Linux Assembly HOWTO Konstantin Boldyshev and Franois-Ren Rideau v0.5e, February 2000 This is the Linux Assembly HOWTO. This document describes how to pro- gram in assembly language using FREE programming tools, focusing on development for or from the Linux Operating System on IA-32 (i386) platform. Included material may or may not be applicable to other hardware and/or software platforms. Contributions about them will be gladly accepted. KKeeyywwoorrddss: assembly, assembler, free, macroprocessor, preprocessor, asm, inline asm, 32-bit, x86, i386, gas, as86, nasm, masm, tasm, OS, kernel, system, small, fast, embedded, hardware, port, I/O, syscall, system call, libc, interrupt. ______________________________________________________________________ Table of Contents 1. INTRODUCTION 1.1 Legal Blurb 1.2 Important Note 1.3 Foreword 1.3.1 How to use this document 1.3.2 Other related documents 1.4 History 1.5 Credits 2. DO YOU NEED ASSEMBLY? 2.1 Pros and Cons 2.1.1 The advantages of Assembly 2.1.2 The disadvantages of Assembly 2.1.3 Assessment 2.2 How to NOT use Assembly 2.2.1 General procedure to achieve efficient code 2.2.2 Languages with optimizing compilers 2.2.3 General procedure to speed your code up 2.2.4 Inspecting compiler-generated code 2.3 Linux and assembly 3. ASSEMBLERS 3.1 GCC Inline Assembly 3.1.1 Where to find GCC 3.1.2 Where to find docs for GCC Inline Asm 3.1.3 Invoking GCC to build proper inline assembly code 3.2 GAS 3.2.1 Where to find it 3.2.2 What is this AT&T syntax 3.2.3 16-bit mode 3.3 GASP 3.3.1 Where to find GASP 3.3.2 How it works 3.4 NASM 3.4.1 Where to find NASM 3.4.2 What it does 3.5 AS86 3.5.1 Where to get AS86 3.5.2 How to invoke the assembler? 3.5.3 Where to find docs 3.5.4 What if I can't compile Linux anymore with this new version ? 3.6 OTHER ASSEMBLERS 3.6.1 Win32Forth assembler 3.6.2 Terse 3.6.3 Non-free and/or Non-32bit x86 assemblers. 4. METAPROGRAMMING/MACROPROCESSING 4.1 What's integrated into the above 4.1.1 GCC 4.1.2 GAS 4.1.3 GASP 4.1.4 NASM 4.1.5 AS86 4.1.6 OTHER ASSEMBLERS 4.2 External Filters 4.2.1 CPP 4.2.2 M4 4.2.3 Macroprocessing with your own filter 4.2.4 Metaprogramming 4.2.4.1 Backends from compilers 4.2.4.2 The New-Jersey Machine-Code Toolkit 4.2.4.3 TUNES 5. CALLING CONVENTIONS 5.1 Linux 5.1.1 Linking to GCC 5.1.2 ELF vs a.out problems 5.1.3 Direct Linux syscalls 5.1.4 Hardware I/O under Linux 5.1.5 Accessing 16-bit drivers from Linux/i386 5.2 DOS 5.3 Windows and Co. 5.4 Your own OS 6. QUICK START 6.1 Tools you need 6.2 Hello, world! 6.2.1 NASM (hello.asm) 6.2.2 GAS (hello.S) 6.3 Producing object code 6.4 Producing executable 7. RESOURCES 7.1 Software projects 7.2 Tutorials 7.3 Mailing list 7.4 Books 7.5 Somehow related projects 7.6 General pointers ______________________________________________________________________ 11.. IINNTTRROODDUUCCTTIIOONN 11..11.. LLeeggaall BBlluurrbb Copyright (C) 1999-2000 Konstantin Boldyshev. Copyright (C) 1996-1999 Franois-Ren Rideau. This document is free; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or any later version. 11..22.. IImmppoorrttaanntt NNoottee This is an interactively evolving document: you are especially invited to ask questions, to answer questions, to correct given answers, to give pointers to new software, to point the current maintainer to bugs or deficiencies in the pages. In one word, contribute! To contribute, please contact the Assembly-HOWTO maintainer. At the time of this writing, it is Konstantin Boldyshev and no more Franois-Ren Rideau . I (Far) had been looking for some time for a serious hacker to replace me as maintainer of this document, and am pleased to announce Konstantin as my worthy successor. 11..33.. FFoorreewwoorrdd This document aims answering questions of those who program or want to program 32-bit x86 assembly using _f_r_e_e _s_o_f_t_w_a_r_e _, particularly under the Linux operating system. It may also point to other documents about non- free, non-x86, or non-32-bit assemblers, although this is not its primary goal. Because the main interest of assembly programming is to build the guts of operating systems, interpreters, compilers, and games, where C compiler fails to provide the needed expressiveness (performance is more and more seldom as issue), we are focusing on development of such kind of software. 11..33..11.. HHooww ttoo uussee tthhiiss ddooccuummeenntt This document contains answers to some frequently asked questions. At many places, Universal Resource Locators (URL) are given for some software or documentation repository. Please see that the most useful repositories are mirrored, and that by accessing a nearer mirror site, you relieve the whole Internet from unneeded network traffic, while saving your own precious time. Particularly, there are large repositories all over the world, that mirror other popular repositories. You should learn and note what are those places near you (networkwise). Sometimes, the list of mirrors is listed in a file, or in a login message. Please heed the advice. Else, you should ask archie about the software you're looking for... The most recent official version of this document is available from Linux Assembly and LDP sites. 11..33..22.. OOtthheerr rreellaatteedd ddooccuummeennttss +o If you don't know what _f_r_e_e software is, please do read _c_a_r_e_f_u_l_l_y the GNU General Public License, which is used in a lot of free software, and is a model for most of their licenses. It generally comes in a file named COPYING, with a library version in a file named COPYING.LIB. Literature from the FSF (free software foundation) might help you, too. +o Particularly, the interesting feature of free software is that it comes with sources that you can consult and correct, or sometimes even borrow from. Read your particular license carefully, and do comply to it. +o There are FAQs and docs about programming on your favorite platform, whatever it is, which you should consult for platform- specific issues, not related directly to assembly programming. +o Refer to ``RESOURCES'' section of this HOWTO for pointers to related documents and projects. 11..44.. HHiissttoorryy Each version includes a few fixes and minor corrections, that need not to be repeatedly mentioned every time. VVeerrssiioonn 00..55ee 1100 FFeebb 22000000 url updates, changes in GAS example VVeerrssiioonn 00..55dd 0011 FFeebb 22000000 RESOURCE (former POINTERS) section completely redone, various url updates. VVeerrssiioonn 00..55cc 0055 DDeecc 11999999 New pointers, updates and some rearrangements. Rewrite of sgml source. VVeerrssiioonn 00..55bb 1199 SSeepp 11999999 Discussion about libc or not libc continues. New web pointers and and overall updates. VVeerrssiioonn 00..55aa 0011 AAuugg 11999999 "QUICK START" section rearranged, added GAS example. Several new web pointers. VVeerrssiioonn 00..55 2255 JJuullyy 11999999 GAS has 16-bit mode. New maintainer (at last): Konstantin Boldyshev. Discussion about libc or not libc. Added section "QUICK START" with examples of using assembly. VVeerrssiioonn 00..44qq 2222 JJuunnee 11999999 process argument passing (argc,argv,environ) in assembly. This is yet another "last release by Far before new maintainer takes over". Nobody knows who might be the new maintainer. VVeerrssiioonn 00..44pp 66 JJuunnee 11999999 clean up and updates. VVeerrssiioonn 00..44oo 11 DDeecceemmbbeerr 11999988 * VVeerrssiioonn 00..44mm 2233 MMaarrcchh 11999988 corrections about gcc invocation VVeerrssiioonn 00..44ll 1166 NNoovveemmbbeerr 11999977 release for LSL 6th edition. VVeerrssiioonn 00..44kk 1199 OOccttoobbeerr 11999977 * VVeerrssiioonn 00..44jj 77 SSeepptteemmbbeerr 11999977 * VVeerrssiioonn 00..44ii 1177 JJuullyy 11999977 info on 16-bit mode access from Linux. VVeerrssiioonn 00..44hh 1199 JJuunn 11999977 still more on "how not to use assembly"; updates on NASM, GAS. VVeerrssiioonn 00..44gg 3300 MMaarr 11999977 * VVeerrssiioonn 00..44ff 2200 MMaarr 11999977 * VVeerrssiioonn 00..44ee 1133 MMaarr 11999977 Release for DrLinux VVeerrssiioonn 00..44dd 2288 FFeebb 11999977 Vapor announce of a new Assembly-HOWTO maintainer. VVeerrssiioonn 00..44cc 99 FFeebb 11999977 Added section "DO YOU NEED ASSEMBLY?" VVeerrssiioonn 00..44bb 33 FFeebb 11999977 NASM moved: now is before AS86 VVeerrssiioonn 00..44aa 2200 JJaann 11999977 CREDITS section added VVeerrssiioonn 00..44 2200 JJaann 11999977 first release of the HOWTO as such. VVeerrssiioonn 00..44pprree11 1133 JJaann 11999977 text mini-HOWTO transformed into a full linuxdoc-sgml HOWTO, to see what the SGML tools are like. VVeerrssiioonn 00..33ll 1111 JJaann 11999977 * VVeerrssiioonn 00..33kk 1199 DDeecc 11999966 What? I had forgotten to point to terse??? VVeerrssiioonn 00..33jj 2244 NNoovv 11999966 point to French translated version VVeerrssiioonn 00..33ii 1166 NNoovv 11999966 NASM is getting pretty slick VVeerrssiioonn 00..33hh 66 NNoovv 11999966 more about cross-compiling -- See on sunsite: devel/msdos/ VVeerrssiioonn 00..33gg 22 NNoovv 11999966 Created the History. Added pointers in cross-compiling section. Added section about I/O programming under Linux (particularly video). VVeerrssiioonn 00..33ff 1177 OOcctt 11999966 * VVeerrssiioonn 00..33cc 1155 JJuunn 11999966 * VVeerrssiioonn 00..22 0044 MMaayy 11999966 * VVeerrssiioonn 00..11 2233 AApprr 11999966 Francois-Rene "Far" Rideau creates and publishes the first mini-HOWTO, because "I'm sick of answering ever the same questions on comp.lang.asm.x86" 11..55.. CCrreeddiittss I would like to thank following persons, by order of appearance: +o Linus Torvalds for Linux +o Bruce Evans for bcc from which as86 is extracted +o Simon Tatham and Julian Hall for NASM +o Greg Hankins and now Tim Bynum for maintaining HOWTOs +o Raymond Moon for his FAQ +o Eric Dumas for his translation of the mini-HOWTO into French (sad thing for the original author to be French and write in English) +o Paul Anderson and Rahim Azizarab for helping me, if not for taking over the HOWTO. +o Marc Lehman for his insight on GCC invocation. +o Abhijit Menon-Sen for helping me figure out the process argument passing convention +o All the people who have contributed ideas, remarks, and moral support. 22.. DDOO YYOOUU NNEEEEDD AASSSSEEMMBBLLYY?? Well, I wouldn't want to interfere with what you're doing, but here is some advice from hard-earned experience. 22..11.. PPrrooss aanndd CCoonnss 22..11..11.. TThhee aaddvvaannttaaggeess ooff AAsssseemmbbllyy Assembly can express very low-level things: +o you can access machine-dependent registers and I/O. +o you can control the exact behavior of code in critical sections that might otherwise involve deadlock between multiple software threads or hardware devices. +o you can break the conventions of your usual compiler, which might allow some optimizations (like temporarily breaking rules about memory allocation, threading, calling conventions, etc). +o you can build interfaces between code fragments using incompatible such conventions (e.g. produced by different compilers, or separated by a low-level interface). +o you can get access to unusual programming modes of your processor (e.g. 16 bit mode to interface startup, firmware, or legacy code on Intel PCs) +o you can produce reasonably fast code for tight loops to cope with a bad non-optimizing compiler (but then, there are free optimizing compilers available!) +o you can produce code where (but only on CPUs with known instruction timings, which generally excludes all current .... +o you can produce hand-optimized code that's perfectly tuned for your particular hardware setup, though not to anyone else's. +o you can write some code for your new language's optimizing compiler (that's something few will ever do, and even they, not often). 22..11..22.. TThhee ddiissaaddvvaannttaaggeess ooff AAsssseemmbbllyy Assembly is a very low-level language (the lowest above hand-coding the binary instruction patterns). This means +o it's long and tedious to write initially, +o it's very bug-prone, +o your bugs will be very difficult to chase, +o it's very difficult to understand and modify, i.e. to maintain. +o the result is very non-portable to other architectures, existing or future, +o your code will be optimized only for a certain implementation of a same architecture: for instance, among Intel-compatible platforms, each CPU design and its variations (relative latency, throughput, and capacity, of processing units, caches, RAM, bus, disks, presence of FPU, MMX, 3DNOW, SIMD extensions, etc) implies potentially completely different optimization techniques. CPU designs already include: Intel 386, 486, Pentium, PPro, Pentium II, Pentium III; Cyrix 5x86, 6x86; AMD K5, K6 (K6-2, K6-III), K7 (Athlon). New designs keep popping up, so don't expect either this listing or your code to be up-to-date. +o your code might also be unportable across different OS platforms on the same architecture, by lack of proper tools. (well, GAS seems to work on all platforms; NASM seems to work or be workable on all Intel platforms). +o you spend more time on a few details, and can't focus on small and large algorithmic design, that are known to bring the largest part of the speed up. [e.g. you might spend some time building very fast list/array manipulation primitives in assembly; only a hash table would have sped up your program much more; or, in another context, a binary tree; or some high-level structure distributed over a cluster of CPUs] +o a small change in algorithmic design might completely invalidate all your existing assembly code. So that either you're ready (and able) to rewrite it all, or you're tied to a particular algorithmic design; +o On code that ain't too far from what's in standard benchmarks, commercial optimizing compilers outperform hand-coded assembly (well, that's less true on the x86 architecture than on RISC architectures, and perhaps less true for widely available/free compilers; anyway, for typical C code, GCC is fairly good); +o And in any case, as says moderator John Levine on comp.compilers , "compilers make it a lot easier to use complex data structures, and compilers don't get bored halfway through and generate reliably pretty good code." They will also _c_o_r_r_e_c_t_l_y propagate code transformations throughout the whole (huge) program when optimizing code between procedures and module boundaries. 22..11..33.. AAsssseessssmmeenntt All in all, you might find that though using assembly is sometimes needed, and might even be useful in a few cases where it is not, you'll want to: +o minimize the use of assembly code, +o encapsulate this code in well-defined interfaces +o have your assembly code automatically generated from patterns expressed in a higher-level language than assembly (e.g. GCC inline assembly macros). +o have automatic tools translate these programs into assembly code +o have this code be optimized if possible +o All of the above, i.e. write (an extension to) an optimizing compiler back-end. Even in cases when Assembly is needed (e.g. OS development), you'll find that not so much of it is, and that the above principles hold. See the Linux kernel sources concerning this: as little assembly as needed, resulting in a fast, reliable, portable, maintainable OS. Even a successful game like DOOM was almost massively written in C, with a tiny part only being written in assembly for speed up. 22..22.. HHooww ttoo NNOOTT uussee AAsssseemmbbllyy 22..22..11.. GGeenneerraall pprroocceedduurree ttoo aacchhiieevvee eeffffiicciieenntt ccooddee As says Charles Fiterman on comp.compilers about human vs computer-generated assembly code, "The human should always win and here is why. +o First the human writes the whole thing in a high level language. +o Second he profiles it to find the hot spots where it spends its time. +o Third he has the compiler produce assembly for those small sections of code. +o Fourth he hand tunes them looking for tiny improvements over the machine generated code. The human wins because he can use the machine." 22..22..22.. LLaanngguuaaggeess wwiitthh ooppttiimmiizziinngg ccoommppiilleerrss Languages like ObjectiveCAML, SML, CommonLISP, Scheme, ADA, Pascal, C, C++, among others, all have free optimizing compilers that'll optimize the bulk of your programs, and often do better than hand-coded assembly even for tight loops, while allowing you to focus on higher- level details, and without forbidding you to grab a few percent of extra performance in the above-mentioned way, once you've reached a stable design. Of course, there are also commercial optimizing compilers for most of these languages, too! Some languages have compilers that produce C code, which can be further optimized by a C compiler: LISP, Scheme, Perl, and many other. Speed is fairly good. 22..22..33.. GGeenneerraall pprroocceedduurree ttoo ssppeeeedd yyoouurr ccooddee uupp As for speeding code up, you should do it only for parts of a program that a profiling tool has consistently identified as being a performance bottleneck. Hence, if you identify some code portion as being too slow, you should +o first try to use a better algorithm; +o then try to compile it rather than interpret it; +o then try to enable and tweak optimization from your compiler; +o then give the compiler hints about how to optimize (typing information in LISP; register usage with GCC; lots of options in most compilers, etc). +o then possibly fallback to assembly programming Finally, before you end up writing assembly, you should inspect generated code, to check that the problem really is with bad code generation, as this might really not be the case: compiler-generated code might be better than what you'd have written, particularly on modern multi-pipelined architectures! Slow parts of a program might be intrinsically so. Biggest problems on modern architectures with fast processors are due to delays from memory access, cache-misses, TLB-misses, and page-faults; register optimization becomes useless, and you'll more profitably re-think data structures and threading to achieve better locality in memory access. Perhaps a completely different approach to the problem might help, then. 22..22..44.. IInnssppeeccttiinngg ccoommppiilleerr--ggeenneerraatteedd ccooddee There are many reasons to inspect compiler-generated assembly code. Here are what you'll do with such code: +o check whether generated code can be obviously enhanced with hand- coded assembly (or by tweaking compiler switches) +o when that's the case, start from generated code and modify it instead of starting from scratch +o more generally, use generated code as stubs to modify, which at least gets right the way your assembly routines interface to the external world +o track down bugs in your compiler (hopefully rarer) The standard way to have assembly code be generated is to invoke your compiler with the -S flag. This works with most Unix compilers, including the GNU C Compiler (GCC), but YMMV. As for GCC, it will produce more understandable assembly code with the -fverbose-asm command-line option. Of course, if you want to get good assembly code, don't forget your usual optimization options and hints! 22..33.. LLiinnuuxx aanndd aasssseemmbbllyy In general case you don't need to use assembly language in Linux programming. Unlike DOS, you do not have to write Linux drivers in assembly (this must be done in C). And with modern optimizing compilers, if you care of speed optimization for different CPU's, it's much simpler to write in C. However, if you're reading this, you might have some reason to use assembly instead of C/C++. You may _n_e_e_d to use assembly, or you may _w_a_n_t to use assembly. Shortly, main practical reasons why you may need to get into Linux assembly are _s_m_a_l_l _c_o_d_e and _l_i_b_c _i_n_d_e_p_e_n_d_e_n_c_e. Non-practical (and most often) reason is being just an old crazy hacker, who has twenty years old habit of doing everything in assembly language. Also, if you're porting Linux to some embedded hardware you can be quite short at size of whole system: you need to fit kernel, libc and all that stuff of (file|find|text|sh|etc.) utils into several hundreds of kilobytes, and every kilobyte costs much. So, one of the ways you've got is to rewrite some (or all) parts of system in assembly, and this will really save you a lot of space. For instance, a simple httpd written in assembly can take less than 800 bytes; you can fit a webserver, consisting of kernel and httpd, in 400 KB or less... Think about it. 33.. AASSSSEEMMBBLLEERRSS 33..11.. GGCCCC IInnlliinnee AAsssseemmbbllyy The well-known GNU C/C++ Compiler (GCC), an optimizing 32-bit compiler at the heart of the GNU project, supports the x86 architecture quite well, and includes the ability to insert assembly code in C programs, in such a way that register allocation can be either specified or left to GCC. GCC works on most available platforms, notably Linux, *BSD, VSTa, OS/2, *DOS, Win*, etc. 33..11..11.. WWhheerree ttoo ffiinndd GGCCCC The original GCC site is the GNU FTP site together with all released application software from the GNU project. Linux-configured and precompiled versions can be found in There exists a lot of FTP mirrors of both sites. everywhere around the world, as well as CD-ROM copies. GCC development has split into two branches some time ago (GCC 2.8 and EGCS), but they merged back, and current GCC webpage is . Sources adapted to your favorite OS, and binaries precompiled for it, should be found at your usual FTP sites. For most popular DOS port of GCC is named DJGPP, and can be found in directories of such name in FTP sites. See: There is also a port of GCC to OS/2 named EMX, that also works under DOS, and includes lots of unix-emulation library routines. See around the following site: . 33..11..22.. WWhheerree ttoo ffiinndd ddooccss ffoorr GGCCCC IInnlliinnee AAssmm The documentation of GCC includes documentation files in texinfo format. You can compile them with tex and print then result, or convert them to .info, and browse them with emacs, or convert them to .html, or nearly whatever you like. convert (with the right tools) to whatever you like, or just read as is. The .info files are generally found on any good installation for GCC. The right section to look for is: C Extensions::Extended Asm:: Section Invoking GCC::Submodel Options::i386 Options:: might help too. Particularly, it gives the i386 specific constraint names for registers: abcdSDB correspond to %eax, %ebx, %ecx, %edx, %esi, %edi and %ebp respectively (no letter for %esp). The DJGPP Games resource (not only for game hackers) had page specifically about assembly, but it's down. Its data have nonetheless been recovered on the DJGPP site , that contains a mine of other useful information: , and in the DJGPP Quick ASM Programming Guide . GCC depends on GAS for assembling, and follow its syntax (see below); do mind that inline asm needs percent characters to be quoted so they be passed to GAS. See the section about GAS below. Find _l_o_t_s of useful examples in the linux/include/asm-i386/ subdirectory of the sources for the Linux kernel. 33..11..33.. IInnvvookkiinngg GGCCCC ttoo bbuuiilldd pprrooppeerr iinnlliinnee aasssseemmbbllyy ccooddee Because assembly routines from the kernel headers (and most likely your own headers, if you try making your assembly programming as clean as it is in the linux kernel) are embedded in extern inline functions, GCC must be invoked with the -O flag (or -O2, -O3, etc), for these routines to be available. If not, your code may compile, but not link properly, since it will be looking for non-inlined extern functions in the libraries against which your program is being linked! Another way is to link against libraries that include fallback versions of the routines. Inline assembly can be disabled with -fno-asm, which will have the compiler die when using extended inline asm syntax, or else generate calls to an external function named asm() that the linker can't resolve. To counter such flag, -fasm restores treatment of the asm keyword. More generally, good compile flags for GCC on the x86 platform are ______________________________________________________________________ gcc -O2 -fomit-frame-pointer -W -Wall ______________________________________________________________________ -O2 is the good optimization level in most cases. Optimizing besides it takes longer, and yields code that is a lot larger, but only a bit faster; such overoptimization might be useful for tight loops only (if any), which you may be doing in assembly anyway. In cases when you need really strong compiler optimization for a few files, do consider using up to -O6. -fomit-frame-pointer allows generated code to skip the stupid frame pointer maintenance, which makes code smaller and faster, and frees a register for further optimizations. It precludes the easy use of debugging tools (gdb), but when you use these, you just don't care about size and speed anymore anyway. -W -Wall enables all warnings and helps you catch obvious stupid errors. You can add some CPU-specific -m486 or such flag so that GCC will produce code that is more adapted to your precise computer. Note that modern GCC has -mpentium and such flags (and PGCC has even more), whereas GCC 2.7.x and older versions do not. A good choice of CPU-specific flags should be in the Linux kernel. Check the texinfo documentation of your current GCC installation for more. -m386 will help optimize for size, hence also for speed on computers whose memory is tight and/or loaded, since big programs cause swap, which more than counters any "optimization" intended by the larger code. In such settings, it might be useful to stop using C, and use instead a language that favors code factorization, such as a functional language and/or FORTH, and use a bytecode- or wordcode- based implementation. Note that you can vary code generation flags from file to file, so performance-critical files will use maximum optimization, whereas other files will be optimized for size. To optimize even more, option -mregparm=2 and/or corresponding function attribute might help, but might pose lots of problems when linking to foreign code, _i_n_c_l_u_d_i_n_g _t_h_e _l_i_b_c. There are ways to correctly declare foreign functions so the right call sequences be generated, or you might want to recompile the foreign libraries to use the same register-based calling convention... Note that you can add make these flags the default by editing file /usr/lib/gcc-lib/i486-linux/2.7.2.3/specs or wherever that is on your system (better not add -W -Wall there, though). The exact location of the GCC specs files on _y_o_u_r system can be found by asking gcc -v. 33..22.. GGAASS GAS is the GNU Assembler, that GCC relies upon. 33..22..11.. WWhheerree ttoo ffiinndd iitt Find it at the same place where you found GCC, in a package named binutils. The latest version is available from HJLu at . 33..22..22.. WWhhaatt iiss tthhiiss AATT&&TT ssyynnttaaxx Because GAS was invented to support a 32-bit unix compiler, it uses standard AT&T syntax, which resembles a lot the syntax for standard m68k assemblers, and is standard in the UNIX world. This syntax is no worse, no better than the Intel syntax. It's just different. When you get used to it, you find it much more regular than the Intel syntax, though a bit boring. Here are the major caveats about GAS syntax: +o Register names are prefixed with %, so that registers are %eax, %dl and so on, instead of just eax, dl, etc. This makes it possible to include external C symbols directly in assembly source, without any risk of confusion, or any need for ugly underscore prefixes. +o The order of operands is source(s) first, and destination last, as opposed to the Intel convention of destination first and sources last. Hence, what in Intel syntax is mov ax,dx (move contents of register dx into register ax) will be in GAS syntax mov %dx, %ax. +o The operand length is specified as a suffix to the instruction name. The suffix is b for (8-bit) byte, w for (16-bit) word, and l for (32-bit) long. For instance, the correct syntax for the above instruction would have been movw %dx,%ax. However, gas does not require strict AT&T syntax, so the suffix is optional when length can be guessed from register operands, and else defaults to 32-bit (with a warning). +o Immediate operands are marked with a $ prefix, as in addl $5,%eax (add immediate long value 5 to register %eax). +o No prefix to an operand indicates it is a memory-address; hence movl $foo,%eax puts the _a_d_d_r_e_s_s of variable foo in register %eax, but movl foo,%eax puts the _c_o_n_t_e_n_t_s of variable foo in register %eax. +o Indexing or indirection is done by enclosing the index register or indirection memory cell address in parentheses, as in testb $0x80,17(%ebp) (test the high bit of the byte value at offset 17 from the cell pointed to by %ebp). A program exists to help you convert programs from TASM syntax to AT&T syntax. See . (Since the original x2ftp site is closing (no more?), use a mirror site ). There also exists a program for the reverse conversion: . GAS has comprehensive documentation in TeXinfo format, which comes at least with the source distribution. Browse extracted .info pages with Emacs or whatever. There used to be a file named gas.doc or as.doc around the GAS source package, but it was merged into the TeXinfo docs. Of course, in case of doubt, the ultimate documentation is the sources themselves! A section that will particularly interest you is Machine Dependencies::i386-Dependent:: Again, the sources for Linux (the OS kernel) come in as excellent examples; see under linux/arch/i386/ the following files: kernel/*.S, boot/compressed/*.S, mathemu/*.S. If you are writing kind of a language, a thread package, etc., you might as well see how other languages (OCaml , Gforth , etc.), or thread packages (QuickThreads, MIT pthreads, LinuxThreads, etc), or whatever, do it. Finally, just compiling a C program to assembly might show you the syntax for the kind of instructions you want. See section ``Do you need Assembly?'' above. 33..22..33.. 1166--bbiitt mmooddee The current stable release of binutils (2.9.1.0.25) now fully supports 16-bit mode (registers _a_n_d addressing) on i386 PCs. Still with its peculiar AT&T syntax, of course. Use .code16 and .code32 to switch between assembly modes. Also, a neat trick used by some (including the oskit authors) is to have GCC produce code for 16-bit real mode, using an inline assembly statement asm(".code16\n"). GCC will still emit only 32-bit addressing modes, but GAS will insert proper 32-bit prefixes for them. 33..33.. GGAASSPP GASP is the GAS Preprocessor. It adds macros and some nice syntax to GAS. 33..33..11.. WWhheerree ttoo ffiinndd GGAASSPP GASP comes together with GAS in the GNU binutils archive. 33..33..22.. HHooww iitt wwoorrkkss It works as a filter, much like cpp and the like. I have no idea on details, but it comes with its own texinfo documentation, so just browse them (in .info), print them, grok them. GAS with GASP looks like a regular macro-assembler to me. 33..44.. NNAASSMM The Netwide Assembler project provides cool i386 assembler, written in C, that should be modular enough to eventually support all known syntaxes and object formats. 33..44..11.. WWhheerree ttoo ffiinndd NNAASSMM Binary release on your usual metalab mirror in devel/lang/asm/ Should also be available as .rpm or .deb in your usual RedHat/Debian distributions' contrib. 33..44..22.. WWhhaatt iitt ddooeess At the time this HOWTO is written, current version of NASM is 0.98. The syntax is Intel-style. Excellent macroprocessing support is integrated. Supported object file formats are bin, aout, coff, elf, as86, (DOS) obj, win32, (their own format) rdf. NASM can be used as a backend for the free LCC compiler (support files included). Unless you're using BCC as a 16-bit compiler (which is out of scope of this 32-bit HOWTO), you should definitely use NASM instead of say AS86 or MASM, because it is actively supported online, and runs on all platforms. Note: NASM also comes with a disassembler, NDISASM. Its hand-written parser makes it much faster than GAS, though of course, it doesn't support three bazillion different architectures. If you like Intel-style syntax, as opposed to GAS syntax, then it should be the assembler of choice... Note: There's a ``converter between GAS AT&T and Intel assembler syntax'', which does conversion in both directions. 33..55.. AASS8866 AS86 is a 80x86 assembler, both 16-bit and 32-bit, part of Bruce Evans' C Compiler (BCC). It has mostly Intel-syntax, though it differs slightly as for addressing modes. 33..55..11.. WWhheerree ttoo ggeett AASS8866 A completely outdated version of AS86 is distributed by HJLu just to compile the Linux kernel, in a package named bin86 (current version 0.4), available in any Linux GCC repository. But I advise no one to use it for anything else but compiling Linux. This version supports only a hacked minix object file format, which is not supported by the GNU binutils or anything, and it has a few bugs in 32-bit mode, so you really should better keep it only for compiling Linux. The most recent versions by Bruce Evans (bde@zeta.org.au) are published together with the FreeBSD distribution. Well, they were: I could not find the sources from distribution 2.1 on :( Hence, I put the sources at my place: The Linux/8086 (aka ELKS) project is somehow maintaining bcc (though I don't think they included the 32-bit patches). See around (or ) and . I haven't followed these developments, and would appreciate a reader contributing on this topic. Among other things, these more recent versions, unlike HJLu's, supports Linux GNU a.out format, so you can link you code to Linux programs, and/or use the usual tools from the GNU binutils package to manipulate your data. This version can co-exist without any harm with the previous one (see according question below). BCC from 12 march 1995 and earlier version has a misfeature that makes all segment pushing/popping 16-bit, which is quite annoying when programming in 32-bit mode. I wrote a patch at a time when the TUNES Project used as86: . Bruce Evans accepted this patch, but since as far as I know he hasn't published a new release of bcc, the ones to ask about integrating it (if not done yet) are the ELKS developers. 33..55..22.. HHooww ttoo iinnvvookkee tthhee aasssseemmbblleerr?? Here's the GNU Makefile entry for using bcc to transform .s asm into both GNU a.out .o object and .l listing: ______________________________________________________________________ %.o %.l: %.s bcc -3 -G -c -A-d -A-l -A$*.l -o $*.o $< ______________________________________________________________________ Remove the %.l, -A-l, and -A$*.l, if you don't want any listing. If you want something else than GNU a.out, you can see the docs of bcc about the other supported formats, and/or use the objcopy utility from the GNU binutils package. 33..55..33.. WWhheerree ttoo ffiinndd ddooccss The docs are what is included in the bcc package. I salvaged the man pages that used to be available from the FreeBSD site at . Maybe ELKS developers know better. When in doubt, the sources themselves are often a good docs: it's not very well commented, but the programming style is straightforward. You might try to see how as86 is used in ELKS or Tunes 0.0.0.25... 33..55..44.. WWhhaatt iiff II ccaann''tt ccoommppiillee LLiinnuuxx aannyymmoorree wwiitthh tthhiiss nneeww vveerrssiioonn ?? Linus is buried alive in mail, and since HJLu (official bin86 maintainer) chose to write hacks around an obsolete version of as86 instead of building clean code around the latest version, I don't think my patch for compiling Linux with a modern as86 has any chance to be accepted if resubmitted. Now, this shouldn't matter: just keep your as86 from the bin86 package in /usr/bin/, and let bcc install the good as86 as /usr/local/libexec/i386/bcc/as where it should be. You never need explicitly call this "good" as86, because bcc does everything right, including conversion to Linux a.out, when invoked with the right options; so assemble files exclusively with bcc as a frontend, not directly with as86. Since GAS now supports 16-bit code, and since H. Peter Anvin, well- known linux hacker, works on NASM, maybe Linux will get rid of AS86, anyway? Who knows! 33..66.. OOTTHHEERR AASSSSEEMMBBLLEERRSS These are other, non-regular, options, in case the previous didn't satisfy you (why?), that I don't recommend in the usual (?) case, but that could prove quite useful if the assembler must be integrated in the software you're designing (i.e. an OS or development environment). 33..66..11.. WWiinn3322FFoorrtthh aasssseemmbblleerr Win32Forth is a _f_r_e_e 32-bit ANS FORTH system that successfully runs under Win32s, Win95, Win/NT. It includes a free 32-bit assembler (either prefix or postfix syntax) integrated into the reflective FORTH language. Macro processing is done with the full power of the reflective language FORTH; however, the only supported input and output contexts is Win32For itself (no dumping of .obj file, but you could add that feature yourself, of course). Find it at . 33..66..22.. TTeerrssee Terse is a programming tool that provides _T_H_E most compact assembler syntax for the x86 family! However, it is evil proprietary software. It is said that there was a project for a free clone somewhere, that was abandoned after worthless pretenses that the syntax would be owned by the original author. Thus, if you're looking for a nifty programming project related to assembly hacking, I invite you to develop a terse-syntax frontend to NASM, if you like that syntax. As an interesting historic remark, on comp.compilers , 1999/07/11 19:36:51, the moderator wrote: "There's no reason that assemblers have to have awful syntax. About 30 years ago I used Niklaus Wirth's PL360, which was basically a S/360 assembler with Algol syntax and a a little syntactic sugar like while loops that turned into the obvious branches. It really was an assembler, e.g., you had to write out your expressions with explicit assignments of values to registers, but it was nice. Wirth used it to write Algol W, a small fast Algol subset, which was a predecessor to Pascal. As is so often the case, Algol W was a significant improvement over many of its successors. -John" 33..66..33.. NNoonn--ffrreeee aanndd//oorr NNoonn--3322bbiitt xx8866 aasssseemmbblleerrss.. You may find more about them, together with the basics of x86 assembly programming, in ``Raymond Moon's FAQ for comp.lang.asm.x86''. Note that all DOS-based assemblers should work inside the Linux DOS Emulator, as well as other similar emulators, so that if you already own one, you can still use it inside a real OS. Recent DOS-based assemblers also support COFF and/or other object file formats that are supported by the GNU BFD library, so that you can use them together with your free 32-bit tools, perhaps using GNU objcopy (part of the binutils) as a conversion filter. 44.. MMEETTAAPPRROOGGRRAAMMMMIINNGG//MMAACCRROOPPRROOCCEESSSSIINNGG Assembly programming is a bore, but for critical parts of programs. You should use the appropriate tool for the right task, so don't choose assembly when it's not fit; C, OCaml, perl, Scheme, might be a better choice for most of your programming. However, there are cases when these tools do not give a fine enough control on the machine, and assembly is useful or needed. In those case, you'll appreciate a system of macroprocessing and metaprogramming that'll allow recurring patterns to be factored each into a one indefinitely reusable definition, which allows safer programming, automatic propagation of pattern modification, etc. Plain assembler often is not enough, even when one is doing only small routines to link with C. 44..11.. WWhhaatt''ss iinntteeggrraatteedd iinnttoo tthhee aabboovvee Yes I know this section does not contain much useful up-to-date information. Feel free to contribute what you discover the hard way... 44..11..11.. GGCCCC GCC allows (and requires) you to specify register constraints in your inline assembly code, so the optimizer always know about it; thus, inline assembly code is really made of patterns, not forcibly exact code. Thus, you can make put your assembly into CPP macros, and inline C functions, so anyone can use it in as any C function/macro. Inline functions resemble macros very much, but are sometimes cleaner to use. Beware that in all those cases, code will be duplicated, so only local labels (of 1: style) should be defined in that asm code. However, a macro would allow the name for a non local defined label to be passed as a parameter (or else, you should use additional meta-programming methods). Also, note that propagating inline asm code will spread potential bugs in them; so watch out doubly for register constraints in such inline asm code. Lastly, the C language itself may be considered as a good abstraction to assembly programming, which relieves you from most of the trouble of assembling. 44..11..22.. GGAASS GAS has some macro capability included, as detailed in the texinfo docs. Moreover, while GCC recognizes .s files as raw assembly to send to GAS, it also recognizes .S files as files to pipe through CPP before to feed them to GAS. Again and again, see Linux sources for examples. 44..11..33.. GGAASSPP It adds all the usual macroassembly tricks to GAS. See its texinfo docs. 44..11..44.. NNAASSMM NASM has some macro support, too. See according docs. If you have some bright idea, you might wanna contact the authors, as they are actively developing it. Meanwhile, see about external filters below. 44..11..55.. AASS8866 It has some simple macro support, but I couldn't find docs. Now the sources are very straightforward, so if you're interested, you should understand them easily. If you need more than the basics, you should use an external filter (see below). 44..11..66.. OOTTHHEERR AASSSSEEMMBBLLEERRSS +o Win32FORTH: CODE and END-CODE are normal that do not switch from interpretation mode to compilation mode, so you have access to the full power of FORTH while assembling. +o TUNES: it doesn't work yet, but the Scheme language is a real high- level language that allows arbitrary meta-programming. 44..22.. EExxtteerrnnaall FFiilltteerrss Whatever is the macro support from your assembler, or whatever language you use (even C !), if the language is not expressive enough to you, you can have files passed through an external filter with a Makefile rule like that: ______________________________________________________________________ %.s: %.S other_dependencies $(FILTER) $(FILTER_OPTIONS) < $< > $@ ______________________________________________________________________ 44..22..11.. CCPPPP CPP is truly not very expressive, but it's enough for easy things, it's standard, and called transparently by GCC. As an example of its limitations, you can't declare objects so that destructors are automatically called at the end of the declaring block; you don't have diversions or scoping, etc. CPP comes with any C compiler. However, considering how mediocre it is, stay away from it if by chance you can make it without C, 44..22..22.. MM44 M4 gives you the full power of macroprocessing, with a Turing equivalent language, recursion, regular expressions, etc. You can do with it everything that CPP cannot. See macro4th (this4th) or the Tunes 0.0.0.25 sources as examples of advanced macroprogramming using m4. However, its disfunctional quoting and unquoting semantics force you to use explicit continuation-passing tail-recursive macro style if you want to do _a_d_v_a_n_c_e_d macro programming (which is remindful of TeX -- BTW, has anyone tried to use TeX as a macroprocessor for anything else than typesetting ?). This is NOT worse than CPP that does not allow quoting and recursion anyway. The right version of m4 to get is GNU m4 1.4 (or later if exists), which has the most features and the least bugs or limitations of all. m4 is designed to be slow for anything but the simplest uses, which might still be ok for most assembly programming (you're not writing million-lines assembly programs, are you?). 44..22..33.. MMaaccrroopprroocceessssiinngg wwiitthh yyoouurr oowwnn ffiilltteerr You can write your own simple macro-expansion filter with the usual tools: perl, awk, sed, etc. That's quick to do, and you control everything. But of course, any power in macroprocessing must be earned the hard way. 44..22..44.. MMeettaapprrooggrraammmmiinngg Instead of using an external filter that expands macros, one way to do things is to write programs that write part or all of other programs. For instance, you could use a program outputting source code +o to generate sine/cosine/whatever lookup tables, +o to extract a source-form representation of a binary file, +o to compile your bitmaps into fast display routines, +o to extract documentation, initialization/finalization code, description tables, as well as normal code from the same source files, +o to have customized assembly code, generated from a perl/shell/scheme script that does arbitrary processing, +o to propagate data defined at one point only into several cross- referencing tables and code chunks. +o etc. Think about it! 44..22..44..11.. BBaacckkeennddss ffrroomm ccoommppiilleerrss Compilers like GCC, SML/NJ, Objective CAML, MIT-Scheme, CMUCL, etc, do have their own generic assembler backend, which you might choose to use, if you intend to generate code semi-automatically from the according languages, or from a language you hack: rather than write great assembly code, you may instead modify a compiler so that it dumps great assembly code! 44..22..44..22.. TThhee NNeeww--JJeerrsseeyy MMaacchhiinnee--CCooddee TToooollkkiitt There is a project, using the programming language Icon (with an experimental ML version), to build a basis for producing assembly- manipulating code. See around 44..22..44..33.. TTUUNNEESS The TUNES Project for a Free Reflective Computing System is developing its own assembler as an extension to the Scheme language, as part of its development process. It doesn't run at all yet, though help is welcome. The assembler manipulates abstract syntax trees, so it could equally serve as the basis for a assembly syntax translator, a disassembler, a common assembler/compiler back-end, etc. Also, the full power of a real language, Scheme, make it unchallenged as for macroprocessing/metaprogramming. 55.. CCAALLLLIINNGG CCOONNVVEENNTTIIOONNSS 55..11.. LLiinnuuxx 55..11..11.. LLiinnkkiinngg ttoo GGCCCC That's the preferred way. Check GCC docs and examples from Linux kernel .S files that go through gas (not those that go through as86). 32-bit arguments are pushed down stack in reverse syntactic order (hence accessed/popped in the right order), above the 32-bit near return address. %ebp, %esi, %edi, %ebx are callee-saved, other registers are caller-saved; %eax is to hold the result, or %edx:%eax for 64-bit results. FP stack: I'm not sure, but I think it's result in st(0), whole stack caller-saved. Note that GCC has options to modify the calling conventions by reserving registers, having arguments in registers, not assuming the FPU, etc. Check the i386 .info pages. Beware that you must then declare the cdecl or regparm(0) attribute for a function that will follow standard GCC calling conventions. See in the GCC info pages the section: C Extensions::Extended Asm::. See also how Linux defines its asmlinkage macro... 55..11..22.. EELLFF vvss aa..oouutt pprroobblleemmss Some C compilers prepend an underscore before every symbol, while others do not. Particularly, Linux a.out GCC does such prepending, while Linux ELF GCC does not. If you need cope with both behaviors at once, see how existing packages do. For instance, get an old Linux source tree, the Elk, qthreads, or OCaml... You can also override the implicit C->asm renaming by inserting statements like ______________________________________________________________________ void foo asm("bar") (void); ______________________________________________________________________ to be sure that the C function foo will be called really bar in assem- bly. Note that the utility objcopy, from the binutils package, should allow you to transform your a.out objects into ELF objects, and perhaps the contrary too, in some cases. More generally, it will do lots of file format conversions. 55..11..33.. DDiirreecctt LLiinnuuxx ssyyssccaallllss Often you will be told that using libc is the only way, and direct system calls are bad. Believe it, _u_n_l_e_s_s of course you're specifically writing your own replacement for the libc, adapted to your specific language or memory requirements or whatever. But you must know that libc is not sacred, and in _m_o_s_t cases libc only does some checks, then calls kernel, and then sets errno. You can easily do this in your program as well (if you need to), and your program will be dozen times smaller, and this will also result in improved performance, just because you're not using shared libraries (static binaries are faster). Using or not using libc in assembly programming is more a question of taste/belief than something practical. Remember, Linux is aiming to be POSIX compliant, so does libc. This means that syntax of almost all libc "system calls" exactly matches syntax of real kernel system calls (and vice versa). Besides, modern libc becomes slower and slower, and eats more and more memory, and so, cases of using direct system calls become quite usual. But.. main drawback of throwing libc away is that possibly you will need to implement several libc specific functions (that are not just syscall wrappers) on your own (printf and Co.).. and you are ready for that, aren't you? :) Here is summary of direct system calls pros and cons. Pros: +o smallest possible size; squeezing the last byte out of the system. +o highest possible speed; squeezing cycles out of your favorite benchmark. +o no pollution by libc cruft. +o no pollution by C calling conventions (if you're developing your own language or environment). +o static binaries make you independent from libc upgrades or crashes, or from dangling #! path to a interpreter (and are faster). +o just for the fun out of it (don't you get a kick out of assembly programming?) Cons: +o If any other program on your computer uses the libc, then duplicating the libc code will actually waste memory, not save it. +o Size is much better saved by having some kind of bytecode, wordcode, or structure interpreter than by writing everything in assembly. (the interpreter itself could be written either in C or assembly.) +o Services redundantly implemented in many static binaries are a waste of memory. But you can put your libc replacement in a shared library. +o The best way to keep multiple binaries small is to not have multiple binaries, but instead to have an interpreter process files with #! prefix. This is how OCaml works when used in wordcode mode (as opposed to optimized native code mode), and it is compatible with using the libc. This is also how Tom Christiansen's Perl PowerTools reimplementation of unix utilities works. Finally, one last way to keep things small, that doesn't depend on an external file with a hardcoded path, be it library or interpreter, is to have only one binary, and have multiply-named hard or soft links to it: the same binary will provide everything you need in an optimal space, with no redundancy of subroutines or useless binary headers; it will dispatch its specific behavior according to its argv[0]; in case it isn't called with a recognized name, it might default to a shell, and be possibly thus also usable as an interpreter! +o You cannot benefit from the many functionalities that libc provides besides mere linux syscalls: that is, functionality described in section 3 of the manual pages, as opposed to section 2, such as malloc, threads, locale, password, high-level network management, etc. +o Consequently, you might have to reimplement large parts of libc, from printf to malloc and gethostbyname. It's redundant with the libc effort, and can be _q_u_i_t_e boring sometimes. Note that some people have already reimplemented "light" replacements for parts of the libc -- check them out! ( Rick Hohensee's libsys , ``asmutils'' project is working on pure assembly libc, can anyone send more pointers?) +o Static libraries prevent your benefitting from libc upgrades as well as from libc add-ons such as the zlibc package, that does on- the-fly transparent decompression of gzip-compressed files. +o The few instructions added by the libc are a _r_i_d_i_c_u_l_o_u_s_l_y small speed overhead as compared to the cost of a system call. If speed is a concern, your main problem is in your usage of system calls, not in their wrapper's implementation. +o Using the standard assembly API for system calls is much slower than using the libc API when running in micro-kernel versions of Linux such as L4Linux, that have their own faster calling convention, and pay high convention-translation overhead when using the standard one (L4Linux comes with libc recompiled with their syscall API; of course, you could recompile your code with their API, too). +o See previous discussion for general speed optimization issue. +o If syscalls are too slow to you, you might want to hack the kernel sources (in C) instead of staying in userland. If you've pondered the above pros and cons, and still want to use direct syscalls (as documented in section 2 of the manual pages), then here is some advice. +o You can easily define your system calling functions in a portable way in C (as opposed to unportable using assembly), by including , and using provided macros. +o Since you're trying to replace it, go get the sources for the libc, and grok them. (And if you think you can do better, then send feedback to the authors!) +o As an example of pure assembly code that does everything you want, examine ``Linux Assembly Projects''. Basically, you issue an int 0x80, with the __NR_syscallname number (from asm/unistd.h) in eax, and parameters (up to five) in ebx, ecx, edx, esi, edi respectively. Result is returned in eax, with a negative result being an error, whose opposite is what libc would put in errno. The user-stack is not touched, so you needn't have a valid one when doing a syscall. As for the invocation arguments passed to a process upon startup, the general principle is that the stack originally contains the number of arguments argc, then the list of pointers that constitute *argv, then a null-terminated sequence of null-terminated variable=value strings for the environment. For more details, do examine ``Linux assembly resources'', read the sources of C startup code from your libc (crt0.S or crt1.S), or those from the Linux kernel (exec.c and binfmt_*.c in linux/fs/). 55..11..44.. HHaarrddwwaarree II//OO uunnddeerr LLiinnuuxx If you want to do direct I/O under Linux, either it's something very simple that needn't OS arbitration, and you should see the IO-Port- Programming mini-HOWTO; or it needs a kernel device driver, and you should try to learn more about kernel hacking, device driver development, kernel modules, etc, for which there are other excellent HOWTOs and documents from the LDP. Particularly, if what you want is Graphics programming, then do join one of the GGI or XFree86 projects. Some people have even done better, writing small and robust XFree86 drivers in an interpreted domain-specific language, GAL , and achieving the efficiency of hand C-written drivers through partial evaluation (drivers not only not in asm, but not even in C!). The problem is that the partial evaluator they used to achieve efficiency is not free software. Any taker for a replacement? Anyway, in all these cases, you'll be better when using GCC inline assembly with the macros from linux/asm/*.h than writing full assembly source files. 55..11..55.. AAcccceessssiinngg 1166--bbiitt ddrriivveerrss ffrroomm LLiinnuuxx//ii338866 Such thing is theoretically possible (proof: see how DOSEMU can selectively grant hardware port access to programs), and I've heard rumors that someone somewhere did actually do it (in the PCI driver? Some VESA access stuff? ISA PnP? dunno). If you have some more precise information on that, you'll be most welcome. Anyway, good places to look for more information are the Linux kernel sources, DOSEMU sources (and other programs in the DOSEMU repository ), and sources for various low-level programs under Linux... (perhaps GGI if it supports VESA). Basically, you must either use 16-bit protected mode or vm86 mode. The first is simpler to setup, but only works with well-behaved code that won't do any kind of segment arithmetics or absolute segment addressing (particularly addressing segment 0), unless by chance it happens that all segments used can be setup in advance in the LDT. The later allows for more "compatibility" with vanilla 16-bit environments, but requires more complicated handling. In both cases, before you can jump to 16-bit code, you must +o mmap any absolute address used in the 16-bit code (such as ROM, video buffers, DMA targets, and memory-mapped I/O) from /dev/mem to your process' address space, +o setup the LDT and/or vm86 mode monitor. +o grab proper I/O permissions from the kernel (see the above section) Again, carefully read the source for the stuff contributed to the DOSEMU project, particularly these mini-emulators for running ELKS and/or simple .COM programs under Linux/i386. 55..22.. DDOOSS Most DOS extenders come with some interface to DOS services. Read their docs about that, but often, they just simulate int 0x21 and such, so you do "as if" you are in real mode (I doubt they have more than stubs and extend things to work with 32-bit operands; they most likely will just reflect the interrupt into the real-mode or vm86 handler). Docs about DPMI (and much more) can be found on (again, the original x2ftp site is closing (no more?), so use a mirror site ). DJGPP comes with its own (limited) glibc derivative/subset/replacement, too. It is possible to cross-compile from Linux to DOS, see the devel/msdos/ directory of your local FTP mirror for metalab.unc.edu Also see the MOSS dos-extender from the Flux project from university of Utah. Other documents and FAQs are more DOS-centered. We do not recommend DOS development. 55..33.. WWiinnddoowwss aanndd CCoo.. This HOWTO is not about Windows programming, you can find lots of documents about it everywhere.. The thing you should know is that Cygnus Solutions developed the cygwin32.dll library, for GNU programs to run on Win32 platform. Thus, you can use GCC, GAS, all the GNU tools, and many other Unix applications. Take a look on their webpage. 55..44.. YYoouurr oowwnn OOSS Control is what attracts many OS developers to assembly, often is what leads to or stems from assembly hacking. Note that any system that allows self-development could be qualified an "OS", though it can run "on the top" of an underlying system (much like Linux over Mach or OpenGenera over Unix). Hence, for easier debugging purpose, you might like to develop your "OS" first as a process running on top of Linux (despite the slowness), then use the Flux OS kit (which grants use of Linux and BSD drivers in your own OS) to make it standalone. When your OS is stable, it is time to write your own hardware drivers if you really love that. This HOWTO will not cover topics such as Boot loader code & getting into 32-bit mode, Handling Interrupts, The basics about Intel protected mode or V86/R86 braindeadness, defining your object format and calling conventions. The main place where to find reliable information about that all, is source code of existing OSes and bootloaders. Lots of pointers are on the following webpage: 66.. QQUUIICCKK SSTTAARRTT Finally, if you still want to try this crazy idea and write something in assembly (if you've reached this section -- you're real assembly fan), I'll herein provide what you will need to get started. As you've read before, you can write for Linux in different ways; I'll show example of using pure system calls. This means that we will not use libc at all, the only thing required for our program to run is kernel. Our code will not be linked to any library, will not use ELF interpreter -- it will communicate directly with kernel. I will show the same sample program in two assemblers, nasm and gas, thus showing Intel and AT&T syntax. 66..11.. TToooollss yyoouu nneeeedd First of all you need assembler (compiler): NASM or GAS. Second, you need linker: ld, assembler produces only object code. Almost all distributions include GAS and ld, in binutils package. As for NASM, you may have to download and install binary packages for Linux and docs from ``NASM webpage''; however, several distributions (Stampede, Debian, SuSe) already include it, check first. If you are going to dig in, you should also install kernel source. I assume that you are using at least Linux 2.0 and ELF. 66..22.. HHeelllloo,, wwoorrlldd!! Linux/i386 is 32bit and has flat memory model. A program can be divided into sections. Main sections are _._t_e_x_t for your code, _._d_a_t_a for your data, _._b_s_s for undefined data. Program must have at least _._t_e_x_t section. Now we will write our first program. Here is sample code: 66..22..11.. NNAASSMM ((hheelllloo..aassmm)) ______________________________________________________________________ section .data ;section declaration msg db "Hello, world!",0xA ;our dear string len equ $ - msg ;length of our dear string section .text ;section declaration ;we must export the entry point to the ELF linker or global _start ;loader. They conventionally recognize _start as their ;entry point. Use ld -e foo to override the default. _start: ;write our string to stdout mov eax,4 ;system call number (sys_write) mov ebx,1 ;first argument: file handle (stdout) mov ecx,msg ;second argument: pointer to message to write mov edx,len ;third argument: message length int 0x80 ;call kernel ;and exit mov eax,1 ;system call number (sys_exit) xor ebx,ebx ;first syscall argument: exit code int 0x80 ;call kernel ______________________________________________________________________ 66..22..22.. GGAASS ((hheelllloo..SS)) ______________________________________________________________________ .data # section declaration msg: .string "Hello, world!\n" # our dear string len = . - msg # length of our dear string .text # section declaration # we must export the entry point to the ELF linker or .global _start # loader. They conventionally recognize _start as their # entry point. Use ld -e foo to override the default. _start: # write our string to stdout movl $4,%eax # system call number (sys_write) movl $1,%ebx # first argument: file handle (stdout) movl $msg,%ecx # second argument: pointer to message to write movl $len,%edx # third argument: message length int $0x80 # call kernel # and exit movl $1,%eax # system call number (sys_exit) xorl %ebx,%ebx # first syscall argument: exit code int $0x80 # call kernel ______________________________________________________________________ 66..33.. PPrroodduucciinngg oobbjjeecctt ccooddee First step of building binary is producing object file from source, by invoking assembler; we must issue the following: For NASM example: $ nasm -f elf hello.asm For GAS example: $ as -o hello.o hello.S This will produce hello.o object file. 66..44.. PPrroodduucciinngg eexxeeccuuttaabbllee Second step is producing executable file itself from object file, by invoking linker: $ ld -s -o hello hello.o This will finally build hello ELF binary. Hey, try to run it... Works? That's it. Pretty simple. If you get interested and want to proceed further, you may want to look through ``Linux assembly projects'', that contain PLENTY of source code and examples. 77.. RREESSOOUURRCCEESS You main resource for Linux assembly programming material is Linux Assembly :). Here are some resources listed there. 77..11.. SSooffttwwaarree pprroojjeeccttss +o asmutils (unix utils written with NASM) +o ec64 (C64 emulator) +o ELF kickers & tiny Linux executables +o ASMIX (unix utils in written with GAS) +o eforth 1.0e 77..22.. TTuuttoorriiaallss +o Introduction to Linux/i386 Assembly Programming +o Linux assembly tutorial (GAS and GDB related) +o A Whirlwind Tutorial on Creating Really Teensy ELF Executables for Linux +o Jan's Linux & Assembler HomePage (mostly about assembly programming with libc) 77..33.. MMaaiilliinngg lliisstt If you're are interested in Linux assembly programming (or have questions, or are just curious) I especially invite you to join Linux assembly programming mailing list. List address is . To subscribe send a blank message to . List archives are available at . 77..44.. BBooookkss Unfortunately there are no ready books I can recommend on the topic. However I'm in the progress of writing a book "Linux Assembly Programming", which [hopefully] will be published somewhere in 2000. 77..55.. SSoommeehhooww rreellaatteedd pprroojjeeccttss +o NASM (portable x86 assembler with Intel syntax) +o BIEW (portable console hex viewer/editor with built-in disassembler) +o UPX (portable executable packer for several formats) +o Intel2gas (converter between AT&T and Intel assembler syntax) +o A2I (converter from AT&T to NASM Intel assembler syntax) +o Assembly Programming Journal (has articles on Linux assembly) 77..66.. GGeenneerraall ppooiinntteerrss +o The Art Of Assembly +o x86 assembly FAQ +o ftp.luth.se mirrors the hornet and x2ftp former archives of msdos assembly coding stuff +o CPU family references: Intel , AMD ; x86 bugs ; Alpha ; PowerPC assembly +o Fun stuff: CoreWars , a fun way to learn assembly in general +o USENET: comp.lang.asm.x86 ; alt.os.assembly _N_o_t_e_: _i_f _y_o_u _k_n_o_w _a_n_y Linux assembly related project/webpage not mentioned here, please send me the url.