} field
5773 would be aligned to a four-byte boundary. However, this sort of
5774 feature is sometimes a configurable option in the C compiler, either
5775 using command-line options or \c{#pragma} lines, so you have to find
5776 out how your own compiler does it.
5779 \S{32cmacro} \i\c{c32.mac}: Helper Macros for the 32-bit C Interface
5781 Included in the NASM archives, in the \I{misc directory}\c{misc}
5782 directory, is a file \c{c32.mac} of macros. It defines three macros:
5783 \i\c{proc}, \i\c{arg} and \i\c{endproc}. These are intended to be
5784 used for C-style procedure definitions, and they automate a lot of
5785 the work involved in keeping track of the calling convention.
5787 An example of an assembly function using the macro set is given
5794 \c mov eax,[ebp + %$i]
5795 \c mov ebx,[ebp + %$j]
5800 This defines \c{_proc32} to be a procedure taking two arguments, the
5801 first (\c{i}) an integer and the second (\c{j}) a pointer to an
5802 integer. It returns \c{i + *j}.
5804 Note that the \c{arg} macro has an \c{EQU} as the first line of its
5805 expansion, and since the label before the macro call gets prepended
5806 to the first line of the expanded macro, the \c{EQU} works, defining
5807 \c{%$i} to be an offset from \c{BP}. A context-local variable is
5808 used, local to the context pushed by the \c{proc} macro and popped
5809 by the \c{endproc} macro, so that the same argument name can be used
5810 in later procedures. Of course, you don't \e{have} to do that.
5812 \c{arg} can take an optional parameter, giving the size of the
5813 argument. If no size is given, 4 is assumed, since it is likely that
5814 many function parameters will be of type \c{int} or pointers.
5817 \H{picdll} Writing NetBSD/FreeBSD/OpenBSD and Linux/ELF \i{Shared
5820 \c{ELF} replaced the older \c{a.out} object file format under Linux
5821 because it contains support for \i{position-independent code}
5822 (\i{PIC}), which makes writing shared libraries much easier. NASM
5823 supports the \c{ELF} position-independent code features, so you can
5824 write Linux \c{ELF} shared libraries in NASM.
5826 \i{NetBSD}, and its close cousins \i{FreeBSD} and \i{OpenBSD}, take
5827 a different approach by hacking PIC support into the \c{a.out}
5828 format. NASM supports this as the \i\c{aoutb} output format, so you
5829 can write \i{BSD} shared libraries in NASM too.
5831 The operating system loads a PIC shared library by memory-mapping
5832 the library file at an arbitrarily chosen point in the address space
5833 of the running process. The contents of the library's code section
5834 must therefore not depend on where it is loaded in memory.
5836 Therefore, you cannot get at your variables by writing code like
5839 \c mov eax,[myvar] ; WRONG
5841 Instead, the linker provides an area of memory called the
5842 \i\e{global offset table}, or \i{GOT}; the GOT is situated at a
5843 constant distance from your library's code, so if you can find out
5844 where your library is loaded (which is typically done using a
5845 \c{CALL} and \c{POP} combination), you can obtain the address of the
5846 GOT, and you can then load the addresses of your variables out of
5847 linker-generated entries in the GOT.
5849 The \e{data} section of a PIC shared library does not have these
5850 restrictions: since the data section is writable, it has to be
5851 copied into memory anyway rather than just paged in from the library
5852 file, so as long as it's being copied it can be relocated too. So
5853 you can put ordinary types of relocation in the data section without
5854 too much worry (but see \k{picglobal} for a caveat).
5857 \S{picgot} Obtaining the Address of the GOT
5859 Each code module in your shared library should define the GOT as an
5862 \c extern _GLOBAL_OFFSET_TABLE_ ; in ELF
5863 \c extern __GLOBAL_OFFSET_TABLE_ ; in BSD a.out
5865 At the beginning of any function in your shared library which plans
5866 to access your data or BSS sections, you must first calculate the
5867 address of the GOT. This is typically done by writing the function
5876 \c add ebx,_GLOBAL_OFFSET_TABLE_+$$-.get_GOT wrt ..gotpc
5878 \c ; the function body comes here
5885 (For BSD, again, the symbol \c{_GLOBAL_OFFSET_TABLE} requires a
5886 second leading underscore.)
5888 The first two lines of this function are simply the standard C
5889 prologue to set up a stack frame, and the last three lines are
5890 standard C function epilogue. The third line, and the fourth to last
5891 line, save and restore the \c{EBX} register, because PIC shared
5892 libraries use this register to store the address of the GOT.
5894 The interesting bit is the \c{CALL} instruction and the following
5895 two lines. The \c{CALL} and \c{POP} combination obtains the address
5896 of the label \c{.get_GOT}, without having to know in advance where
5897 the program was loaded (since the \c{CALL} instruction is encoded
5898 relative to the current position). The \c{ADD} instruction makes use
5899 of one of the special PIC relocation types: \i{GOTPC relocation}.
5900 With the \i\c{WRT ..gotpc} qualifier specified, the symbol
5901 referenced (here \c{_GLOBAL_OFFSET_TABLE_}, the special symbol
5902 assigned to the GOT) is given as an offset from the beginning of the
5903 section. (Actually, \c{ELF} encodes it as the offset from the operand
5904 field of the \c{ADD} instruction, but NASM simplifies this
5905 deliberately, so you do things the same way for both \c{ELF} and
5906 \c{BSD}.) So the instruction then \e{adds} the beginning of the section,
5907 to get the real address of the GOT, and subtracts the value of
5908 \c{.get_GOT} which it knows is in \c{EBX}. Therefore, by the time
5909 that instruction has finished, \c{EBX} contains the address of the GOT.
5911 If you didn't follow that, don't worry: it's never necessary to
5912 obtain the address of the GOT by any other means, so you can put
5913 those three instructions into a macro and safely ignore them:
5920 \c add ebx,_GLOBAL_OFFSET_TABLE_+$$-%%getgot wrt ..gotpc
5924 \S{piclocal} Finding Your Local Data Items
5926 Having got the GOT, you can then use it to obtain the addresses of
5927 your data items. Most variables will reside in the sections you have
5928 declared; they can be accessed using the \I{GOTOFF
5929 relocation}\c{..gotoff} special \I\c{WRT ..gotoff}\c{WRT} type. The
5930 way this works is like this:
5932 \c lea eax,[ebx+myvar wrt ..gotoff]
5934 The expression \c{myvar wrt ..gotoff} is calculated, when the shared
5935 library is linked, to be the offset to the local variable \c{myvar}
5936 from the beginning of the GOT. Therefore, adding it to \c{EBX} as
5937 above will place the real address of \c{myvar} in \c{EAX}.
5939 If you declare variables as \c{GLOBAL} without specifying a size for
5940 them, they are shared between code modules in the library, but do
5941 not get exported from the library to the program that loaded it.
5942 They will still be in your ordinary data and BSS sections, so you
5943 can access them in the same way as local variables, using the above
5944 \c{..gotoff} mechanism.
5946 Note that due to a peculiarity of the way BSD \c{a.out} format
5947 handles this relocation type, there must be at least one non-local
5948 symbol in the same section as the address you're trying to access.
5951 \S{picextern} Finding External and Common Data Items
5953 If your library needs to get at an external variable (external to
5954 the \e{library}, not just to one of the modules within it), you must
5955 use the \I{GOT relocations}\I\c{WRT ..got}\c{..got} type to get at
5956 it. The \c{..got} type, instead of giving you the offset from the
5957 GOT base to the variable, gives you the offset from the GOT base to
5958 a GOT \e{entry} containing the address of the variable. The linker
5959 will set up this GOT entry when it builds the library, and the
5960 dynamic linker will place the correct address in it at load time. So
5961 to obtain the address of an external variable \c{extvar} in \c{EAX},
5964 \c mov eax,[ebx+extvar wrt ..got]
5966 This loads the address of \c{extvar} out of an entry in the GOT. The
5967 linker, when it builds the shared library, collects together every
5968 relocation of type \c{..got}, and builds the GOT so as to ensure it
5969 has every necessary entry present.
5971 Common variables must also be accessed in this way.
5974 \S{picglobal} Exporting Symbols to the Library User
5976 If you want to export symbols to the user of the library, you have
5977 to declare whether they are functions or data, and if they are data,
5978 you have to give the size of the data item. This is because the
5979 dynamic linker has to build \I{PLT}\i{procedure linkage table}
5980 entries for any exported functions, and also moves exported data
5981 items away from the library's data section in which they were
5984 So to export a function to users of the library, you must use
5986 \c global func:function ; declare it as a function
5992 And to export a data item such as an array, you would have to code
5994 \c global array:data array.end-array ; give the size too
5999 Be careful: If you export a variable to the library user, by
6000 declaring it as \c{GLOBAL} and supplying a size, the variable will
6001 end up living in the data section of the main program, rather than
6002 in your library's data section, where you declared it. So you will
6003 have to access your own global variable with the \c{..got} mechanism
6004 rather than \c{..gotoff}, as if it were external (which,
6005 effectively, it has become).
6007 Equally, if you need to store the address of an exported global in
6008 one of your data sections, you can't do it by means of the standard
6011 \c dataptr: dd global_data_item ; WRONG
6013 NASM will interpret this code as an ordinary relocation, in which
6014 \c{global_data_item} is merely an offset from the beginning of the
6015 \c{.data} section (or whatever); so this reference will end up
6016 pointing at your data section instead of at the exported global
6017 which resides elsewhere.
6019 Instead of the above code, then, you must write
6021 \c dataptr: dd global_data_item wrt ..sym
6023 which makes use of the special \c{WRT} type \I\c{WRT ..sym}\c{..sym}
6024 to instruct NASM to search the symbol table for a particular symbol
6025 at that address, rather than just relocating by section base.
6027 Either method will work for functions: referring to one of your
6028 functions by means of
6030 \c funcptr: dd my_function
6032 will give the user the address of the code you wrote, whereas
6034 \c funcptr: dd my_function wrt .sym
6036 will give the address of the procedure linkage table for the
6037 function, which is where the calling program will \e{believe} the
6038 function lives. Either address is a valid way to call the function.
6041 \S{picproc} Calling Procedures Outside the Library
6043 Calling procedures outside your shared library has to be done by
6044 means of a \i\e{procedure linkage table}, or \i{PLT}. The PLT is
6045 placed at a known offset from where the library is loaded, so the
6046 library code can make calls to the PLT in a position-independent
6047 way. Within the PLT there is code to jump to offsets contained in
6048 the GOT, so function calls to other shared libraries or to routines
6049 in the main program can be transparently passed off to their real
6052 To call an external routine, you must use another special PIC
6053 relocation type, \I{PLT relocations}\i\c{WRT ..plt}. This is much
6054 easier than the GOT-based ones: you simply replace calls such as
6055 \c{CALL printf} with the PLT-relative version \c{CALL printf WRT
6059 \S{link} Generating the Library File
6061 Having written some code modules and assembled them to \c{.o} files,
6062 you then generate your shared library with a command such as
6064 \c ld -shared -o library.so module1.o module2.o # for ELF
6065 \c ld -Bshareable -o library.so module1.o module2.o # for BSD
6067 For ELF, if your shared library is going to reside in system
6068 directories such as \c{/usr/lib} or \c{/lib}, it is usually worth
6069 using the \i\c{-soname} flag to the linker, to store the final
6070 library file name, with a version number, into the library:
6072 \c ld -shared -soname library.so.1 -o library.so.1.2 *.o
6074 You would then copy \c{library.so.1.2} into the library directory,
6075 and create \c{library.so.1} as a symbolic link to it.
6078 \C{mixsize} Mixing 16 and 32 Bit Code
6080 This chapter tries to cover some of the issues, largely related to
6081 unusual forms of addressing and jump instructions, encountered when
6082 writing operating system code such as protected-mode initialisation
6083 routines, which require code that operates in mixed segment sizes,
6084 such as code in a 16-bit segment trying to modify data in a 32-bit
6085 one, or jumps between different-size segments.
6088 \H{mixjump} Mixed-Size Jumps\I{jumps, mixed-size}
6090 \I{operating system, writing}\I{writing operating systems}The most
6091 common form of \i{mixed-size instruction} is the one used when
6092 writing a 32-bit OS: having done your setup in 16-bit mode, such as
6093 loading the kernel, you then have to boot it by switching into
6094 protected mode and jumping to the 32-bit kernel start address. In a
6095 fully 32-bit OS, this tends to be the \e{only} mixed-size
6096 instruction you need, since everything before it can be done in pure
6097 16-bit code, and everything after it can be pure 32-bit.
6099 This jump must specify a 48-bit far address, since the target
6100 segment is a 32-bit one. However, it must be assembled in a 16-bit
6101 segment, so just coding, for example,
6103 \c jmp 0x1234:0x56789ABC ; wrong!
6105 will not work, since the offset part of the address will be
6106 truncated to \c{0x9ABC} and the jump will be an ordinary 16-bit far
6109 The Linux kernel setup code gets round the inability of \c{as86} to
6110 generate the required instruction by coding it manually, using
6111 \c{DB} instructions. NASM can go one better than that, by actually
6112 generating the right instruction itself. Here's how to do it right:
6114 \c jmp dword 0x1234:0x56789ABC ; right
6116 \I\c{JMP DWORD}The \c{DWORD} prefix (strictly speaking, it should
6117 come \e{after} the colon, since it is declaring the \e{offset} field
6118 to be a doubleword; but NASM will accept either form, since both are
6119 unambiguous) forces the offset part to be treated as far, in the
6120 assumption that you are deliberately writing a jump from a 16-bit
6121 segment to a 32-bit one.
6123 You can do the reverse operation, jumping from a 32-bit segment to a
6124 16-bit one, by means of the \c{WORD} prefix:
6126 \c jmp word 0x8765:0x4321 ; 32 to 16 bit
6128 If the \c{WORD} prefix is specified in 16-bit mode, or the \c{DWORD}
6129 prefix in 32-bit mode, they will be ignored, since each is
6130 explicitly forcing NASM into a mode it was in anyway.
6133 \H{mixaddr} Addressing Between Different-Size Segments\I{addressing,
6134 mixed-size}\I{mixed-size addressing}
6136 If your OS is mixed 16 and 32-bit, or if you are writing a DOS
6137 extender, you are likely to have to deal with some 16-bit segments
6138 and some 32-bit ones. At some point, you will probably end up
6139 writing code in a 16-bit segment which has to access data in a
6140 32-bit segment, or vice versa.
6142 If the data you are trying to access in a 32-bit segment lies within
6143 the first 64K of the segment, you may be able to get away with using
6144 an ordinary 16-bit addressing operation for the purpose; but sooner
6145 or later, you will want to do 32-bit addressing from 16-bit mode.
6147 The easiest way to do this is to make sure you use a register for
6148 the address, since any effective address containing a 32-bit
6149 register is forced to be a 32-bit address. So you can do
6151 \c mov eax,offset_into_32_bit_segment_specified_by_fs
6152 \c mov dword [fs:eax],0x11223344
6154 This is fine, but slightly cumbersome (since it wastes an
6155 instruction and a register) if you already know the precise offset
6156 you are aiming at. The x86 architecture does allow 32-bit effective
6157 addresses to specify nothing but a 4-byte offset, so why shouldn't
6158 NASM be able to generate the best instruction for the purpose?
6160 It can. As in \k{mixjump}, you need only prefix the address with the
6161 \c{DWORD} keyword, and it will be forced to be a 32-bit address:
6163 \c mov dword [fs:dword my_offset],0x11223344
6165 Also as in \k{mixjump}, NASM is not fussy about whether the
6166 \c{DWORD} prefix comes before or after the segment override, so
6167 arguably a nicer-looking way to code the above instruction is
6169 \c mov dword [dword fs:my_offset],0x11223344
6171 Don't confuse the \c{DWORD} prefix \e{outside} the square brackets,
6172 which controls the size of the data stored at the address, with the
6173 one \c{inside} the square brackets which controls the length of the
6174 address itself. The two can quite easily be different:
6176 \c mov word [dword 0x12345678],0x9ABC
6178 This moves 16 bits of data to an address specified by a 32-bit
6181 You can also specify \c{WORD} or \c{DWORD} prefixes along with the
6182 \c{FAR} prefix to indirect far jumps or calls. For example:
6184 \c call dword far [fs:word 0x4321]
6186 This instruction contains an address specified by a 16-bit offset;
6187 it loads a 48-bit far pointer from that (16-bit segment and 32-bit
6188 offset), and calls that address.
6191 \H{mixother} Other Mixed-Size Instructions
6193 The other way you might want to access data might be using the
6194 string instructions (\c{LODSx}, \c{STOSx} and so on) or the
6195 \c{XLATB} instruction. These instructions, since they take no
6196 parameters, might seem to have no easy way to make them perform
6197 32-bit addressing when assembled in a 16-bit segment.
6199 This is the purpose of NASM's \i\c{a16} and \i\c{a32} prefixes. If
6200 you are coding \c{LODSB} in a 16-bit segment but it is supposed to
6201 be accessing a string in a 32-bit segment, you should load the
6202 desired address into \c{ESI} and then code
6206 The prefix forces the addressing size to 32 bits, meaning that
6207 \c{LODSB} loads from \c{[DS:ESI]} instead of \c{[DS:SI]}. To access
6208 a string in a 16-bit segment when coding in a 32-bit one, the
6209 corresponding \c{a16} prefix can be used.
6211 The \c{a16} and \c{a32} prefixes can be applied to any instruction
6212 in NASM's instruction table, but most of them can generate all the
6213 useful forms without them. The prefixes are necessary only for
6214 instructions with implicit addressing:
6215 \# \c{CMPSx} (\k{insCMPSB}),
6216 \# \c{SCASx} (\k{insSCASB}), \c{LODSx} (\k{insLODSB}), \c{STOSx}
6217 \# (\k{insSTOSB}), \c{MOVSx} (\k{insMOVSB}), \c{INSx} (\k{insINSB}),
6218 \# \c{OUTSx} (\k{insOUTSB}), and \c{XLATB} (\k{insXLATB}).
6219 \c{CMPSx}, \c{SCASx}, \c{LODSx}, \c{STOSx}, \c{MOVSx}, \c{INSx},
6220 \c{OUTSx}, and \c{XLATB}.
6222 various push and pop instructions (\c{PUSHA} and \c{POPF} as well as
6223 the more usual \c{PUSH} and \c{POP}) can accept \c{a16} or \c{a32}
6224 prefixes to force a particular one of \c{SP} or \c{ESP} to be used
6225 as a stack pointer, in case the stack segment in use is a different
6226 size from the code segment.
6228 \c{PUSH} and \c{POP}, when applied to segment registers in 32-bit
6229 mode, also have the slightly odd behaviour that they push and pop 4
6230 bytes at a time, of which the top two are ignored and the bottom two
6231 give the value of the segment register being manipulated. To force
6232 the 16-bit behaviour of segment-register push and pop instructions,
6233 you can use the operand-size prefix \i\c{o16}:
6238 This code saves a doubleword of stack space by fitting two segment
6239 registers into the space which would normally be consumed by pushing
6242 (You can also use the \i\c{o32} prefix to force the 32-bit behaviour
6243 when in 16-bit mode, but this seems less useful.)
6246 \C{64bit} Writing 64-bit Code (Unix, Win64)
6248 This chapter attempts to cover some of the common issues involved when
6249 writing 64-bit code, to run under \i{Win64} or Unix. It covers how to
6250 write assembly code to interface with 64-bit C routines, and how to
6251 write position-independent code for shared libraries.
6253 All 64-bit code uses a flat memory model, since segmentation is not
6254 available in 64-bit mode. The one exception is the \c{FS} and \c{GS}
6255 registers, which still add their bases.
6257 Position independence in 64-bit mode is significantly simpler, since
6258 the processor supports \c{RIP}-relative addressing directly; see the
6259 \c{REL} keyword (\k{effaddr}). On most 64-bit platforms, it is
6260 probably desirable to make that the default, using the directive
6261 \c{DEFAULT REL} (\k{default}).
6263 64-bit programming is relatively similar to 32-bit programming, but
6264 of course pointers are 64 bits long; additionally, all existing
6265 platforms pass arguments in registers rather than on the stack.
6266 Furthermore, 64-bit platforms use SSE2 by default for floating point.
6267 Please see the ABI documentation for your platform.
6269 64-bit platforms differ in the sizes of the fundamental datatypes, not
6270 just from 32-bit platforms but from each other. If a specific size
6271 data type is desired, it is probably best to use the types defined in
6272 the Standard C header \c{<inttypes.h>}.
6274 \H{unix64} Interfacing to 64-bit C Programs (Unix)
6276 On Unix, the 64-bit ABI is defined by the document:
6278 \W{http://www.x86-64.org/documentation/abi.pdf}\c{http://www.x86-64.org/documentation/abi.pdf}
6280 Although written for AT&T-syntax assembly, the concepts apply equally
6281 well for NASM-style assembly. What follows is a simplified summary.
6283 The first six integer arguments (from the left) are passed in \c{RDI},
6284 \c{RSI}, \c{RDX}, \c{RCX}, \c{R8}, and \c{R9}, in that order.
6285 Additional integer arguments are passed on the stack. These
6286 registers, plus \c{RAX}, \c{R10} and \c{R11} are destroyed by function
6287 calls, and thus are available for use by the function without saving.
6289 Integer return values are passed in \c{RAX} and \c{RDX}, in that order.
6291 Floating point is done using SSE registers, except for \c{long
6292 double}. Floating-point arguments are passed in \c{XMM0} to \c{XMM7};
6293 return is \c{XMM0} and \c{XMM1}. \c{long double} are passed on the
6294 stack, and returned in \c{ST(0)} and \c{ST(1)}.
6296 All SSE and x87 registers are destroyed by function calls.
6298 On 64-bit Unix, \c{long} is 64 bits.
6300 \H{win64} Interfacing to 64-bit C Programs (Win64)
6302 The Win64 ABI is described at:
6304 \W{http://msdn2.microsoft.com/en-gb/library/ms794533.aspx}\c{http://msdn2.microsoft.com/en-gb/library/ms794533.aspx}
6306 What follows is a simplified summary.
6308 The first four integer arguments are passwd in \c{RCX}, \c{RDX},
6309 \c{R8} and \c{R9}, in that order. Additional integer arguments are
6310 passed on the stack. These registers, plus \c{RAX}, \c{R10} and
6311 \c{R11} are destroyed by function calls, and thus are available for
6312 use by the function without saving.
6314 Integer return values are passed in \c{RAX} only.
6316 Floating point is done using SSE registers, except for \c{long
6317 double}. Floating-point arguments are passed in \c{XMM0} to \c{XMM3};
6318 return is \c{XMM0} only.
6320 On Win64, \c{long} is 32 bits; \c{long long} or \c{_int64} is 64 bits.
6322 \C{trouble} Troubleshooting
6324 This chapter describes some of the common problems that users have
6325 been known to encounter with NASM, and answers them. It also gives
6326 instructions for reporting bugs in NASM if you find a difficulty
6327 that isn't listed here.
6330 \H{problems} Common Problems
6332 \S{inefficient} NASM Generates \i{Inefficient Code}
6334 We sometimes get `bug' reports about NASM generating inefficient, or
6335 even `wrong', code on instructions such as \c{ADD ESP,8}. This is a
6336 deliberate design feature, connected to predictability of output:
6337 NASM, on seeing \c{ADD ESP,8}, will generate the form of the
6338 instruction which leaves room for a 32-bit offset. You need to code
6339 \I\c{BYTE}\c{ADD ESP,BYTE 8} if you want the space-efficient form of
6340 the instruction. This isn't a bug, it's user error: if you prefer to
6341 have NASM produce the more efficient code automatically enable
6342 optimization with the \c{-On} option (see \k{opt-On}).
6345 \S{jmprange} My Jumps are Out of Range\I{out of range, jumps}
6347 Similarly, people complain that when they issue \i{conditional
6348 jumps} (which are \c{SHORT} by default) that try to jump too far,
6349 NASM reports `short jump out of range' instead of making the jumps
6352 This, again, is partly a predictability issue, but in fact has a
6353 more practical reason as well. NASM has no means of being told what
6354 type of processor the code it is generating will be run on; so it
6355 cannot decide for itself that it should generate \i\c{Jcc NEAR} type
6356 instructions, because it doesn't know that it's working for a 386 or
6357 above. Alternatively, it could replace the out-of-range short
6358 \c{JNE} instruction with a very short \c{JE} instruction that jumps
6359 over a \c{JMP NEAR}; this is a sensible solution for processors
6360 below a 386, but hardly efficient on processors which have good
6361 branch prediction \e{and} could have used \c{JNE NEAR} instead. So,
6362 once again, it's up to the user, not the assembler, to decide what
6363 instructions should be generated. See \k{opt-On}.
6366 \S{proborg} \i\c{ORG} Doesn't Work
6368 People writing \i{boot sector} programs in the \c{bin} format often
6369 complain that \c{ORG} doesn't work the way they'd like: in order to
6370 place the \c{0xAA55} signature word at the end of a 512-byte boot
6371 sector, people who are used to MASM tend to code
6375 \c ; some boot sector code
6380 This is not the intended use of the \c{ORG} directive in NASM, and
6381 will not work. The correct way to solve this problem in NASM is to
6382 use the \i\c{TIMES} directive, like this:
6386 \c ; some boot sector code
6388 \c TIMES 510-($-$$) DB 0
6391 The \c{TIMES} directive will insert exactly enough zero bytes into
6392 the output to move the assembly point up to 510. This method also
6393 has the advantage that if you accidentally fill your boot sector too
6394 full, NASM will catch the problem at assembly time and report it, so
6395 you won't end up with a boot sector that you have to disassemble to
6396 find out what's wrong with it.
6399 \S{probtimes} \i\c{TIMES} Doesn't Work
6401 The other common problem with the above code is people who write the
6406 by reasoning that \c{$} should be a pure number, just like 510, so
6407 the difference between them is also a pure number and can happily be
6410 NASM is a \e{modular} assembler: the various component parts are
6411 designed to be easily separable for re-use, so they don't exchange
6412 information unnecessarily. In consequence, the \c{bin} output
6413 format, even though it has been told by the \c{ORG} directive that
6414 the \c{.text} section should start at 0, does not pass that
6415 information back to the expression evaluator. So from the
6416 evaluator's point of view, \c{$} isn't a pure number: it's an offset
6417 from a section base. Therefore the difference between \c{$} and 510
6418 is also not a pure number, but involves a section base. Values
6419 involving section bases cannot be passed as arguments to \c{TIMES}.
6421 The solution, as in the previous section, is to code the \c{TIMES}
6424 \c TIMES 510-($-$$) DB 0
6426 in which \c{$} and \c{$$} are offsets from the same section base,
6427 and so their difference is a pure number. This will solve the
6428 problem and generate sensible code.
6431 \H{bugs} \i{Bugs}\I{reporting bugs}
6433 We have never yet released a version of NASM with any \e{known}
6434 bugs. That doesn't usually stop there being plenty we didn't know
6435 about, though. Any that you find should be reported firstly via the
6437 \W{https://sourceforge.net/projects/nasm/}\c{https://sourceforge.net/projects/nasm/}
6438 (click on "Bugs"), or if that fails then through one of the
6439 contacts in \k{contact}.
6441 Please read \k{qstart} first, and don't report the bug if it's
6442 listed in there as a deliberate feature. (If you think the feature
6443 is badly thought out, feel free to send us reasons why you think it
6444 should be changed, but don't just send us mail saying `This is a
6445 bug' if the documentation says we did it on purpose.) Then read
6446 \k{problems}, and don't bother reporting the bug if it's listed
6449 If you do report a bug, \e{please} give us all of the following
6452 \b What operating system you're running NASM under. DOS, Linux,
6453 NetBSD, Win16, Win32, VMS (I'd be impressed), whatever.
6455 \b If you're running NASM under DOS or Win32, tell us whether you've
6456 compiled your own executable from the DOS source archive, or whether
6457 you were using the standard distribution binaries out of the
6458 archive. If you were using a locally built executable, try to
6459 reproduce the problem using one of the standard binaries, as this
6460 will make it easier for us to reproduce your problem prior to fixing
6463 \b Which version of NASM you're using, and exactly how you invoked
6464 it. Give us the precise command line, and the contents of the
6465 \c{NASMENV} environment variable if any.
6467 \b Which versions of any supplementary programs you're using, and
6468 how you invoked them. If the problem only becomes visible at link
6469 time, tell us what linker you're using, what version of it you've
6470 got, and the exact linker command line. If the problem involves
6471 linking against object files generated by a compiler, tell us what
6472 compiler, what version, and what command line or options you used.
6473 (If you're compiling in an IDE, please try to reproduce the problem
6474 with the command-line version of the compiler.)
6476 \b If at all possible, send us a NASM source file which exhibits the
6477 problem. If this causes copyright problems (e.g. you can only
6478 reproduce the bug in restricted-distribution code) then bear in mind
6479 the following two points: firstly, we guarantee that any source code
6480 sent to us for the purposes of debugging NASM will be used \e{only}
6481 for the purposes of debugging NASM, and that we will delete all our
6482 copies of it as soon as we have found and fixed the bug or bugs in
6483 question; and secondly, we would prefer \e{not} to be mailed large
6484 chunks of code anyway. The smaller the file, the better. A
6485 three-line sample file that does nothing useful \e{except}
6486 demonstrate the problem is much easier to work with than a
6487 fully fledged ten-thousand-line program. (Of course, some errors
6488 \e{do} only crop up in large files, so this may not be possible.)
6490 \b A description of what the problem actually \e{is}. `It doesn't
6491 work' is \e{not} a helpful description! Please describe exactly what
6492 is happening that shouldn't be, or what isn't happening that should.
6493 Examples might be: `NASM generates an error message saying Line 3
6494 for an error that's actually on Line 5'; `NASM generates an error
6495 message that I believe it shouldn't be generating at all'; `NASM
6496 fails to generate an error message that I believe it \e{should} be
6497 generating'; `the object file produced from this source code crashes
6498 my linker'; `the ninth byte of the output file is 66 and I think it
6499 should be 77 instead'.
6501 \b If you believe the output file from NASM to be faulty, send it to
6502 us. That allows us to determine whether our own copy of NASM
6503 generates the same file, or whether the problem is related to
6504 portability issues between our development platforms and yours. We
6505 can handle binary files mailed to us as MIME attachments, uuencoded,
6506 and even BinHex. Alternatively, we may be able to provide an FTP
6507 site you can upload the suspect files to; but mailing them is easier
6510 \b Any other information or data files that might be helpful. If,
6511 for example, the problem involves NASM failing to generate an object
6512 file while TASM can generate an equivalent file without trouble,
6513 then send us \e{both} object files, so we can see what TASM is doing
6514 differently from us.
6517 \A{ndisasm} \i{Ndisasm}
6519 The Netwide Disassembler, NDISASM
6521 \H{ndisintro} Introduction
6524 The Netwide Disassembler is a small companion program to the Netwide
6525 Assembler, NASM. It seemed a shame to have an x86 assembler,
6526 complete with a full instruction table, and not make as much use of
6527 it as possible, so here's a disassembler which shares the
6528 instruction table (and some other bits of code) with NASM.
6530 The Netwide Disassembler does nothing except to produce
6531 disassemblies of \e{binary} source files. NDISASM does not have any
6532 understanding of object file formats, like \c{objdump}, and it will
6533 not understand \c{DOS .EXE} files like \c{debug} will. It just
6537 \H{ndisstart} Getting Started: Installation
6539 See \k{install} for installation instructions. NDISASM, like NASM,
6540 has a \c{man page} which you may want to put somewhere useful, if you
6541 are on a Unix system.
6544 \H{ndisrun} Running NDISASM
6546 To disassemble a file, you will typically use a command of the form
6548 \c ndisasm -b {16|32|64} filename
6550 NDISASM can disassemble 16-, 32- or 64-bit code equally easily,
6551 provided of course that you remember to specify which it is to work
6552 with. If no \i\c{-b} switch is present, NDISASM works in 16-bit mode
6553 by default. The \i\c{-u} switch (for USE32) also invokes 32-bit mode.
6555 Two more command line options are \i\c{-r} which reports the version
6556 number of NDISASM you are running, and \i\c{-h} which gives a short
6557 summary of command line options.
6560 \S{ndiscom} COM Files: Specifying an Origin
6562 To disassemble a \c{DOS .COM} file correctly, a disassembler must assume
6563 that the first instruction in the file is loaded at address \c{0x100},
6564 rather than at zero. NDISASM, which assumes by default that any file
6565 you give it is loaded at zero, will therefore need to be informed of
6568 The \i\c{-o} option allows you to declare a different origin for the
6569 file you are disassembling. Its argument may be expressed in any of
6570 the NASM numeric formats: decimal by default, if it begins with `\c{$}'
6571 or `\c{0x}' or ends in `\c{H}' it's \c{hex}, if it ends in `\c{Q}' it's
6572 \c{octal}, and if it ends in `\c{B}' it's \c{binary}.
6574 Hence, to disassemble a \c{.COM} file:
6576 \c ndisasm -o100h filename.com
6581 \S{ndissync} Code Following Data: Synchronisation
6583 Suppose you are disassembling a file which contains some data which
6584 isn't machine code, and \e{then} contains some machine code. NDISASM
6585 will faithfully plough through the data section, producing machine
6586 instructions wherever it can (although most of them will look
6587 bizarre, and some may have unusual prefixes, e.g. `\c{FS OR AX,0x240A}'),
6588 and generating `DB' instructions ever so often if it's totally stumped.
6589 Then it will reach the code section.
6591 Supposing NDISASM has just finished generating a strange machine
6592 instruction from part of the data section, and its file position is
6593 now one byte \e{before} the beginning of the code section. It's
6594 entirely possible that another spurious instruction will get
6595 generated, starting with the final byte of the data section, and
6596 then the correct first instruction in the code section will not be
6597 seen because the starting point skipped over it. This isn't really
6600 To avoid this, you can specify a `\i\c{synchronisation}' point, or indeed
6601 as many synchronisation points as you like (although NDISASM can
6602 only handle 8192 sync points internally). The definition of a sync
6603 point is this: NDISASM guarantees to hit sync points exactly during
6604 disassembly. If it is thinking about generating an instruction which
6605 would cause it to jump over a sync point, it will discard that
6606 instruction and output a `\c{db}' instead. So it \e{will} start
6607 disassembly exactly from the sync point, and so you \e{will} see all
6608 the instructions in your code section.
6610 Sync points are specified using the \i\c{-s} option: they are measured
6611 in terms of the program origin, not the file position. So if you
6612 want to synchronize after 32 bytes of a \c{.COM} file, you would have to
6615 \c ndisasm -o100h -s120h file.com
6619 \c ndisasm -o100h -s20h file.com
6621 As stated above, you can specify multiple sync markers if you need
6622 to, just by repeating the \c{-s} option.
6625 \S{ndisisync} Mixed Code and Data: Automatic (Intelligent) Synchronisation
6628 Suppose you are disassembling the boot sector of a \c{DOS} floppy (maybe
6629 it has a virus, and you need to understand the virus so that you
6630 know what kinds of damage it might have done you). Typically, this
6631 will contain a \c{JMP} instruction, then some data, then the rest of the
6632 code. So there is a very good chance of NDISASM being \e{misaligned}
6633 when the data ends and the code begins. Hence a sync point is
6636 On the other hand, why should you have to specify the sync point
6637 manually? What you'd do in order to find where the sync point would
6638 be, surely, would be to read the \c{JMP} instruction, and then to use
6639 its target address as a sync point. So can NDISASM do that for you?
6641 The answer, of course, is yes: using either of the synonymous
6642 switches \i\c{-a} (for automatic sync) or \i\c{-i} (for intelligent
6643 sync) will enable \c{auto-sync} mode. Auto-sync mode automatically
6644 generates a sync point for any forward-referring PC-relative jump or
6645 call instruction that NDISASM encounters. (Since NDISASM is one-pass,
6646 if it encounters a PC-relative jump whose target has already been
6647 processed, there isn't much it can do about it...)
6649 Only PC-relative jumps are processed, since an absolute jump is
6650 either through a register (in which case NDISASM doesn't know what
6651 the register contains) or involves a segment address (in which case
6652 the target code isn't in the same segment that NDISASM is working
6653 in, and so the sync point can't be placed anywhere useful).
6655 For some kinds of file, this mechanism will automatically put sync
6656 points in all the right places, and save you from having to place
6657 any sync points manually. However, it should be stressed that
6658 auto-sync mode is \e{not} guaranteed to catch all the sync points, and
6659 you may still have to place some manually.
6661 Auto-sync mode doesn't prevent you from declaring manual sync
6662 points: it just adds automatically generated ones to the ones you
6663 provide. It's perfectly feasible to specify \c{-i} \e{and} some \c{-s}
6666 Another caveat with auto-sync mode is that if, by some unpleasant
6667 fluke, something in your data section should disassemble to a
6668 PC-relative call or jump instruction, NDISASM may obediently place a
6669 sync point in a totally random place, for example in the middle of
6670 one of the instructions in your code section. So you may end up with
6671 a wrong disassembly even if you use auto-sync. Again, there isn't
6672 much I can do about this. If you have problems, you'll have to use
6673 manual sync points, or use the \c{-k} option (documented below) to
6674 suppress disassembly of the data area.
6677 \S{ndisother} Other Options
6679 The \i\c{-e} option skips a header on the file, by ignoring the first N
6680 bytes. This means that the header is \e{not} counted towards the
6681 disassembly offset: if you give \c{-e10 -o10}, disassembly will start
6682 at byte 10 in the file, and this will be given offset 10, not 20.
6684 The \i\c{-k} option is provided with two comma-separated numeric
6685 arguments, the first of which is an assembly offset and the second
6686 is a number of bytes to skip. This \e{will} count the skipped bytes
6687 towards the assembly offset: its use is to suppress disassembly of a
6688 data section which wouldn't contain anything you wanted to see
6692 \H{ndisbugs} Bugs and Improvements
6694 There are no known bugs. However, any you find, with patches if
6695 possible, should be sent to
6696 \W{mailto:nasm-bugs@lists.sourceforge.net}\c{nasm-bugs@lists.sourceforge.net}, or to the
6698 \W{https://sourceforge.net/projects/nasm/}\c{https://sourceforge.net/projects/nasm/}
6699 and we'll try to fix them. Feel free to send contributions and
6700 new features as well.
6702 Future plans include awareness of which processors certain
6703 instructions will run on, and marking of instructions that are too
6704 advanced for some processor (or are \c{FPU} instructions, or are
6705 undocumented opcodes, or are privileged protected-mode instructions,
6710 I hope NDISASM is of some use to somebody. Including me. :-)
6712 I don't recommend taking NDISASM apart to see how an efficient
6713 disassembler works, because as far as I know, it isn't an efficient
6714 one anyway. You have been warned.