2 \# NASM revision history in nasmdoc format
5 \H{cl-2.xx} NASM 2 Series
7 The NASM 2 series support x86-64, and is the production version of NASM
11 \S{cl-2.06} Version 2.06
13 \b Support for indirect macro expansion (\c{%[...]}). See \k{indmacro}.
15 \b \c{%pop} can now take an argument, see \k{pushpop}.
17 \b The argument to \c{%use} is no longer macro-expanded. Use
18 \c{%[...]} if macro expansion is desired.
20 \b Support for thread-local storage in ELF32 and ELF64. See \k{elftls}.
22 \b Fix crash on \c{%ifmacro} without an argument.
24 \b Correct the arguments to the \c{POPCNT} instruction.
26 \b Fix section alignment in the Mach-O format.
28 \b Update AVX support to version 5 of the Intel specification.
30 \b Fix the handling of accesses to context-local macros from higher
31 levels in the context stack.
33 \b Treat \c{WAIT} as a prefix rather than as an instruction, thereby
34 allowing constructs like \c{O16 FSAVE} to work correctly.
37 \S{cl-2.05.01} Version 2.05.01
39 \b Fix the \c{-w}/\c{-W} option parsing, which was broken in NASM 2.05.
42 \S{cl-2.05} Version 2.05
44 \b Fix redundant REX.W prefix on \c{JMP reg64}.
46 \b Make the behaviour of \c{-O0} match NASM 0.98 legacy behavior.
49 \b \c{-w-user} can be used to suppress the output of \c{%warning} directives.
52 \b Fix bug where \c{ALIGN} would issue a full alignment datum instead of
55 \b Fix offsets in list files.
57 \b Fix \c{%include} inside multi-line macros or loops.
59 \b Fix error where NASM would generate a spurious warning on valid
60 optimizations of immediate values.
62 \b Fix arguments to a number of the \c{CVT} SSE instructions.
64 \b Fix RIP-relative offsets when the instruction carries an immediate.
66 \b Massive overhaul of the ELF64 backend for spec compliance.
68 \b Fix the Geode \c{PFRCPV} and \c{PFRSQRTV} instruction.
70 \b Fix the SSE 4.2 \c{CRC32} instruction.
73 \S{cl-2.04} Version 2.04
75 \b Sanitize macro handing in the \c{%error} directive.
77 \b New \c{%warning} directive to issue user-controlled warnings.
79 \b \c{%error} directives are now deferred to the final assembly phase.
81 \b New \c{%fatal} directive to immediately terminate assembly.
83 \b New \c{%strcat} directive to join quoted strings together.
85 \b New \c{%use} macro directive to support standard macro directives. See
88 \b Excess default parameters to \c{%macro} now issues a warning by default.
91 \b Fix \c{%ifn} and \c{%elifn}.
93 \b Fix nested \c{%else} clauses.
95 \b Correct the handling of nested \c{%rep}s.
97 \b New \c{%unmacro} directive to undeclare a multi-line macro.
100 \b Builtin macro \c{__PASS__} which expands to the current assembly pass.
103 \b \c{__utf16__} and \c{__utf32__} operators to generate UTF-16 and UTF-32
104 strings. See \k{unicode}.
106 \b Fix bug in case-insensitive matching when compiled on platforms that
107 don't use the \c{configure} script. Of the official release binaries,
108 that only affected the OS/2 binary.
110 \b Support for x87 packed BCD constants. See \k{bcdconst}.
112 \b Correct the \c{LTR} and \c{SLDT} instructions in 64-bit mode.
114 \b Fix unnecessary REX.W prefix on indirect jumps in 64-bit mode.
116 \b Add AVX versions of the AES instructions (\c{VAES}...).
118 \b Fix the 256-bit FMA instructions.
120 \b Add 256-bit AVX stores per the latest AVX spec.
122 \b VIA XCRYPT instructions can now be written either with or without
123 \c{REP}, apparently different versions of the VIA spec wrote them
126 \b Add missing 64-bit \c{MOVNTI} instruction.
128 \b Fix the operand size of \c{VMREAD} and \c{VMWRITE}.
130 \b Numerous bug fixes, especially to the AES, AVX and VTX instructions.
132 \b The optimizer now always runs until it converges. It also runs even
133 when disabled, but doesn't optimize. This allows most forward references
134 to be resolved properly.
136 \b \c{%push} no longer needs a context identifier; omitting the context
137 identifier results in an anonymous context.
140 \S{cl-2.03.01} Version 2.03.01
142 \b Fix buffer overflow in the listing module.
144 \b Fix the handling of hexadecimal escape codes in `...` strings.
146 \b The Postscript/PDF documentation has been reformatted.
148 \b The \c{-F} option now implies \c{-g}.
151 \S{cl-2.03} Version 2.03
153 \b Add support for Intel AVX, CLMUL and FMA instructions,
154 including YMM registers.
156 \b \c{dy}, \c{resy} and \c{yword} for 32-byte operands.
158 \b Fix some SSE5 instructions.
160 \b Intel \c{INVEPT}, \c{INVVPID} and \c{MOVBE} instructions.
162 \b Fix checking for critical expressions when the optimizer is enabled.
164 \b Support the DWARF debugging format for ELF targets.
166 \b Fix optimizations of signed bytes.
168 \b Fix operation on bigendian machines.
170 \b Fix buffer overflow in the preprocessor.
172 \b \c{SAFESEH} support for Win32, \c{IMAGEREL} for Win64 (SEH).
174 \b \c{%?} and \c{%??} to refer to the name of a macro itself. In particular,
175 \c{%idefine keyword $%?} can be used to make a keyword "disappear".
177 \b New options for dependency generation: \c{-MD}, \c{-MF},
178 \c{-MP}, \c{-MT}, \c{-MQ}.
180 \b New preprocessor directives \c{%pathsearch} and \c{%depend}; INCBIN
181 reimplemented as a macro.
183 \b \c{%include} now resolves macros in a sane manner.
185 \b \c{%substr} can now be used to get other than one-character substrings.
187 \b New type of character/string constants, using backquotes (\c{`...`}),
188 which support C-style escape sequences.
190 \b \c{%defstr} and \c{%idefstr} to stringize macro definitions before
193 \b Fix forward references used in \c{EQU} statements.
195 \S{cl-2.02} Version 2.02
198 \b Additional fixes for MMX operands with explicit \c{qword}, as well as
199 (hopefully) SSE operands with \c{oword}.
201 \b Fix handling of truncated strings with \c{DO}.
203 \b Fix segfaults due to memory overwrites when floating-point constants
206 \b Fix segfaults due to missing include files.
208 \b Fix OpenWatcom Makefiles for DOS and OS/2.
210 \b Add autogenerated instruction list back into the documentation.
212 \b ELF: Fix segfault when generating stabs, and no symbols have been
215 \b ELF: Experimental support for DWARF debugging information.
217 \b New compile date and time standard macros.
219 \b \c{%ifnum} now returns true for negative numbers.
221 \b New \c{%iftoken} test for a single token.
223 \b New \c{%ifempty} test for empty expansion.
225 \b Add support for the \c{XSAVE} instruction group.
227 \b Makefile for Netware/gcc.
229 \b Fix issue with some warnings getting emitted way too many times.
231 \b Autogenerated instruction list added to the documentation.
233 \S{cl-2.01} Version 2.01
236 \b Fix the handling of MMX registers with explicit \c{qword} tags on
237 memory (broken in 2.00 due to 64-bit changes.)
239 \b Fix the PREFETCH instructions.
241 \b Fix the documentation.
243 \b Fix debugging info when using \c{-f elf}
244 (backwards compatibility alias for \c{-f elf32}).
246 \b Man pages for rdoff tools (from the Debian project.)
248 \b ELF: handle large numbers of sections.
250 \b Fix corrupt output when the optimizer runs out of passes.
253 \S{cl-2.00} Version 2.00
255 \b Added c99 data-type compliance.
257 \b Added general x86-64 support.
259 \b Added win64 (x86-64 COFF) output format.
261 \b Added \c{__BITS__} standard macro.
263 \b Renamed the \c{elf} output format to \c{elf32} for clarity.
265 \b Added \c{elf64} and \c{macho} (MacOS X) output formats.
267 \b Added Numeric constants in \c{dq} directive.
269 \b Added \c{oword}, \c{do} and \c{reso} pseudo operands.
271 \b Allow underscores in numbers.
273 \b Added 8-, 16- and 128-bit floating-point formats.
275 \b Added binary, octal and hexadecimal floating-point.
277 \b Correct the generation of floating-point constants.
279 \b Added floating-point option control.
281 \b Added Infinity and NaN floating point support.
283 \b Added ELF Symbol Visibility support.
285 \b Added setting OSABI value in ELF header directive.
287 \b Added Generate Makefile Dependencies option.
289 \b Added Unlimited Optimization Passes option.
291 \b Added \c{%IFN} and \c{%ELIFN} support.
293 \b Added Logical Negation Operator.
295 \b Enhanced Stack Relative Preprocessor Directives.
297 \b Enhanced ELF Debug Formats.
299 \b Enhanced Send Errors to a File option.
301 \b Added SSSE3, SSE4.1, SSE4.2, SSE5 support.
303 \b Added a large number of additional instructions.
305 \b Significant performance improvements.
307 \b \c{-w+warning} and \c{-w-warning} can now be written as -Wwarning and
308 -Wno-warning, respectively. See \k{opt-w}.
310 \b Add \c{-w+error} to treat warnings as errors. See \k{opt-w}.
312 \b Add \c{-w+all} and \c{-w-all} to enable or disable all suppressible
313 warnings. See \k{opt-w}.
316 \H{cl-0.98.xx} NASM 0.98 Series
318 The 0.98 series was the production versions of NASM from 1999 to 2007.
321 \S{cl-0.98.39} Version 0.98.39
323 \b fix buffer overflow
325 \b fix outas86's \c{.bss} handling
327 \b "make spotless" no longer deletes config.h.in.
329 \b \c{%(el)if(n)idn} insensitivity to string quotes difference (#809300).
331 \b (nasm.c)\c{__OUTPUT_FORMAT__} changed to string value instead of symbol.
333 \S{cl-0.98.38} Version 0.98.38
336 \b Add Makefile for 16-bit DOS binaries under OpenWatcom, and modify
337 \c{mkdep.pl} to be able to generate completely pathless dependencies, as
338 required by OpenWatcom wmake (it supports path searches, but not
341 \b Fix the \c{STR} instruction.
343 \b Fix the ELF output format, which was broken under certain
344 circumstances due to the addition of stabs support.
346 \b Quick-fix Borland format debug-info for \c{-f obj}
348 \b Fix for \c{%rep} with no arguments (#560568)
350 \b Fix concatenation of preprocessor function call (#794686)
352 \b Fix long label causes coredump (#677841)
354 \b Use autoheader as well as autoconf to keep configure from generating
355 ridiculously long command lines.
357 \b Make sure that all of the formats which support debugging output
358 actually will suppress debugging output when \c{-g} not specified.
360 \S{cl-0.98.37} Version 0.98.37
363 \b Paths given in \c{-I} switch searched for \c{incbin}-ed as
364 well as \c{%include}-ed files.
366 \b Added stabs debugging for the ELF output format, patch from
369 \b Fix \c{output/outbin.c} to allow origin > 80000000h.
371 \b Make \c{-U} switch work.
373 \b Fix the use of relative offsets with explicit prefixes, e.g.
376 \b Remove \c{backslash()}.
378 \b Fix the \c{SMSW} and \c{SLDT} instructions.
380 \b \c{-O2} and \c{-O3} are no longer aliases for \c{-O10} and \c{-O15}.
381 If you mean the latter, please say so! :)
383 \S{cl-0.98.36} Version 0.98.36
386 \b Update rdoff - librarian/archiver - common rec - docs!
388 \b Fix signed/unsigned problems.
390 \b Fix \c{JMP FAR label} and \c{CALL FAR label}.
392 \b Add new multisection support - map files - fix align bug
394 \b Fix sysexit, movhps/movlps reg,reg bugs in insns.dat
396 \b \c{Q} or \c{O} suffixes indicate octal
398 \b Support Prescott new instructions (PNI).
400 \b Cyrix \c{XSTORE} instruction.
403 \S{cl-0.98.35} Version 0.98.35
405 \b Fix build failure on 16-bit DOS (Makefile.bc3 workaround for compiler bug.)
407 \b Fix dependencies and compiler warnings.
409 \b Add "const" in a number of places.
411 \b Add -X option to specify error reporting format (use -Xvc to
412 integrate with Microsoft Visual Studio.)
414 \b Minor changes for code legibility.
416 \b Drop use of tmpnam() in rdoff (security fix.)
419 \S{cl-0.98.34} Version 0.98.34
421 \b Correct additional address-size vs. operand-size confusions.
423 \b Generate dependencies for all Makefiles automatically.
425 \b Add support for unimplemented (but theoretically available)
426 registers such as tr0 and cr5. Segment registers 6 and 7 are called
427 segr6 and segr7 for the operations which they can be represented.
429 \b Correct some disassembler bugs related to redundant address-size prefixes.
430 Some work still remains in this area.
432 \b Correctly generate an error for things like "SEG eax".
434 \b Add the JMPE instruction, enabled by "CPU IA64".
436 \b Correct compilation on newer gcc/glibc platforms.
438 \b Issue an error on things like "jmp far eax".
441 \S{cl-0.98.33} Version 0.98.33
443 \b New __NASM_PATCHLEVEL__ and __NASM_VERSION_ID__ standard macros to
444 round out the version-query macros. version.pl now understands
445 X.YYplWW or X.YY.ZZplWW as a version number, equivalent to
446 X.YY.ZZ.WW (or X.YY.0.WW, as appropriate).
448 \b New keyword "strict" to disable the optimization of specific
451 \b Fix the handing of size overrides with JMP instructions
452 (instructions such as "jmp dword foo".)
454 \b Fix the handling of "ABSOLUTE label", where "label" points into a
457 \b Fix OBJ output format with lots of externs.
459 \b More documentation updates.
461 \b Add -Ov option to get verbose information about optimizations.
463 \b Undo a braindead change which broke \c{%elif} directives.
468 \S{cl-0.98.32} Version 0.98.32
470 \b Fix NASM crashing when \c{%macro} directives were left unterminated.
472 \b Lots of documentation updates.
474 \b Complete rewrite of the PostScript/PDF documentation generator.
476 \b The MS Visual C++ Makefile was updated and corrected.
478 \b Recognize .rodata as a standard section name in ELF.
480 \b Fix some obsolete Perl4-isms in Perl scripts.
482 \b Fix configure.in to work with autoconf 2.5x.
484 \b Fix a couple of "make cleaner" misses.
486 \b Make the normal "./configure && make" work with Cygwin.
489 \S{cl-0.98.31} Version 0.98.31
491 \b Correctly build in a separate object directory again.
493 \b Derive all references to the version number from the version file.
495 \b New standard macros __NASM_SUBMINOR__ and __NASM_VER__ macros.
497 \b Lots of Makefile updates and bug fixes.
499 \b New \c{%ifmacro} directive to test for multiline macros.
501 \b Documentation updates.
503 \b Fixes for 16-bit OBJ format output.
505 \b Changed the NASM environment variable to NASMENV.
508 \S{cl-0.98.30} Version 0.98.30
510 \b Changed doc files a lot: completely removed old READMExx and
511 Wishlist files, incorporating all information in CHANGES and TODO.
513 \b I waited a long time to rename zoutieee.c to (original) outieee.c
515 \b moved all output modules to output/ subdirectory.
517 \b Added 'make strip' target to strip debug info from nasm & ndisasm.
519 \b Added INSTALL file with installation instructions.
521 \b Added -v option description to nasm man.
523 \b Added dist makefile target to produce source distributions.
525 \b 16-bit support for ELF output format (GNU extension, but useful.)
528 \S{cl-0.98.28} Version 0.98.28
530 \b Fastcooked this for Debian's Woody release:
531 Frank applied the INCBIN bug patch to 0.98.25alt and called
532 it 0.98.28 to not confuse poor little apt-get.
535 \S{cl-0.98.26} Version 0.98.26
537 \b Reorganised files even better from 0.98.25alt
540 \S{cl-0.98.25alt} Version 0.98.25alt
542 \b Prettified the source tree. Moved files to more reasonable places.
544 \b Added findleak.pl script to misc/ directory.
546 \b Attempted to fix doc.
549 \S{cl-0.98.25} Version 0.98.25
551 \b Line continuation character \c{\\}.
553 \b Docs inadvertantly reverted - "dos packaging".
556 \S{cl-0.98.24p1} Version 0.98.24p1
558 \b FIXME: Someone, document this please.
561 \S{cl-0.98.24} Version 0.98.24
563 \b Documentation - Ndisasm doc added to Nasm.doc.
566 \S{cl-0.98.23} Version 0.98.23
568 \b Attempted to remove rdoff version1
570 \b Lino Mastrodomenico's patches to preproc.c (%$$ bug?).
573 \S{cl-0.98.22} Version 0.98.22
575 \b Update rdoff2 - attempt to remove v1.
578 \S{cl-0.98.21} Version 0.98.21
580 \b Optimization fixes.
583 \S{cl-0.98.20} Version 0.98.20
585 \b Optimization fixes.
588 \S{cl-0.98.19} Version 0.98.19
590 \b H. J. Lu's patch back out.
593 \S{cl-0.98.18} Version 0.98.18
595 \b Added ".rdata" to "-f win32".
598 \S{cl-0.98.17} Version 0.98.17
600 \b H. J. Lu's "bogus elf" patch. (Red Hat problem?)
603 \S{cl-0.98.16} Version 0.98.16
605 \b Fix whitespace before "[section ..." bug.
608 \S{cl-0.98.15} Version 0.98.15
610 \b Rdoff changes (?).
612 \b Fix fixes to memory leaks.
615 \S{cl-0.98.14} Version 0.98.14
620 \S{cl-0.98.13} Version 0.98.13
622 \b There was no 0.98.13
625 \S{cl-0.98.12} Version 0.98.12
627 \b Update optimization (new function of "-O1")
629 \b Changes to test/bintest.asm (?).
632 \S{cl-0.98.11} Version 0.98.11
634 \b Optimization changes.
639 \S{cl-0.98.10} Version 0.98.10
641 \b There was no 0.98.10
644 \S{cl-0.98.09} Version 0.98.09
646 \b Add multiple sections support to "-f bin".
648 \b Changed GLOBAL_TEMP_BASE in outelf.c from 6 to 15.
650 \b Add "-v" as an alias to the "-r" switch.
652 \b Remove "#ifdef" from Tasm compatibility options.
654 \b Remove redundant size-overrides on "mov ds, ex", etc.
656 \b Fixes to SSE2, other insns.dat (?).
658 \b Enable uppercase "I" and "P" switches.
660 \b Case insinsitive "seg" and "wrt".
662 \b Update install.sh (?).
664 \b Allocate tokens in blocks.
666 \b Improve "invalid effective address" messages.
669 \S{cl-0.98.08} Version 0.98.08
671 \b Add "\c{%strlen}" and "\c{%substr}" macro operators
673 \b Fixed broken c16.mac.
675 \b Unterminated string error reported.
677 \b Fixed bugs as per 0.98bf
680 \S{cl-0.98.09b with John Coffman patches released 28-Oct-2001} Version 0.98.09b with John Coffman patches released 28-Oct-2001
682 Changes from 0.98.07 release to 98.09b as of 28-Oct-2001
684 \b More closely compatible with 0.98 when -O0 is implied
685 or specified. Not strictly identical, since backward
686 branches in range of short offsets are recognized, and signed
687 byte values with no explicit size specification will be
688 assembled as a single byte.
690 \b More forgiving with the PUSH instruction. 0.98 requires
691 a size to be specified always. 0.98.09b will imply the size
692 from the current BITS setting (16 or 32).
694 \b Changed definition of the optimization flag:
696 -O0 strict two-pass assembly, JMP and Jcc are
697 handled more like 0.98, except that back-
698 ward JMPs are short, if possible.
700 -O1 strict two-pass assembly, but forward
701 branches are assembled with code guaranteed
702 to reach; may produce larger code than
703 -O0, but will produce successful assembly
704 more often if branch offset sizes are not
707 -O2 multi-pass optimization, minimize branch
708 offsets; also will minimize signed immed-
709 iate bytes, overriding size specification.
711 -O3 like -O2, but more passes taken, if needed
714 \S{cl-0.98.07 released 01/28/01} Version 0.98.07 released 01/28/01
716 \b Added Stepane Denis' SSE2 instructions to a *working*
717 version of the code - some earlier versions were based on
718 broken code - sorry 'bout that. version "0.98.07"
724 \b Cosmetic modifications to nasm.c, nasm.h,
728 \S{cl-0.98.06f released 01/18/01} Version 0.98.06f released 01/18/01
731 \b - Add "metalbrain"s jecxz bug fix in insns.dat
732 - alter nasmdoc.src to match - version "0.98.06f"
735 \S{cl-0.98.06e released 01/09/01} Version 0.98.06e released 01/09/01
738 \b Removed the "outforms.h" file - it appears to be
739 someone's old backup of "outform.h". version "0.98.06e"
743 \b fbk - finally added the fix for the "multiple %includes bug",
744 known since 7/27/99 - reported originally (?) and sent to
745 us by Austin Lunnen - he reports that John Fine had a fix
746 within the day. Here it is...
748 \b Nelson Rush resigns from the group. Big thanks to Nelson for
749 his leadership and enthusiasm in getting these changes
750 incorporated into Nasm!
752 \b fbk - [list +], [list -] directives - ineptly implemented, should
753 be re-written or removed, perhaps.
755 \b Brian Raiter / fbk - "elfso bug" fix - applied to aoutb format
756 as well - testing might be desirable...
760 \b James Seter - -postfix, -prefix command line switches.
762 \b Yuri Zaporogets - rdoff utility changes.
765 \S{cl-0.98p1} Version 0.98p1
767 \b GAS-like palign (Panos Minos)
769 \b FIXME: Someone, fill this in with details
772 \S{cl-0.98bf (bug-fixed)} Version 0.98bf (bug-fixed)
774 \b Fixed - elf and aoutb bug - shared libraries
775 - multiple "%include" bug in "-f obj"
777 - unrecognized option bug in ndisasm
779 \S{cl-0.98.03 with John Coffman's changes released 27-Jul-2000} Version 0.98.03 with John Coffman's changes released 27-Jul-2000
781 \b Added signed byte optimizations for the 0x81/0x83 class
782 of instructions: ADC, ADD, AND, CMP, OR, SBB, SUB, XOR:
783 when used as 'ADD reg16,imm' or 'ADD reg32,imm.' Also
784 optimization of signed byte form of 'PUSH imm' and 'IMUL
785 reg,imm'/'IMUL reg,reg,imm.' No size specification is needed.
787 \b Added multi-pass JMP and Jcc offset optimization. Offsets
788 on forward references will preferentially use the short form,
789 without the need to code a specific size (short or near) for
790 the branch. Added instructions for 'Jcc label' to use the
791 form 'Jnotcc $+3/JMP label', in cases where a short offset
792 is out of bounds. If compiling for a 386 or higher CPU, then
793 the 386 form of Jcc will be used instead.
795 This feature is controlled by a new command-line switch: "O",
796 (upper case letter O). "-O0" reverts the assembler to no
797 extra optimization passes, "-O1" allows up to 5 extra passes,
798 and "-O2"(default), allows up to 10 extra optimization passes.
800 \b Added a new directive: 'cpu XXX', where XXX is any of:
801 8086, 186, 286, 386, 486, 586, pentium, 686, PPro, P2, P3 or
802 Katmai. All are case insensitive. All instructions will
803 be selected only if they apply to the selected cpu or lower.
804 Corrected a couple of bugs in cpu-dependence in 'insns.dat'.
806 \b Added to 'standard.mac', the "use16" and "use32" forms of
807 the "bits 16/32" directive. This is nothing new, just conforms
808 to a lot of other assemblers. (minor)
810 \b Changed label allocation from 320/32 (10000 labels @ 200K+)
811 to 32/37 (1000 labels); makes running under DOS much easier.
812 Since additional label space is allocated dynamically, this
813 should have no effect on large programs with lots of labels.
814 The 37 is a prime, believed to be better for hashing. (minor)
817 \S{cl-0.98.03} Version 0.98.03
819 "Integrated patchfile 0.98-0.98.01. I call this version 0.98.03 for
820 historical reasons: 0.98.02 was trashed." --John Coffman
821 <johninsd@san.rr.com>, 27-Jul-2000
823 \b Kendall Bennett's SciTech MGL changes
825 \b Note that you must define "TASM_COMPAT" at compile-time
826 to get the Tasm Ideal Mode compatibility.
828 \b All changes can be compiled in and out using the TASM_COMPAT macros,
829 and when compiled without TASM_COMPAT defined we get the exact same
830 binary as the unmodified 0.98 sources.
832 \b standard.mac, macros.c: Added macros to ignore TASM directives before
835 \b nasm.h: Added extern declaration for tasm_compatible_mode
837 \b nasm.c: Added global variable tasm_compatible_mode
839 \b Added command line switch for TASM compatible mode (-t)
841 \b Changed version command line to reflect when compiled with TASM additions
843 \b Added response file processing to allow all arguments on a single
844 line (response file is @resp rather than -@resp for NASM format).
846 \b labels.c: Changes islocal() macro to support TASM style @@local labels.
848 \b Added islocalchar() macro to support TASM style @@local labels.
850 \b parser.c: Added support for TASM style memory references (ie: mov
851 [DWORD eax],10 rather than the NASM style mov DWORD [eax],10).
853 \b preproc.c: Added new directives, \c{%arg}, \c{%local}, \c{%stacksize} to directives
856 \b Added support for TASM style directives without a leading % symbol.
858 \b Integrated a block of changes from Andrew Zabolotny <bit@eltech.ru>:
860 \b A new keyword \c{%xdefine} and its case-insensitive counterpart \c{%ixdefine}.
861 They work almost the same way as \c{%define} and \c{%idefine} but expand
862 the definition immediately, not on the invocation. Something like a cross
863 between \c{%define} and \c{%assign}. The "x" suffix stands for "eXpand", so
864 "xdefine" can be deciphered as "expand-and-define". Thus you can do
870 \c %xdefine %1 dword [esp+ofs]
874 \b Changed the place where the expansion of %$name macros are expanded.
875 Now they are converted into ..@ctxnum.name form when detokenizing, so
876 there are no quirks as before when using %$name arguments to macros,
877 in macros etc. For example:
886 Now last line will be expanded into "hello" as expected. This also allows
887 for lots of goodies, a good example are extended "proc" macros included
890 \b Added a check for "cstk" in smacro_defined() before calling get_ctx() -
891 this allows for things like:
896 to work without warnings even in no context.
898 \b Added a check for "cstk" in %if*ctx and %elif*ctx directives -
899 this allows to use \c{%ifctx} without excessive warnings. If there is
900 no active context, \c{%ifctx} goes through "false" branch.
902 \b Removed "user error: " prefix with \c{%error} directive: it just clobbers the
903 output and has absolutely no functionality. Besides, this allows to write
904 macros that does not differ from built-in functions in any way.
906 \b Added expansion of string that is output by \c{%error} directive. Now you
909 \c %define hello(x) Hello, x!
911 \c %define %$name andy
912 \c %error "hello(%$name)"
914 Same happened with \c{%include} directive.
916 \b Now all directives that expect an identifier will try to expand and
917 concatenate everything without whitespaces in between before usage.
918 For example, with "unfixed" nasm the commands
920 \c %define %$abc hello
921 \c %define __%$abc goodbye
924 would produce "incorrect" output: last line will expand to
926 \c hello goodbyehello
928 Not quite what you expected, eh? :-) The answer is that preprocessor
929 treats the \c{%define} construct as if it would be
931 \c %define __ %$abc goodbye
933 (note the white space between __ and %$abc). After my "fix" it
934 will "correctly" expand into
938 as expected. Note that I use quotes around words "correct", "incorrect"
939 etc because this is rather a feature not a bug; however current behaviour
940 is more logical (and allows more advanced macro usage :-).
942 Same change was applied to:
943 \c{%push},\c{%macro},\c{%imacro},\c{%define},\c{%idefine},\c{%xdefine},\c{%ixdefine},
944 \c{%assign},\c{%iassign},\c{%undef}
946 \b A new directive [WARNING {+|-}warning-id] have been added. It works only
947 if the assembly phase is enabled (i.e. it doesn't work with nasm -e).
949 \b A new warning type: macro-selfref. By default this warning is disabled;
950 when enabled NASM warns when a macro self-references itself; for example
951 the following source:
953 \c [WARNING macro-selfref]
964 will produce a warning, but if we remove the first line we won't see it
965 anymore (which is The Right Thing To Do {tm} IMHO since C preprocessor
966 eats such constructs without warnings at all).
968 \b Added a "error" routine to preprocessor which always will set ERR_PASS1
969 bit in severity_code. This removes annoying repeated errors on first
970 and second passes from preprocessor.
972 \b Added the %+ operator in single-line macros for concatenating two
973 identifiers. Usage example:
975 \c %define _myfunc _otherfunc
976 \c %define cextern(x) _ %+ x
979 After first expansion, third line will become "_myfunc". After this
980 expansion is performed again so it becomes "_otherunc".
982 \b Now if preprocessor is in a non-emitting state, no warning or error
983 will be emitted. Example:
988 \c put anything you want between these two brackets,
989 \c even macro-parameter references %1 or local
990 \c labels %$zz or macro-local labels %%zz - no
991 \c warning will be emitted.
994 \b Context-local variables on expansion as a last resort are looked up
995 in outer contexts. For example, the following piece:
1005 will expand correctly the fourth line to [esp]; if we'll define another
1006 %$a inside the "inner" context, it will take precedence over outer
1007 definition. However, this modification has been applied only to
1008 expand_smacro and not to smacro_define: as a consequence expansion
1009 looks in outer contexts, but \c{%ifdef} won't look in outer contexts.
1011 This behaviour is needed because we don't want nested contexts to
1012 act on already defined local macros. Example:
1014 \c %define %$arg1 [esp+4]
1020 In this example the "if" mmacro enters into the "if" context, so %$arg1
1021 is not valid anymore inside "if". Of course it could be worked around
1022 by using explicitely %$$arg1 but this is ugly IMHO.
1024 \b Fixed memory leak in \c{%undef}. The origline wasn't freed before
1027 \b Fixed trap in preprocessor when line expanded to empty set of tokens.
1028 This happens, for example, in the following case:
1030 \c #define SOMETHING
1034 \S{cl-0.98} Version 0.98
1036 All changes since NASM 0.98p3 have been produced by H. Peter Anvin <hpa@zytor.com>.
1038 \b The documentation comment delimiter is \# not #.
1040 \b Allow EQU definitions to refer to external labels; reported by
1043 \b Re-enable support for RDOFF v1; reported by Pedro Gimeno.
1045 \b Updated License file per OK from Simon and Julian.
1048 \S{cl-0.98p9} Version 0.98p9
1050 \b Update documentation (although the instruction set reference will
1051 have to wait; I don't want to hold up the 0.98 release for it.)
1053 \b Verified that the NASM implementation of the PEXTRW and PMOVMSKB
1054 instructions is correct. The encoding differs from what the Intel
1055 manuals document, but the Pentium III behaviour matches NASM, not
1058 \b Fix handling of implicit sizes in PSHUFW and PINSRW, reported by
1061 \b Resurrect the -s option, which was removed when changing the
1062 diagnostic output to stdout.
1065 \S{cl-0.98p8} Version 0.98p8
1067 \b Fix for "DB" when NASM is running on a bigendian machine.
1069 \b Invoke insns.pl once for each output script, making Makefile.in
1070 legal for "make -j".
1072 \b Improve the Unix configure-based makefiles to make package
1075 \b Included an RPM .spec file for building RPM (RedHat Package Manager)
1076 packages on Linux or Unix systems.
1078 \b Fix Makefile dependency problems.
1080 \b Change src/rdsrc.pl to include sectioning information in info
1081 output; required for install-info to work.
1083 \b Updated the RDOFF distribution to version 2 from Jules; minor
1084 massaging to make it compile in my environment.
1086 \b Split doc files that can be built by anyone with a Perl interpreter off
1087 into a separate archive.
1089 \b "Dress rehearsal" release!
1092 \S{cl-0.98p7} Version 0.98p7
1094 \b Fixed opcodes with a third byte-sized immediate argument to not
1095 complain if given "byte" on the immediate.
1097 \b Allow \c{%undef} to remove single-line macros with arguments. This
1098 matches the behaviour of #undef in the C preprocessor.
1100 \b Allow -d, -u, -i and -p to be specified as -D, -U, -I and -P for
1101 compatibility with most C compilers and preprocessors. This allows
1102 Makefile options to be shared between cc and nasm, for example.
1106 \b Went through the list of Katmai instructions and hopefully fixed the
1107 (rather few) mistakes in it.
1109 \b (Hopefully) fixed a number of disassembler bugs related to ambiguous
1110 instructions (disambiguated by -p) and SSE instructions with REP.
1112 \b Fix for bug reported by Mark Junger: "call dword 0x12345678" should
1113 work and may add an OSP (affected CALL, JMP, Jcc).
1115 \b Fix for environments when "stderr" isn't a compile-time constant.
1118 \S{cl-0.98p6} Version 0.98p6
1121 \b Took officially over coordination of the 0.98 release; so drop
1122 the p3.x notation. Skipped p4 and p5 to avoid confusion with John
1123 Fine's J4 and J5 releases.
1125 \b Update the documentation; however, it still doesn't include
1126 documentation for the various new instructions. I somehow wonder if
1127 it makes sense to have an instruction set reference in the assembler
1128 manual when Intel et al have PDF versions of their manuals online.
1130 \b Recognize "idt" or "centaur" for the -p option to ndisasm.
1132 \b Changed error messages back to stderr where they belong, but add an
1133 -E option to redirect them elsewhere (the DOS shell cannot redirect
1136 \b -M option to generate Makefile dependencies (based on code from Alex
1139 \b \c{%undef} preprocessor directive, and -u option, that undefines a
1142 \b OS/2 Makefile (Mkfiles/Makefile.os2) for Borland under OS/2; from
1145 \b Various minor bugfixes (reported by):
1146 - Dangling \c{%s} in preproc.c (Martin Junker)
1148 \b THERE ARE KNOWN BUGS IN SSE AND THE OTHER KATMAI INSTRUCTIONS. I am
1149 on a trip and didn't bring the Katmai instruction reference, so I
1150 can't work on them right now.
1152 \b Updated the License file per agreement with Simon and Jules to
1153 include a GPL distribution clause.
1156 \S{cl-0.98p3.7} Version 0.98p3.7
1158 \b (Hopefully) fixed the canned Makefiles to include the outrdf2 and
1161 \b Renamed changes.asm to changed.asm.
1164 \S{cl-0.98p3.6} Version 0.98p3.6
1166 \b Fixed a bunch of instructions that were added in 0.98p3.5 which had
1167 memory operands, and the address-size prefix was missing from the
1168 instruction pattern.
1171 \S{cl-0.98p3.5} Version 0.98p3.5
1173 \b Merged in changes from John S. Fine's 0.98-J5 release. John's based
1174 0.98-J5 on my 0.98p3.3 release; this merges the changes.
1176 \b Expanded the instructions flag field to a long so we can fit more
1177 flags; mark SSE (KNI) and AMD or Katmai-specific instructions as
1180 \b Fix the "PRIV" flag on a bunch of instructions, and create new
1181 "PROT" flag for protected-mode-only instructions (orthogonal to if
1182 the instruction is privileged!) and new "SMM" flag for SMM-only
1185 \b Added AMD-only SYSCALL and SYSRET instructions.
1187 \b Make SSE actually work, and add new Katmai MMX instructions.
1189 \b Added a -p (preferred vendor) option to ndisasm so that it can
1190 distinguish e.g. Cyrix opcodes also used in SSE. For example:
1192 \c ndisasm -p cyrix aliased.bin
1193 \c 00000000 670F514310 paddsiw mm0,[ebx+0x10]
1194 \c 00000005 670F514320 paddsiw mm0,[ebx+0x20]
1195 \c ndisasm -p intel aliased.bin
1196 \c 00000000 670F514310 sqrtps xmm0,[ebx+0x10]
1197 \c 00000005 670F514320 sqrtps xmm0,[ebx+0x20]
1199 \b Added a bunch of Cyrix-specific instructions.
1202 \S{cl-0.98p3.4} Version 0.98p3.4
1204 \b Made at least an attempt to modify all the additional Makefiles (in
1205 the Mkfiles directory). I can't test it, but this was the best I
1208 \b DOS DJGPP+"Opus Make" Makefile from John S. Fine.
1210 \b changes.asm changes from John S. Fine.
1213 \S{cl-0.98p3.3} Version 0.98p3.3
1215 \b Patch from Conan Brink to allow nesting of \c{%rep} directives.
1217 \b If we're going to allow INT01 as an alias for INT1/ICEBP (one of
1218 Jules 0.98p3 changes), then we should allow INT03 as an alias for INT3
1221 \b Updated changes.asm to include the latest changes.
1223 \b Tried to clean up the <CR>s that had snuck in from a DOS/Windows
1224 environment into my Unix environment, and try to make sure than
1225 DOS/Windows users get them back.
1227 \b We would silently generate broken tools if insns.dat wasn't sorted
1228 properly. Change insns.pl so that the order doesn't matter.
1230 \b Fix bug in insns.pl (introduced by me) which would cause conditional
1231 instructions to have an extra "cc" in disassembly, e.g. "jnz"
1232 disassembled as "jccnz".
1235 \S{cl-0.98p3.2} Version 0.98p3.2
1237 \b Merged in John S. Fine's changes from his 0.98-J4 prerelease; see
1238 http://www.csoft.net/cz/johnfine/
1240 \b Changed previous "spotless" Makefile target (appropriate for distribution)
1241 to "distclean", and added "cleaner" target which is same as "clean"
1242 except deletes files generated by Perl scripts; "spotless" is union.
1244 \b Removed BASIC programs from distribution. Get a Perl interpreter
1245 instead (see below.)
1247 \b Calling this "pre-release 3.2" rather than "p3-hpa2" because of
1248 John's contributions.
1250 \b Actually link in the IEEE output format (zoutieee.c); fix a bunch of
1251 compiler warnings in that file. Note I don't know what IEEE output
1252 is supposed to look like, so these changes were made "blind".
1255 \S{cl-0.98p3-hpa} Version 0.98p3-hpa
1257 \b Merged nasm098p3.zip with nasm-0.97.tar.gz to create a fully
1258 buildable version for Unix systems (Makefile.in updates, etc.)
1260 \b Changed insns.pl to create the instruction tables in nasm.h and
1261 names.c, so that a new instruction can be added by adding it *only*
1264 \b Added the following new instructions: SYSENTER, SYSEXIT, FXSAVE,
1265 FXRSTOR, UD1, UD2 (the latter two are two opcodes that Intel
1266 guarantee will never be used; one of them is documented as UD2 in
1267 Intel documentation, the other one just as "Undefined Opcode" --
1268 calling it UD1 seemed to make sense.)
1270 \b MAX_SYMBOL was defined to be 9, but LOADALL286 and LOADALL386 are 10
1271 characters long. Now MAX_SYMBOL is derived from insns.dat.
1273 \b A note on the BASIC programs included: forget them. insns.bas is
1274 already out of date. Get yourself a Perl interpreter for your
1275 platform of choice at
1276 \W{http://www.cpan.org/ports/index.html}{http://www.cpan.org/ports/index.html}.
1279 \S{cl-0.98p3} Version 0.98 pre-release 3
1281 \b added response file support, improved command line handling, new layout
1284 \b fixed limit checking bug, 'OUT byte nn, reg' bug, and a couple of rdoff
1285 related bugs, updated Wishlist; 0.98 Prerelease 3.
1288 \S{cl-0.98p2} Version 0.98 pre-release 2
1290 \b fixed bug in outcoff.c to do with truncating section names longer
1291 than 8 characters, referencing beyond end of string; 0.98 pre-release 2
1294 \S{cl-0.98p1} Version 0.98 pre-release 1
1296 \b Fixed a bug whereby STRUC didn't work at all in RDF.
1298 \b Fixed a problem with group specification in PUBDEFs in OBJ.
1300 \b Improved ease of adding new output formats. Contribution due to
1303 \b Fixed a bug in relocations in the `bin' format: was showing up when
1304 a relocatable reference crossed an 8192-byte boundary in any output
1307 \b Fixed a bug in local labels: local-label lookups were inconsistent
1308 between passes one and two if an EQU occurred between the definition
1309 of a global label and the subsequent use of a local label local to
1312 \b Fixed a seg-fault in the preprocessor (again) which happened when
1313 you use a blank line as the first line of a multi-line macro
1314 definition and then defined a label on the same line as a call to
1317 \b Fixed a stale-pointer bug in the handling of the NASM environment
1318 variable. Thanks to Thomas McWilliams.
1320 \b ELF had a hard limit on the number of sections which caused
1321 segfaults when transgressed. Fixed.
1323 \b Added ability for ndisasm to read from stdin by using `-' as the
1326 \b ndisasm wasn't outputting the TO keyword. Fixed.
1328 \b Fixed error cascade on bogus expression in \c{%if} - an error in
1329 evaluation was causing the entire \c{%if} to be discarded, thus creating
1330 trouble later when the \c{%else} or \c{%endif} was encountered.
1332 \b Forward reference tracking was instruction-granular not operand-
1333 granular, which was causing 286-specific code to be generated
1334 needlessly on code of the form `shr word [forwardref],1'. Thanks to
1335 Jim Hague for sending a patch.
1337 \b All messages now appear on stdout, as sending them to stderr serves
1338 no useful purpose other than to make redirection difficult.
1340 \b Fixed the problem with EQUs pointing to an external symbol - this
1341 now generates an error message.
1343 \b Allowed multiple size prefixes to an operand, of which only the first
1344 is taken into account.
1346 \b Incorporated John Fine's changes, including fixes of a large number
1347 of preprocessor bugs, some small problems in OBJ, and a reworking of
1348 label handling to define labels before their line is assembled, rather
1351 \b Reformatted a lot of the source code to be more readable. Included
1352 'coding.txt' as a guideline for how to format code for contributors.
1354 \b Stopped nested \c{%reps} causing a panic - they now cause a slightly more
1355 friendly error message instead.
1357 \b Fixed floating point constant problems (patch by Pedro Gimeno)
1359 \b Fixed the return value of insn_size() not being checked for -1, indicating
1362 \b Incorporated 3Dnow! instructions.
1364 \b Fixed the 'mov eax, eax + ebx' bug.
1366 \b Fixed the GLOBAL EQU bug in ELF. Released developers release 3.
1368 \b Incorporated John Fine's command line parsing changes
1370 \b Incorporated David Lindauer's OMF debug support
1372 \b Made changes for LCC 4.0 support (\c{__NASM_CDecl__}, removed register size
1373 specification warning when sizes agree).
1376 \H{cl-0.9x} NASM 0.9 Series
1378 Revisions before 0.98.
1381 \S{cl-0.97} Version 0.97 released December 1997
1383 \b This was entirely a bug-fix release to 0.96, which seems to have got
1386 \b Fixed stupid mistake in OBJ which caused `MOV EAX,<constant>' to
1387 fail. Caused by an error in the `MOV EAX,<segment>' support.
1389 \b ndisasm hung at EOF when compiled with lcc on Linux because lcc on
1390 Linux somehow breaks feof(). ndisasm now does not rely on feof().
1392 \b A heading in the documentation was missing due to a markup error in
1393 the indexing. Fixed.
1395 \b Fixed failure to update all pointers on realloc() within extended-
1396 operand code in parser.c. Was causing wrong behaviour and seg faults
1397 on lines such as `dd 0.0,0.0,0.0,0.0,...'
1399 \b Fixed a subtle preprocessor bug whereby invoking one multi-line
1400 macro on the first line of the expansion of another, when the second
1401 had been invoked with a label defined before it, didn't expand the
1404 \b Added internal.doc back in to the distribution archives - it was
1405 missing in 0.96 *blush*
1407 \b Fixed bug causing 0.96 to be unable to assemble its own test files,
1408 specifically objtest.asm. *blush again*
1410 \b Fixed seg-faults and bogus error messages caused by mismatching
1411 \c{%rep} and \c{%endrep} within multi-line macro definitions.
1413 \b Fixed a problem with buffer overrun in OBJ, which was causing
1414 corruption at ends of long PUBDEF records.
1416 \b Separated DOS archives into main-program and documentation to reduce
1420 \S{cl-0.96} Version 0.96 released November 1997
1422 \b Fixed a bug whereby, if `nasm sourcefile' would cause a filename
1423 collision warning and put output into `nasm.out', then `nasm
1424 sourcefile -o outputfile' still gave the warning even though the
1426 Fixed name pollution under Digital UNIX: one of its header files
1427 defined R_SP, which broke the enum in nasm.h.
1429 \b Fixed minor instruction table problems: FUCOM and FUCOMP didn't have
1430 two-operand forms; NDISASM didn't recognise the longer register
1431 forms of PUSH and POP (eg FF F3 for PUSH BX); TEST mem,imm32 was
1432 flagged as undocumented; the 32-bit forms of CMOV had 16-bit operand
1433 size prefixes; `AAD imm' and `AAM imm' are no longer flagged as
1434 undocumented because the Intel Architecture reference documents
1437 \b Fixed a problem with the local-label mechanism, whereby strange
1438 types of symbol (EQUs, auto-defined OBJ segment base symbols)
1439 interfered with the `previous global label' value and screwed up
1442 \b Fixed a bug whereby the stub preprocessor didn't communicate with
1443 the listing file generator, so that the -a and -l options in
1444 conjunction would produce a useless listing file.
1446 \b Merged `os2' object file format back into `obj', after discovering
1447 that `obj' _also_ shouldn't have a link pass separator in a module
1448 containing a non-trivial MODEND. Flat segments are now declared
1449 using the FLAT attribute. `os2' is no longer a valid object format
1452 \b Removed the fixed-size temporary storage in the evaluator. Very very
1453 long expressions (like `mov ax,1+1+1+1+...' for two hundred 1s or
1454 so) should now no longer crash NASM.
1456 \b Fixed a bug involving segfaults on disassembly of MMX instructions,
1457 by changing the meaning of one of the operand-type flags in nasm.h.
1458 This may cause other apparently unrelated MMX problems; it needs to
1459 be tested thoroughly.
1461 \b Fixed some buffer overrun problems with large OBJ output files.
1462 Thanks to DJ Delorie for the bug report and fix.
1464 \b Made preprocess-only mode actually listen to the \c{%line} markers as it
1465 prints them, so that it can report errors more sanely.
1467 \b Re-designed the evaluator to keep more sensible track of expressions
1468 involving forward references: can now cope with previously-nightmare
1475 \b Added the ALIGN and ALIGNB standard macros.
1477 \b Added PIC support in ELF: use of WRT to obtain the four extra
1478 relocation types needed.
1480 \b Added the ability for output file formats to define their own
1481 extensions to the GLOBAL, COMMON and EXTERN directives.
1483 \b Implemented common-variable alignment, and global-symbol type and
1484 size declarations, in ELF.
1486 \b Implemented NEAR and FAR keywords for common variables, plus
1487 far-common element size specification, in OBJ.
1489 \b Added a feature whereby EXTERNs and COMMONs in OBJ can be given a
1490 default WRT specification (either a segment or a group).
1492 \b Transformed the Unix NASM archive into an auto-configuring package.
1494 \b Added a sanity-check for people applying SEG to things which are
1495 already segment bases: this previously went unnoticed by the SEG
1496 processing and caused OBJ-driver panics later.
1498 \b Added the ability, in OBJ format, to deal with `MOV EAX,<segment>'
1499 type references: OBJ doesn't directly support dword-size segment
1500 base fixups, but as long as the low two bytes of the constant term
1501 are zero, a word-size fixup can be generated instead and it will
1504 \b Added the ability to specify sections' alignment requirements in
1505 Win32 object files and pure binary files.
1507 \b Added preprocess-time expression evaluation: the \c{%assign} (and
1508 \c{%iassign}) directive and the bare \c{%if} (and \c{%elif}) conditional. Added
1509 relational operators to the evaluator, for use only in \c{%if}
1510 constructs: the standard relationals = < > <= >= <> (and C-like
1511 synonyms == and !=) plus low-precedence logical operators &&, ^^ and
1514 \b Added a preprocessor repeat construct: \c{%rep} / \c{%exitrep} / \c{%endrep}.
1516 \b Added the __FILE__ and __LINE__ standard macros.
1518 \b Added a sanity check for number constants being greater than
1519 0xFFFFFFFF. The warning can be disabled.
1521 \b Added the %0 token whereby a variadic multi-line macro can tell how
1522 many parameters it's been given in a specific invocation.
1524 \b Added \c{%rotate}, allowing multi-line macro parameters to be cycled.
1526 \b Added the `*' option for the maximum parameter count on multi-line
1527 macros, allowing them to take arbitrarily many parameters.
1529 \b Added the ability for the user-level forms of EXTERN, GLOBAL and
1530 COMMON to take more than one argument.
1532 \b Added the IMPORT and EXPORT directives in OBJ format, to deal with
1535 \b Added some more preprocessor \c{%if} constructs: \c{%ifidn} / \c{%ifidni} (exact
1536 textual identity), and \c{%ifid} / \c{%ifnum} / \c{%ifstr} (token type testing).
1538 \b Added the ability to distinguish SHL AX,1 (the 8086 version) from
1539 SHL AX,BYTE 1 (the 286-and-upwards version whose constant happens to
1542 \b Added NetBSD/FreeBSD/OpenBSD's variant of a.out format, complete
1543 with PIC shared library features.
1545 \b Changed NASM's idiosyncratic handling of FCLEX, FDISI, FENI, FINIT,
1546 FSAVE, FSTCW, FSTENV, and FSTSW to bring it into line with the
1547 otherwise accepted standard. The previous behaviour, though it was a
1548 deliberate feature, was a deliberate feature based on a
1549 misunderstanding. Apologies for the inconvenience.
1551 \b Improved the flexibility of ABSOLUTE: you can now give it an
1552 expression rather than being restricted to a constant, and it can
1553 take relocatable arguments as well.
1555 \b Added the ability for a variable to be declared as EXTERN multiple
1556 times, and the subsequent definitions are just ignored.
1558 \b We now allow instruction prefixes (CS, DS, LOCK, REPZ etc) to be
1559 alone on a line (without a following instruction).
1561 \b Improved sanity checks on whether the arguments to EXTERN, GLOBAL
1562 and COMMON are valid identifiers.
1564 \b Added misc/exebin.mac to allow direct generation of .EXE files by
1565 hacking up an EXE header using DB and DW; also added test/binexe.asm
1566 to demonstrate the use of this. Thanks to Yann Guidon for
1567 contributing the EXE header code.
1569 \b ndisasm forgot to check whether the input file had been successfully
1570 opened. Now it does. Doh!
1572 \b Added the Cyrix extensions to the MMX instruction set.
1574 \b Added a hinting mechanism to allow [EAX+EBX] and [EBX+EAX] to be
1575 assembled differently. This is important since [ESI+EBP] and
1576 [EBP+ESI] have different default base segment registers.
1578 \b Added support for the PharLap OMF extension for 4096-byte segment
1582 \S{cl-0.95 released July 1997} Version 0.95 released July 1997
1584 \b Fixed yet another ELF bug. This one manifested if the user relied on
1585 the default segment, and attempted to define global symbols without
1586 first explicitly declaring the target segment.
1588 \b Added makefiles (for NASM and the RDF tools) to build Win32 console
1589 apps under Symantec C++. Donated by Mark Junker.
1591 \b Added `macros.bas' and `insns.bas', QBasic versions of the Perl
1592 scripts that convert `standard.mac' to `macros.c' and convert
1593 `insns.dat' to `insnsa.c' and `insnsd.c'. Also thanks to Mark
1596 \b Changed the diassembled forms of the conditional instructions so
1597 that JB is now emitted as JC, and other similar changes. Suggested
1598 list by Ulrich Doewich.
1600 \b Added `@' to the list of valid characters to begin an identifier
1603 \b Documentary changes, notably the addition of the `Common Problems'
1604 section in nasm.doc.
1606 \b Fixed a bug relating to 32-bit PC-relative fixups in OBJ.
1608 \b Fixed a bug in perm_copy() in labels.c which was causing exceptions
1609 in cleanup_labels() on some systems.
1611 \b Positivity sanity check in TIMES argument changed from a warning to
1612 an error following a further complaint.
1614 \b Changed the acceptable limits on byte and word operands to allow
1615 things like `~10111001b' to work.
1617 \b Fixed a major problem in the preprocessor which caused seg-faults if
1618 macro definitions contained blank lines or comment-only lines.
1620 \b Fixed inadequate error checking on the commas separating the
1621 arguments to `db', `dw' etc.
1623 \b Fixed a crippling bug in the handling of macros with operand counts
1624 defined with a `+' modifier.
1626 \b Fixed a bug whereby object file formats which stored the input file
1627 name in the output file (such as OBJ and COFF) weren't doing so
1628 correctly when the output file name was specified on the command
1631 \b Removed [INC] and [INCLUDE] support for good, since they were
1634 \b Fixed a bug in OBJ which caused all fixups to be output in 16-bit
1635 (old-format) FIXUPP records, rather than putting the 32-bit ones in
1636 FIXUPP32 (new-format) records.
1638 \b Added, tentatively, OS/2 object file support (as a minor variant on
1641 \b Updates to Fox Cutter's Borland C makefile, Makefile.bc2.
1643 \b Removed a spurious second fclose() on the output file.
1645 \b Added the `-s' command line option to redirect all messages which
1646 would go to stderr (errors, help text) to stdout instead.
1648 \b Added the `-w' command line option to selectively suppress some
1649 classes of assembly warning messages.
1651 \b Added the `-p' pre-include and `-d' pre-define command-line options.
1653 \b Added an include file search path: the `-i' command line option.
1655 \b Fixed a silly little preprocessor bug whereby starting a line with a
1656 `%!' environment-variable reference caused an `unknown directive'
1659 \b Added the long-awaited listing file support: the `-l' command line
1662 \b Fixed a problem with OBJ format whereby, in the absence of any
1663 explicit segment definition, non-global symbols declared in the
1664 implicit default segment generated spurious EXTDEF records in the
1667 \b Added the NASM environment variable.
1669 \b From this version forward, Win32 console-mode binaries will be
1670 included in the DOS distribution in addition to the 16-bit binaries.
1671 Added Makefile.vc for this purpose.
1673 \b Added `return 0;' to test/objlink.c to prevent compiler warnings.
1675 \b Added the __NASM_MAJOR__ and __NASM_MINOR__ standard defines.
1677 \b Added an alternative memory-reference syntax in which prefixing an
1678 operand with `&' is equivalent to enclosing it in square brackets,
1679 at the request of Fox Cutter.
1681 \b Errors in pass two now cause the program to return a non-zero error
1682 code, which they didn't before.
1684 \b Fixed the single-line macro cycle detection, which didn't work at
1685 all on macros with no parameters (caused an infinite loop). Also
1686 changed the behaviour of single-line macro cycle detection to work
1687 like cpp, so that macros like `extrn' as given in the documentation
1690 \b Fixed the implementation of WRT, which was too restrictive in that
1691 you couldn't do `mov ax,[di+abc wrt dgroup]' because (di+abc) wasn't
1692 a relocatable reference.
1695 \S{cl-0.94 released April 1997} Version 0.94 released April 1997
1698 \b Major item: added the macro processor.
1700 \b Added undocumented instructions SMI, IBTS, XBTS and LOADALL286. Also
1701 reorganised CMPXCHG instruction into early-486 and Pentium forms.
1702 Thanks to Thobias Jones for the information.
1704 \b Fixed two more stupid bugs in ELF, which were causing `ld' to
1705 continue to seg-fault in a lot of non-trivial cases.
1707 \b Fixed a seg-fault in the label manager.
1709 \b Stopped FBLD and FBSTP from _requiring_ the TWORD keyword, which is
1710 the only option for BCD loads/stores in any case.
1712 \b Ensured FLDCW, FSTCW and FSTSW can cope with the WORD keyword, if
1713 anyone bothers to provide it. Previously they complained unless no
1714 keyword at all was present.
1716 \b Some forms of FDIV/FDIVR and FSUB/FSUBR were still inverted: a
1717 vestige of a bug that I thought had been fixed in 0.92. This was
1718 fixed, hopefully for good this time...
1720 \b Another minor phase error (insofar as a phase error can _ever_ be
1721 minor) fixed, this one occurring in code of the form
1723 \c rol ax,forward_reference
1724 \c forward_reference equ 1
1726 \b The number supplied to TIMES is now sanity-checked for positivity,
1727 and also may be greater than 64K (which previously didn't work on
1730 \b Added Watcom C makefiles, and misc/pmw.bat, donated by Dominik Behr.
1732 \b Added the INCBIN pseudo-opcode.
1734 \b Due to the advent of the preprocessor, the [INCLUDE] and [INC]
1735 directives have become obsolete. They are still supported in this
1736 version, with a warning, but won't be in the next.
1738 \b Fixed a bug in OBJ format, which caused incorrect object records to
1739 be output when absolute labels were made global.
1741 \b Updates to RDOFF subdirectory, and changes to outrdf.c.
1744 \S{cl-0.93 released January 1997} Version 0.93 released January 1997
1746 This release went out in a great hurry after semi-crippling bugs
1749 \b Really \e{did} fix the stack overflows this time. *blush*
1751 \b Had problems with EA instruction sizes changing between passes, when
1752 an offset contained a forward reference and so 4 bytes were
1753 allocated for the offset in pass one; by pass two the symbol had
1754 been defined and happened to be a small absolute value, so only 1
1755 byte got allocated, causing instruction size mismatch between passes
1756 and hence incorrect address calculations. Fixed.
1758 \b Stupid bug in the revised ELF section generation fixed (associated
1759 string-table section for .symtab was hard-coded as 7, even when this
1760 didn't fit with the real section table). Was causing `ld' to
1761 seg-fault under Linux.
1763 \b Included a new Borland C makefile, Makefile.bc2, donated by Fox
1764 Cutter <lmb@comtch.iea.com>.
1767 \S{cl-0.92 released January 1997} Version 0.92 released January 1997
1769 \b The FDIVP/FDIVRP and FSUBP/FSUBRP pairs had been inverted: this was
1770 fixed. This also affected the LCC driver.
1772 \b Fixed a bug regarding 32-bit effective addresses of the form
1773 \c{[other_register+ESP]}.
1775 \b Documentary changes, notably documentation of the fact that Borland
1776 Win32 compilers use `obj' rather than `win32' object format.
1778 \b Fixed the COMENT record in OBJ files, which was formatted
1781 \b Fixed a bug causing segfaults in large RDF files.
1783 \b OBJ format now strips initial periods from segment and group
1784 definitions, in order to avoid complications with the local label
1787 \b Fixed a bug in disassembling far calls and jumps in NDISASM.
1789 \b Added support for user-defined sections in COFF and ELF files.
1791 \b Compiled the DOS binaries with a sensible amount of stack, to
1792 prevent stack overflows on any arithmetic expression containing
1795 \b Fixed a bug in handling of files that do not terminate in a newline.
1798 \S{cl-0.91 released November 1996} Version 0.91 released November 1996
1800 \b Loads of bug fixes.
1802 \b Support for RDF added.
1804 \b Support for DBG debugging format added.
1806 \b Support for 32-bit extensions to Microsoft OBJ format added.
1808 \b Revised for Borland C: some variable names changed, makefile added.
1810 \b LCC support revised to actually work.
1812 \b JMP/CALL NEAR/FAR notation added.
1814 \b `a16', `o16', `a32' and `o32' prefixes added.
1816 \b Range checking on short jumps implemented.
1818 \b MMX instruction support added.
1820 \b Negative floating point constant support added.
1822 \b Memory handling improved to bypass 64K barrier under DOS.
1824 \b \c{$} prefix to force treatment of reserved words as identifiers added.
1826 \b Default-size mechanism for object formats added.
1828 \b Compile-time configurability added.
1830 \b \c{#}, \c{@}, \c{~} and c\{?} are now valid characters in labels.
1832 \b \c{-e} and \c{-k} options in NDISASM added.
1835 \S{cl-0.90 released October 1996} Version 0.90 released October 1996
1837 First release version. First support for object file output. Other
1838 changes from previous version (0.3x) too numerous to document.