The Yasm Modular Assembler is a portable, retargetable assembler written under the
(2 or 3 clause) BSD license. Yasm currently supports the x86 and AMD64 instruction sets, accepts NASM and GAS assembler syntaxes, outputs binary, ELF32, ELF64, COFF, Win32, and Win64 object formats, and generates source debugging information in STABS, DWARF 2, and CodeView 8 formats.
YASM consists of the
command, libyasm, the core backend library, and a large number of modules. Currently, libyasm and the loadable modules are statically built into the
command assembles the file infile and directs output to the file
if specified. If
is not specified,
will derive a default output file name from the name of its input file, usually by appending
.obj, or by removing all extensions for a raw binary file. Failing that, the output file name will be
If called with an
assembles the standard input and directs output to the file
Many options may be given in one of two forms: either a dash followed by a single letter, or two dashes followed by a long option name. Options are listed in alphabetical order.
-a arch or --arch=arch: Select target architecture
Selects the target architecture. The default architecture is
"x86", which supports both the IA-32 and derivatives and AMD64 instruction sets. To print a list of available architectures to standard output, use
for a list of supported architectures.
-f format or --oformat=format: Select object format
Selects the output object format. The default object format is
"bin", which is a flat format binary with no relocation. To print a list of available object formats to standard output, use
for a list of supported object formats.
-g debug or --dformat=debug: Select debugging format
Selects the debugging format for debug information. Debugging information can be used by a debugger to associate executable code back to the source file or get data structure and type information. Available debug formats vary between different object formats;
will error when an invalid combination is selected. The default object format is selected by the object format. To print a list of available debugging formats to standard output, use
for a list of supported debugging formats.
-L list or --lformat=list: Select list file format
Selects the format/style of the output list file. List files typically intermix the original source with the machine code generated by the assembler. The default list format is
"nasm", which mimics the NASM list file format. To print a list of available list file formats to standard output, use
-l listfile or --list=listfile: Specify list filename
Specifies the name of the output list file. If this option is not used, no list file is generated.
-m machine or --machine=machine: Select target machine architecture
Selects the target machine architecture. Essentially a subtype of the selected architecture, the machine type selects between major subsets of an architecture. For example, for the
architecture, the two available machines are
"x86", which is used for the IA-32 and derivative 32-bit instruction set, and
"amd64", which is used for the 64-bit instruction set. This differentiation is required to generate the proper object file for relocatable object formats such as COFF and ELF. To print a list of available machines for a given architecture to standard output, use
and the given architecture using
-a arch. See
for more details.
-o filename or --objfile=filename: Specify object filename
Specifies the name of the output file, overriding any default name generated by Yasm.
-p parser or --parser=parser: Select parser
Selects the parser (the assembler syntax). The default parser is
"nasm", which emulates the syntax of NASM, the Netwide Assembler. Another available parser is
"gas", which emulates the syntax of GNU AS. To print a list of available parsers to standard output, use
for a list of supported parsers.
-r preproc or --preproc=preproc: Select preprocessor
Selects the preprocessor to use on the input file before passing it to the parser. Preprocessors often provide macro functionality that is not included in the main parser. The default preprocessor is
"nasm", which is an imported version of the actual NASM preprocessor. A
preprocessor is also available, which simply skips the preprocessing step, passing the input file directly to the parser. To print a list of available preprocessors to standard output, use
-h or --help: Print a summary of options
Prints a summary of invocation options. All other options are ignored, and no output file is generated.
--version: Get the Yasm version
This option causes Yasm to prints the version number of Yasm as well as a license summary to standard output. All other options are ignored, and no output file is generated.
options have two contrary forms:
-Wno-name. Only the non-default forms are shown here.
The warning options are handled in the order given on the command line, so if
is followed by
-Worphan-labels, all warnings are turned off
-w: Inhibit all warning messages
This option causes Yasm to inhibit all warning messages. As discussed above, this option may be followed by other options to re-enable specified warnings.
-Werror: Treat warnings as errors
This option causes Yasm to treat all warnings as errors. Normally warnings do not prevent an object file from being generated and do not result in a failure exit status from
yasm, whereas errors do. This option makes warnings equivalent to errors in terms of this behavior.
-Wno-unrecognized-char: Do not warn on unrecognized input characters
Causes Yasm to not warn on unrecognized characters found in the input. Normally Yasm will generate a warning for any non-ASCII character found in the input file.
-Worphan-labels: Warn on labels lacking a trailing option
When using the NASM-compatible parser, causes Yasm to warn about labels found alone on a line without a trailing colon. While these are legal labels in NASM syntax, they may be unintentional, due to typos or macro definition ordering.
-X style: Change error/warning reporting style
Selects a specific output style for error and warning messages. The default is
style, which mimics the output of
style is also available, which mimics the output of Microsoft's Visual C++ compiler.
This option is available so that Yasm integrates more naturally into IDE environments such as
Emacs, allowing the IDE to correctly recognize the error/warning message as such and link back to the offending line of source code.
While these preprocessor options theoretically will affect any preprocessor, the only preprocessor currently in Yasm is the
-D macro[=value]: Pre-define a macro
Pre-defines a single-line macro. The value is optional (if no value is given, the macro is still defined, but to an empty value).
-e or --preproc-only: Only preprocess
Stops assembly after the preprocessing stage; preprocessed output is sent to the specified output name or, if no output name is specified, the standard output. No object file is produced.
-I path: Add include file path
to the search path for include files. The search path defaults to only including the directory in which the source file resides.
-P filename: Pre-include a file
filename, making it look as though
was prepended to the input. Can be useful for prepending multi-line macros that the
-U macro: Undefine a macro
Undefines a single-line macro (may be either a built-in macro or one defined earlier in the command line with
To assemble NASM syntax, 32-bit x86 source
into ELF file
source.o, warning on orphan labels:
yasm -f elf32 -Worphan-labels source.asm
To assemble NASM syntax AMD64 source
into Win64 file
yasm -f win64 -o object.obj x.asm
To assemble already preprocessed NASM syntax x86 source
into flat binary file
yasm -f bin -r raw -o y.com y.asm
command exits 0 on success, and nonzero if an error occurs.
Yasm's NASM parser and preprocessor, while they strive to be as compatible as possible with NASM, have a few incompatibilities due to YASM's different internal structure.
Yasm's GAS parser and preprocessor are missing a number of features present in GNU AS.
As object files are often architecture and machine dependent, not all combinations of object formats, architectures, and machines are legal; trying to use an invalid combination will result in an error.
There is no support for symbol maps.
When using the
architecture, it is overly easy to generate AMD64 code (using the
directive) and generate a 32-bit object file (by failing to specify
or selecting a 64-bit object format such as ELF64 on the command line).
Peter Johnson <email@example.com>
Copyright © 2004, 2005, 2006, 2007 Peter Johnson