8 * Changed doc files a lot: completely removed old READMExx and
9 Wishlist files, incorporating all information in CHANGES and TODO.
10 * I waited a long time to rename zoutieee.c to (original) outieee.c
11 * moved all output modules to output/ subdirectory.
12 * Added 'make strip' target to strip debug info from nasm & ndisasm.
13 * Added INSTALL file with installation instructions.
14 * Added -v option description to nasm man.
15 * Added dist makefile target to produce source distributions.
21 * Fastcooked this for Debian's Woody release:
22 Frank applied the INCBIN bug patch to 0.98.25alt and called
23 it 0.98.28 to not confuse poor little apt-get.
29 * Reorganised files even better from 0.98.25alt
35 * Prettified the source tree. Moved files to more
37 * Added findleak.pl script to misc/ directory.
38 * FIXME: Frank, document this please.
44 * FIXME: Frank, document this please.
50 * FIXME: Frank, document this please.
56 * FIXME: Frank, document this please.
62 * FIXME: Frank, document this please.
68 * FIXME: Frank, document this please.
74 * FIXME: Frank, document this please.
80 * FIXME: Frank, document this please.
86 * FIXME: Frank, document this please.
92 * FIXME: Frank, document this please.
98 * FIXME: Frank, document this please.
104 * FIXME: Frank, document this please.
110 * FIXME: Frank, document this please.
116 * FIXME: Frank, document this please.
122 * FIXME: Frank, document this please.
128 * FIXME: Frank, document this please.
131 0.98.09b with John Coffman patches released 28-Oct-2001
132 -------------------------------------------------------
134 Changes from 0.98.07 release to 98.09b as of 28-Oct-2001
136 * More closely compatible with 0.98 when -O0 is implied
137 or specified. Not strictly identical, since backward
138 branches in range of short offsets are recognized, and signed
139 byte values with no explicit size specification will be
140 assembled as a single byte.
142 * More forgiving with the PUSH instruction. 0.98 requires
143 a size to be specified always. 0.98.09b will imply the size
144 from the current BITS setting (16 or 32).
146 * Changed definition of the optimization flag:
148 -O0 strict two-pass assembly, JMP and Jcc are
149 handled more like 0.98, except that back-
150 ward JMPs are short, if possible.
152 -O1 strict two-pass assembly, but forward
153 branches are assembled with code guaranteed
154 to reach; may produce larger code than
155 -O0, but will produce successful assembly
156 more often if branch offset sizes are not
159 -O2 multi-pass optimization, minimize branch
160 offsets; also will minimize signed immed-
161 iate bytes, overriding size specification.
163 -O3 like -O2, but more passes taken, if needed
166 0.98.07 released 01/28/01
167 -------------------------
169 * fbk - added Stepane Denis' SSE2 instructions to a *working*
170 version of the code - some earlier versions were based on
171 broken code - sorry 'bout that. version "0.98.07"
177 * fbk - cosmetic modifications to nasm.c, nasm.h,
181 0.98.06f released 01/18/01
182 --------------------------
184 * fbk - "metalbrain"s jecxz bug fix in insns.dat
185 - alter nasmdoc.src to match - version "0.98.06f"
188 0.98.06e released 01/09/01
189 --------------------------
191 * fbk - removed the "outforms.h" file - it appears to be
192 someone's old backup of "outform.h". version "0.98.06e"
198 * fbk - finally added the fix for the "multiple %includes bug",
199 known since 7/27/99 - reported originally (?) and sent to
200 us by Austin Lunnen - he reports that John Fine had a fix
201 within the day. Here it is...
203 * Nelson Rush resigns from the group. Big thanks to Nelson for
204 his leadership and enthusiasm in getting these changes
205 incorporated into Nasm!
207 * fbk - [list +], [list -] directives - ineptly implemented, should
208 be re-written or removed, perhaps.
210 * Brian Raiter / fbk - "elfso bug" fix - applied to aoutb format
211 as well - testing might be desirable...
217 * James Seter - -postfix, -prefix command line switches.
218 * Yuri Zaporogets - rdoff utility changes.
224 * GAS-like palign (Panos Minos)
225 * FIXME: Frank, fill this in with details
228 0.98bf (bug-fixed, aka brain-fuck)
229 ----------------------------------
231 * FIXME: Frank, fill this in
234 0.98.03 with John Coffman's changes released 27-Jul-2000
235 --------------------------------------------------------
237 * Added signed byte optimizations for the 0x81/0x83 class
238 of instructions: ADC, ADD, AND, CMP, OR, SBB, SUB, XOR:
239 when used as 'ADD reg16,imm' or 'ADD reg32,imm.' Also
240 optimization of signed byte form of 'PUSH imm' and 'IMUL
241 reg,imm'/'IMUL reg,reg,imm.' No size specification is needed.
243 * Added multi-pass JMP and Jcc offset optimization. Offsets
244 on forward references will preferentially use the short form,
245 without the need to code a specific size (short or near) for
246 the branch. Added instructions for 'Jcc label' to use the
247 form 'Jnotcc $+3/JMP label', in cases where a short offset
248 is out of bounds. If compiling for a 386 or higher CPU, then
249 the 386 form of Jcc will be used instead.
251 This feature is controlled by a new command-line switch: "O",
252 (upper case letter O). "-O0" reverts the assembler to no
253 extra optimization passes, "-O1" allows up to 5 extra passes,
254 and "-O2"(default), allows up to 10 extra optimization passes.
256 * Added a new directive: 'cpu XXX', where XXX is any of:
257 8086, 186, 286, 386, 486, 586, pentium, 686, PPro, P2, P3 or
258 Katmai. All are case insensitive. All instructions will
259 be selected only if they apply to the selected cpu or lower.
260 Corrected a couple of bugs in cpu-dependence in 'insns.dat'.
262 * Added to 'standard.mac', the "use16" and "use32" forms of
263 the "bits 16/32" directive. This is nothing new, just conforms
264 to a lot of other assemblers. (minor)
266 * Changed label allocation from 320/32 (10000 labels @ 200K+)
267 to 32/37 (1000 labels); makes running under DOS much easier.
268 Since additional label space is allocated dynamically, this
269 should have no effect on large programs with lots of labels.
270 The 37 is a prime, believed to be better for hashing. (minor)
272 * Integrated patchfile 0.98-0.98.01. I call this version
273 0.98.03, for historical reasons: 0.98.02 was trashed.
275 --John Coffman <johninsd@san.rr.com> 27-Jul-2000
278 Kendall Bennett's SciTech MGL changes
279 -------------------------------------
280 Note that you must define "TASM_COMPAT" at compile-time
281 to get the Tasm Ideal Mode compatibility.
283 All changes can be compiled in and out using the TASM_COMPAT macros,
284 and when compiled without TASM_COMPAT defined we get the exact same
285 binary as the unmodified 0.98 sources.
287 standard.mac, macros.c:
288 . Added macros to ignore TASM directives before first include
291 . Added extern declaration for tasm_compatible_mode
294 . Added global variable tasm_compatible_mode
295 . Added command line switch for TASM compatible mode (-t)
296 . Changed version command line to reflect when compiled with TASM additions
297 . Added response file processing to allow all arguments on a single
298 line (response file is @resp rather than -@resp for NASM format).
301 . Changes islocal() macro to support TASM style @@local labels.
302 . Added islocalchar() macro to support TASM style @@local labels.
305 . Added support for TASM style memory references (ie: mov [DWORD eax],10
306 rather than the NASM style mov DWORD [eax],10).
309 . Added new directives, %arg, %local, %stacksize to directives table
310 . Added support for TASM style directives without a leading % symbol.
312 Integrated a block of changes from Andrew Zabolotny <bit@eltech.ru>:
314 * A new keyword %xdefine and its case-insensitive counterpart %ixdefine.
315 They work almost the same way as %define and %idefine but expand
316 the definition immediately, not on the invocation. Something like a cross
317 between %define and %assign. The "x" suffix stands for "eXpand", so
318 "xdefine" can be deciphered as "expand-and-define". Thus you can do
324 %xdefine %1 dword [esp+ofs]
328 * Changed the place where the expansion of %$name macros are expanded.
329 Now they are converted into ..@ctxnum.name form when detokenizing, so
330 there are no quirks as before when using %$name arguments to macros,
331 in macros etc. For example:
340 Now last line will be expanded into "hello" as expected. This also allows
341 for lots of goodies, a good example are extended "proc" macros included
344 * Added a check for "cstk" in smacro_defined() before calling get_ctx() -
345 this allows for things like:
350 to work without warnings even in no context.
352 * Added a check for "cstk" in %if*ctx and %elif*ctx directives -
353 this allows to use %ifctx without excessive warnings. If there is
354 no active context, %ifctx goes through "false" branch.
356 * Removed "user error: " prefix with %error directive: it just clobbers the
357 output and has absolutely no functionality. Besides, this allows to write
358 macros that does not differ from built-in functions in any way.
360 * Added expansion of string that is output by %error directive. Now you
363 %define hello(x) Hello, x!
366 %error "hello(%$name)"
368 Same happened with %include directive.
370 * Now all directives that expect an identifier will try to expand and
371 concatenate everything without whitespaces in between before usage.
372 For example, with "unfixed" nasm the commands
375 %define __%$abc goodbye
378 would produce "incorrect" output: last line will expand to
382 Not quite what you expected, eh? :-) The answer is that preprocessor
383 treats the %define construct as if it would be
385 %define __ %$abc goodbye
387 (note the white space between __ and %$abc). After my "fix" it
388 will "correctly" expand into
392 as expected. Note that I use quotes around words "correct", "incorrect"
393 etc because this is rather a feature not a bug; however current behaviour
394 is more logical (and allows more advanced macro usage :-).
396 Same change was applied to:
397 %push,%macro,%imacro,%define,%idefine,%xdefine,%ixdefine,
398 %assign,%iassign,%undef
400 * A new directive [WARNING {+|-}warning-id] have been added. It works only
401 if the assembly phase is enabled (i.e. it doesn't work with nasm -e).
403 * A new warning type: macro-selfref. By default this warning is disabled;
404 when enabled NASM warns when a macro self-references itself; for example
405 the following source:
407 [WARNING macro-selfref]
418 will produce a warning, but if we remove the first line we won't see it
419 anymore (which is The Right Thing To Do {tm} IMHO since C preprocessor
420 eats such constructs without warnings at all).
422 * Added a "error" routine to preprocessor which always will set ERR_PASS1
423 bit in severity_code. This removes annoying repeated errors on first
424 and second passes from preprocessor.
426 * Added the %+ operator in single-line macros for concatenating two
427 identifiers. Usage example:
429 %define _myfunc _otherfunc
430 %define cextern(x) _ %+ x
433 After first expansion, third line will become "_myfunc". After this
434 expansion is performed again so it becomes "_otherunc".
436 * Now if preprocessor is in a non-emmitting state, no warning or error
437 will be emmitted. Example:
442 put anything you want between these two brackets,
443 even macro-parameter references %1 or local labels %$zz
444 or macro-local labels %%zz - no warning will be emmitted.
447 * Context-local variables on expansion as a last resort are looked up
448 in outer contexts. For example, the following piece:
458 will expand correctly the fourth line to [esp]; if we'll define another
459 %$a inside the "inner" context, it will take precedence over outer
460 definition. However, this modification has been applied only to
461 expand_smacro and not to smacro_define: as a consequence expansion
462 looks in outer contexts, but %ifdef won't look in outer contexts.
464 This behaviour is needed because we don't want nested contexts to
465 act on already defined local macros. Example:
467 %define %$arg1 [esp+4]
473 In this example the "if" mmacro enters into the "if" context, so %$arg1
474 is not valid anymore inside "if". Of course it could be worked around
475 by using explicitely %$$arg1 but this is ugly IMHO.
477 * Fixed memory leak in %undef. The origline wasn't freed before
480 * Fixed trap in preprocessor when line expanded to empty set of tokens.
481 This happens, for example, in the following case:
490 All changes since NASM 0.98p3 have been produced by H. Peter Anvin <hpa@zytor.com>.
492 * The documentation comment delimiter is \# not #.
493 * Allow EQU definitions to refer to external labels; reported by
495 * Re-enable support for RDOFF v1; reported by Pedro Gimeno.
496 * Updated License file per OK from Simon and Julian.
502 * Update documentation (although the instruction set reference will
503 have to wait; I don't want to hold up the 0.98 release for it.)
504 * Verified that the NASM implementation of the PEXTRW and PMOVMSKB
505 instructions is correct. The encoding differs from what the Intel
506 manuals document, but the Pentium III behaviour matches NASM, not
508 * Fix handling of implicit sizes in PSHUFW and PINSRW, reported by
510 * Resurrect the -s option, which was removed when changing the
511 diagnostic output to stdout.
517 * Fix for "DB" when NASM is running on a bigendian machine.
518 * Invoke insns.pl once for each output script, making Makefile.in
520 * Improve the Unix configure-based makefiles to make package
522 * Included an RPM .spec file for building RPM (RedHat Package Manager)
523 packages on Linux or Unix systems.
524 * Fix Makefile dependency problems.
525 * Change src/rdsrc.pl to include sectioning information in info
526 output; required for install-info to work.
527 * Updated the RDOFF distribution to version 2 from Jules; minor
528 massaging to make it compile in my environment.
529 * Split doc files that can be built by anyone with a Perl interpreter off
530 into a separate archive.
531 * "Dress rehearsal" release!
537 * Fixed opcodes with a third byte-sized immediate argument to not
538 complain if given "byte" on the immediate.
539 * Allow %undef to remove single-line macros with arguments. This
540 matches the behaviour of #undef in the C preprocessor.
541 * Allow -d, -u, -i and -p to be specified as -D, -U, -I and -P for
542 compatibility with most C compilers and preprocessors. This allows
543 Makefile options to be shared between cc and nasm, for example.
545 * Went through the list of Katmai instructions and hopefully fixed the
546 (rather few) mistakes in it.
547 * (Hopefully) fixed a number of disassembler bugs related to ambiguous
548 instructions (disambiguated by -p) and SSE instructions with REP.
549 * Fix for bug reported by Mark Junger: "call dword 0x12345678" should
550 work and may add an OSP (affected CALL, JMP, Jcc).
551 * Fix for environments when "stderr" isn't a compile-time constant.
557 * Took officially over coordination of the 0.98 release; so drop
558 the p3.x notation. Skipped p4 and p5 to avoid confusion with John
559 Fine's J4 and J5 releases.
560 * Update the documentation; however, it still doesn't include
561 documentation for the various new instructions. I somehow wonder if
562 it makes sense to have an instruction set reference in the assembler
563 manual when Intel et al have PDF versions of their manuals online.
564 * Recognize "idt" or "centaur" for the -p option to ndisasm.
565 * Changed error messages back to stderr where they belong, but add an
566 -E option to redirect them elsewhere (the DOS shell cannot redirect
568 * -M option to generate Makefile dependencies (based on code from Alex
570 * %undef preprocessor directive, and -u option, that undefines a
572 * OS/2 Makefile (Mkfiles/Makefile.os2) for Borland under OS/2; from
574 * Various minor bugfixes (reported by):
575 - Dangling %s in preproc.c (Martin Junker)
576 * THERE ARE KNOWN BUGS IN SSE AND THE OTHER KATMAI INSTRUCTIONS. I am
577 on a trip and didn't bring the Katmai instruction reference, so I
578 can't work on them right now.
579 * Updated the License file per agreement with Simon and Jules to
580 include a GPL distribution clause.
586 * (Hopefully) fixed the canned Makefiles to include the outrdf2 and
588 * Renamed changes.asm to changed.asm.
594 * Fixed a bunch of instructions that were added in 0.98p3.5 which had
595 memory operands, and the address-size prefix was missing from the
602 * Merged in changes from John S. Fine's 0.98-J5 release. John's based
603 0.98-J5 on my 0.98p3.3 release; this merges the changes.
604 * Expanded the instructions flag field to a long so we can fit more
605 flags; mark SSE (KNI) and AMD or Katmai-specific instructions as
607 * Fix the "PRIV" flag on a bunch of instructions, and create new
608 "PROT" flag for protected-mode-only instructions (orthogonal to if
609 the instruction is privileged!) and new "SMM" flag for SMM-only
611 * Added AMD-only SYSCALL and SYSRET instructions.
612 * Make SSE actually work, and add new Katmai MMX instructions.
613 * Added a -p (preferred vendor) option to ndisasm so that it can
614 distinguish e.g. Cyrix opcodes also used in SSE. For example:
616 ndisasm -p cyrix aliased.bin
617 00000000 670F514310 paddsiw mm0,[ebx+0x10]
618 00000005 670F514320 paddsiw mm0,[ebx+0x20]
619 ndisasm -p intel aliased.bin
620 00000000 670F514310 sqrtps xmm0,[ebx+0x10]
621 00000005 670F514320 sqrtps xmm0,[ebx+0x20]
622 * Added a bunch of Cyrix-specific instructions.
628 * Made at least an attempt to modify all the additional Makefiles (in
629 the Mkfiles directory). I can't test it, but this was the best I
631 * DOS DJGPP+"Opus Make" Makefile from John S. Fine.
632 * changes.asm changes from John S. Fine.
638 * Patch from Conan Brink to allow nesting of %rep directives.
639 * If we're going to allow INT01 as an alias for INT1/ICEBP (one of
640 Jules 0.98p3 changes), then we should allow INT03 as an alias for INT3
642 * Updated changes.asm to include the latest changes.
643 * Tried to clean up the <CR>s that had snuck in from a DOS/Windows
644 environment into my Unix environment, and try to make sure than
645 DOS/Windows users get them back.
646 * We would silently generate broken tools if insns.dat wasn't sorted
647 properly. Change insns.pl so that the order doesn't matter.
648 * Fix bug in insns.pl (introduced by me) which would cause conditional
649 instructions to have an extra "cc" in disassembly, e.g. "jnz"
650 disassembled as "jccnz".
656 * Merged in John S. Fine's changes from his 0.98-J4 prerelease; see
657 http://www.csoft.net/cz/johnfine/
658 * Changed previous "spotless" Makefile target (appropriate for distribution)
659 to "distclean", and added "cleaner" target which is same as "clean"
660 except deletes files generated by Perl scripts; "spotless" is union.
661 * Removed BASIC programs from distribution. Get a Perl interpreter
663 * Calling this "pre-release 3.2" rather than "p3-hpa2" because of
664 John's contributions.
665 * Actually link in the IEEE output format (zoutieee.c); fix a bunch of
666 compiler warnings in that file. Note I don't know what IEEE output
667 is supposed to look like, so these changes were made "blind".
673 * Merged nasm098p3.zip with nasm-0.97.tar.gz to create a fully
674 buildable version for Unix systems (Makefile.in updates, etc.)
675 * Changed insns.pl to create the instruction tables in nasm.h and
676 names.c, so that a new instruction can be added by adding it *only*
678 * Added the following new instructions: SYSENTER, SYSEXIT, FXSAVE,
679 FXRSTOR, UD1, UD2 (the latter two are two opcodes that Intel
680 guarantee will never be used; one of them is documented as UD2 in
681 Intel documentation, the other one just as "Undefined Opcode" --
682 calling it UD1 seemed to make sense.)
683 * MAX_SYMBOL was defined to be 9, but LOADALL286 and LOADALL386 are 10
684 characters long. Now MAX_SYMBOL is derived from insns.dat.
685 * A note on the BASIC programs included: forget them. insns.bas is
686 already out of date. Get yourself a Perl interpreter for your
687 platform of choice at:
689 http://www.cpan.org/ports/index.html
695 added response file support, improved command line handling, new layout
698 fixed limit checking bug, 'OUT byte nn, reg' bug, and a couple of rdoff
699 related bugs, updated Wishlist; 0.98 Prerelease 3.
705 fixed bug in outcoff.c to do with truncating section names longer
706 than 8 characters, referencing beyond end of string; 0.98 pre-release 2
709 0.98 pre-released May 1999
710 --------------------------
712 Fixed a bug whereby STRUC didn't work at all in RDF.
714 Fixed a problem with group specification in PUBDEFs in OBJ.
716 Improved ease of adding new output formats. Contribution due to
719 Fixed a bug in relocations in the `bin' format: was showing up when
720 a relocatable reference crossed an 8192-byte boundary in any output
723 Fixed a bug in local labels: local-label lookups were inconsistent
724 between passes one and two if an EQU occurred between the definition
725 of a global label and the subsequent use of a local label local to
728 Fixed a seg-fault in the preprocessor (again) which happened when
729 you use a blank line as the first line of a multi-line macro
730 definition and then defined a label on the same line as a call to
733 Fixed a stale-pointer bug in the handling of the NASM environment
734 variable. Thanks to Thomas McWilliams.
736 ELF had a hard limit on the number of sections which caused
737 segfaults when transgressed. Fixed.
739 Added ability for ndisasm to read from stdin by using `-' as the
742 ndisasm wasn't outputting the TO keyword. Fixed.
744 Fixed error cascade on bogus expression in %if - an error in
745 evaluation was causing the entire %if to be discarded, thus creating
746 trouble later when the %else or %endif was encountered.
748 Forward reference tracking was instruction-granular not operand-
749 granular, which was causing 286-specific code to be generated
750 needlessly on code of the form `shr word [forwardref],1'. Thanks to
751 Jim Hague for sending a patch.
753 All messages now appear on stdout, as sending them to stderr serves
754 no useful purpose other than to make redirection difficult.
756 Fixed the problem with EQUs pointing to an external symbol - this
757 now generates an error message.
759 Allowed multiple size prefixes to an operand, of which only the first
760 is taken into account.
762 Incorporated John Fine's changes, including fixes of a large number
763 of preprocessor bugs, some small problems in OBJ, and a reworking of
764 label handling to define labels before their line is assembled, rather
767 Reformatted a lot of the source code to be more readable. Included
768 'coding.txt' as a guideline for how to format code for contributors.
770 Stopped nested %reps causing a panic - they now cause a slightly more
771 friendly error message instead.
773 Fixed floating point constant problems (patch by Pedro Gimeno)
775 Fixed the return value of insn_size() not being checked for -1, indicating
778 Incorporated 3D now instructions.
780 Fixed the 'mov eax, eax + ebx' bug.
782 Fixed the GLOBAL EQU bug in ELF. Released developers release 3.
784 Incorporated John Fine's command line parsing changes
786 Incorporated David Lindauer's OMF debug support
788 Made changes for LCC 4.0 support (__NASM_CDecl__, removed register size
789 specification warning when sizes agree).
791 Released NASM 0.98 Pre-release 1
794 0.97 released December 1997
795 ---------------------------
797 This was entirely a bug-fix release to 0.96, which seems to have got
800 Fixed stupid mistake in OBJ which caused `MOV EAX,<constant>' to
801 fail. Caused by an error in the `MOV EAX,<segment>' support.
803 ndisasm hung at EOF when compiled with lcc on Linux because lcc on
804 Linux somehow breaks feof(). ndisasm now does not rely on feof().
806 A heading in the documentation was missing due to a markup error in
809 Fixed failure to update all pointers on realloc() within extended-
810 operand code in parser.c. Was causing wrong behaviour and seg faults
811 on lines such as `dd 0.0,0.0,0.0,0.0,...'
813 Fixed a subtle preprocessor bug whereby invoking one multi-line
814 macro on the first line of the expansion of another, when the second
815 had been invoked with a label defined before it, didn't expand the
818 Added internal.doc back in to the distribution archives - it was
819 missing in 0.96 *blush*
821 Fixed bug causing 0.96 to be unable to assemble its own test files,
822 specifically objtest.asm. *blush again*
824 Fixed seg-faults and bogus error messages caused by mismatching
825 %rep and %endrep within multi-line macro definitions.
827 Fixed a problem with buffer overrun in OBJ, which was causing
828 corruption at ends of long PUBDEF records.
830 Separated DOS archives into main-program and documentation to reduce
834 0.96 released November 1997
835 ---------------------------
837 Fixed a bug whereby, if `nasm sourcefile' would cause a filename
838 collision warning and put output into `nasm.out', then `nasm
839 sourcefile -o outputfile' still gave the warning even though the
842 Fixed name pollution under Digital UNIX: one of its header files
843 defined R_SP, which broke the enum in nasm.h.
845 Fixed minor instruction table problems: FUCOM and FUCOMP didn't have
846 two-operand forms; NDISASM didn't recognise the longer register
847 forms of PUSH and POP (eg FF F3 for PUSH BX); TEST mem,imm32 was
848 flagged as undocumented; the 32-bit forms of CMOV had 16-bit operand
849 size prefixes; `AAD imm' and `AAM imm' are no longer flagged as
850 undocumented because the Intel Architecture reference documents
853 Fixed a problem with the local-label mechanism, whereby strange
854 types of symbol (EQUs, auto-defined OBJ segment base symbols)
855 interfered with the `previous global label' value and screwed up
858 Fixed a bug whereby the stub preprocessor didn't communicate with
859 the listing file generator, so that the -a and -l options in
860 conjunction would produce a useless listing file.
862 Merged `os2' object file format back into `obj', after discovering
863 that `obj' _also_ shouldn't have a link pass separator in a module
864 containing a non-trivial MODEND. Flat segments are now declared
865 using the FLAT attribute. `os2' is no longer a valid object format
868 Removed the fixed-size temporary storage in the evaluator. Very very
869 long expressions (like `mov ax,1+1+1+1+...' for two hundred 1s or
870 so) should now no longer crash NASM.
872 Fixed a bug involving segfaults on disassembly of MMX instructions,
873 by changing the meaning of one of the operand-type flags in nasm.h.
874 This may cause other apparently unrelated MMX problems; it needs to
875 be tested thoroughly.
877 Fixed some buffer overrun problems with large OBJ output files.
878 Thanks to DJ Delorie for the bug report and fix.
880 Made preprocess-only mode actually listen to the %line markers as it
881 prints them, so that it can report errors more sanely.
883 Re-designed the evaluator to keep more sensible track of expressions
884 involving forward references: can now cope with previously-nightmare
890 Added the ALIGN and ALIGNB standard macros.
892 Added PIC support in ELF: use of WRT to obtain the four extra
893 relocation types needed.
895 Added the ability for output file formats to define their own
896 extensions to the GLOBAL, COMMON and EXTERN directives.
898 Implemented common-variable alignment, and global-symbol type and
899 size declarations, in ELF.
901 Implemented NEAR and FAR keywords for common variables, plus
902 far-common element size specification, in OBJ.
904 Added a feature whereby EXTERNs and COMMONs in OBJ can be given a
905 default WRT specification (either a segment or a group).
907 Transformed the Unix NASM archive into an auto-configuring package.
909 Added a sanity-check for people applying SEG to things which are
910 already segment bases: this previously went unnoticed by the SEG
911 processing and caused OBJ-driver panics later.
913 Added the ability, in OBJ format, to deal with `MOV EAX,<segment>'
914 type references: OBJ doesn't directly support dword-size segment
915 base fixups, but as long as the low two bytes of the constant term
916 are zero, a word-size fixup can be generated instead and it will
919 Added the ability to specify sections' alignment requirements in
920 Win32 object files and pure binary files.
922 Added preprocess-time expression evaluation: the %assign (and
923 %iassign) directive and the bare %if (and %elif) conditional. Added
924 relational operators to the evaluator, for use only in %if
925 constructs: the standard relationals = < > <= >= <> (and C-like
926 synonyms == and !=) plus low-precedence logical operators &&, ^^ and
929 Added a preprocessor repeat construct: %rep / %exitrep / %endrep.
931 Added the __FILE__ and __LINE__ standard macros.
933 Added a sanity check for number constants being greater than
934 0xFFFFFFFF. The warning can be disabled.
936 Added the %0 token whereby a variadic multi-line macro can tell how
937 many parameters it's been given in a specific invocation.
939 Added %rotate, allowing multi-line macro parameters to be cycled.
941 Added the `*' option for the maximum parameter count on multi-line
942 macros, allowing them to take arbitrarily many parameters.
944 Added the ability for the user-level forms of EXTERN, GLOBAL and
945 COMMON to take more than one argument.
947 Added the IMPORT and EXPORT directives in OBJ format, to deal with
950 Added some more preprocessor %if constructs: %ifidn / %ifidni (exact
951 textual identity), and %ifid / %ifnum / %ifstr (token type testing).
953 Added the ability to distinguish SHL AX,1 (the 8086 version) from
954 SHL AX,BYTE 1 (the 286-and-upwards version whose constant happens to
957 Added NetBSD/FreeBSD/OpenBSD's variant of a.out format, complete
958 with PIC shared library features.
960 Changed NASM's idiosyncratic handling of FCLEX, FDISI, FENI, FINIT,
961 FSAVE, FSTCW, FSTENV, and FSTSW to bring it into line with the
962 otherwise accepted standard. The previous behaviour, though it was a
963 deliberate feature, was a deliberate feature based on a
964 misunderstanding. Apologies for the inconvenience.
966 Improved the flexibility of ABSOLUTE: you can now give it an
967 expression rather than being restricted to a constant, and it can
968 take relocatable arguments as well.
970 Added the ability for a variable to be declared as EXTERN multiple
971 times, and the subsequent definitions are just ignored.
973 We now allow instruction prefixes (CS, DS, LOCK, REPZ etc) to be
974 alone on a line (without a following instruction).
976 Improved sanity checks on whether the arguments to EXTERN, GLOBAL
977 and COMMON are valid identifiers.
979 Added misc/exebin.mac to allow direct generation of .EXE files by
980 hacking up an EXE header using DB and DW; also added test/binexe.asm
981 to demonstrate the use of this. Thanks to Yann Guidon for
982 contributing the EXE header code.
984 ndisasm forgot to check whether the input file had been successfully
985 opened. Now it does. Doh!
987 Added the Cyrix extensions to the MMX instruction set.
989 Added a hinting mechanism to allow [EAX+EBX] and [EBX+EAX] to be
990 assembled differently. This is important since [ESI+EBP] and
991 [EBP+ESI] have different default base segment registers.
993 Added support for the PharLap OMF extension for 4096-byte segment
997 0.95 released July 1997
998 -----------------------
1000 Fixed yet another ELF bug. This one manifested if the user relied on
1001 the default segment, and attempted to define global symbols without
1002 first explicitly declaring the target segment.
1004 Added makefiles (for NASM and the RDF tools) to build Win32 console
1005 apps under Symantec C++. Donated by Mark Junker.
1007 Added `macros.bas' and `insns.bas', QBasic versions of the Perl
1008 scripts that convert `standard.mac' to `macros.c' and convert
1009 `insns.dat' to `insnsa.c' and `insnsd.c'. Also thanks to Mark
1012 Changed the diassembled forms of the conditional instructions so
1013 that JB is now emitted as JC, and other similar changes. Suggested
1014 list by Ulrich Doewich.
1016 Added `@' to the list of valid characters to begin an identifier
1019 Documentary changes, notably the addition of the `Common Problems'
1020 section in nasm.doc.
1022 Fixed a bug relating to 32-bit PC-relative fixups in OBJ.
1024 Fixed a bug in perm_copy() in labels.c which was causing exceptions
1025 in cleanup_labels() on some systems.
1027 Positivity sanity check in TIMES argument changed from a warning to
1028 an error following a further complaint.
1030 Changed the acceptable limits on byte and word operands to allow
1031 things like `~10111001b' to work.
1033 Fixed a major problem in the preprocessor which caused seg-faults if
1034 macro definitions contained blank lines or comment-only lines.
1036 Fixed inadequate error checking on the commas separating the
1037 arguments to `db', `dw' etc.
1039 Fixed a crippling bug in the handling of macros with operand counts
1040 defined with a `+' modifier.
1042 Fixed a bug whereby object file formats which stored the input file
1043 name in the output file (such as OBJ and COFF) weren't doing so
1044 correctly when the output file name was specified on the command
1047 Removed [INC] and [INCLUDE] support for good, since they were
1050 Fixed a bug in OBJ which caused all fixups to be output in 16-bit
1051 (old-format) FIXUPP records, rather than putting the 32-bit ones in
1052 FIXUPP32 (new-format) records.
1054 Added, tentatively, OS/2 object file support (as a minor variant on
1057 Updates to Fox Cutter's Borland C makefile, Makefile.bc2.
1059 Removed a spurious second fclose() on the output file.
1061 Added the `-s' command line option to redirect all messages which
1062 would go to stderr (errors, help text) to stdout instead.
1064 Added the `-w' command line option to selectively suppress some
1065 classes of assembly warning messages.
1067 Added the `-p' pre-include and `-d' pre-define command-line options.
1069 Added an include file search path: the `-i' command line option.
1071 Fixed a silly little preprocessor bug whereby starting a line with a
1072 `%!' environment-variable reference caused an `unknown directive'
1075 Added the long-awaited listing file support: the `-l' command line
1078 Fixed a problem with OBJ format whereby, in the absence of any
1079 explicit segment definition, non-global symbols declared in the
1080 implicit default segment generated spurious EXTDEF records in the
1083 Added the NASM environment variable.
1085 From this version forward, Win32 console-mode binaries will be
1086 included in the DOS distribution in addition to the 16-bit binaries.
1087 Added Makefile.vc for this purpose.
1089 Added `return 0;' to test/objlink.c to prevent compiler warnings.
1091 Added the __NASM_MAJOR__ and __NASM_MINOR__ standard defines.
1093 Added an alternative memory-reference syntax in which prefixing an
1094 operand with `&' is equivalent to enclosing it in square brackets,
1095 at the request of Fox Cutter.
1097 Errors in pass two now cause the program to return a non-zero error
1098 code, which they didn't before.
1100 Fixed the single-line macro cycle detection, which didn't work at
1101 all on macros with no parameters (caused an infinite loop). Also
1102 changed the behaviour of single-line macro cycle detection to work
1103 like cpp, so that macros like `extrn' as given in the documentation
1106 Fixed the implementation of WRT, which was too restrictive in that
1107 you couldn't do `mov ax,[di+abc wrt dgroup]' because (di+abc) wasn't
1108 a relocatable reference.
1111 0.94 released April 1997
1112 ------------------------
1114 Major item: added the macro processor.
1116 Added undocumented instructions SMI, IBTS, XBTS and LOADALL286. Also
1117 reorganised CMPXCHG instruction into early-486 and Pentium forms.
1118 Thanks to Thobias Jones for the information.
1120 Fixed two more stupid bugs in ELF, which were causing `ld' to
1121 continue to seg-fault in a lot of non-trivial cases.
1123 Fixed a seg-fault in the label manager.
1125 Stopped FBLD and FBSTP from _requiring_ the TWORD keyword, which is
1126 the only option for BCD loads/stores in any case.
1128 Ensured FLDCW, FSTCW and FSTSW can cope with the WORD keyword, if
1129 anyone bothers to provide it. Previously they complained unless no
1130 keyword at all was present.
1132 Some forms of FDIV/FDIVR and FSUB/FSUBR were still inverted: a
1133 vestige of a bug that I thought had been fixed in 0.92. This was
1134 fixed, hopefully for good this time...
1136 Another minor phase error (insofar as a phase error can _ever_ be
1137 minor) fixed, this one occurring in code of the form
1138 rol ax,forward_reference
1139 forward_reference equ 1
1141 The number supplied to TIMES is now sanity-checked for positivity,
1142 and also may be greater than 64K (which previously didn't work on
1145 Added Watcom C makefiles, and misc/pmw.bat, donated by Dominik Behr.
1147 Added the INCBIN pseudo-opcode.
1149 Due to the advent of the preprocessor, the [INCLUDE] and [INC]
1150 directives have become obsolete. They are still supported in this
1151 version, with a warning, but won't be in the next.
1153 Fixed a bug in OBJ format, which caused incorrect object records to
1154 be output when absolute labels were made global.
1156 Updates to RDOFF subdirectory, and changes to outrdf.c.
1159 0.93 released January 1997
1160 --------------------------
1162 This release went out in a great hurry after semi-crippling bugs
1165 Really _did_ fix the stack overflows this time. *blush*
1167 Had problems with EA instruction sizes changing between passes, when
1168 an offset contained a forward reference and so 4 bytes were
1169 allocated for the offset in pass one; by pass two the symbol had
1170 been defined and happened to be a small absolute value, so only 1
1171 byte got allocated, causing instruction size mismatch between passes
1172 and hence incorrect address calculations. Fixed.
1174 Stupid bug in the revised ELF section generation fixed (associated
1175 string-table section for .symtab was hard-coded as 7, even when this
1176 didn't fit with the real section table). Was causing `ld' to
1177 seg-fault under Linux.
1179 Included a new Borland C makefile, Makefile.bc2, donated by Fox
1180 Cutter <lmb@comtch.iea.com>.
1183 0.92 released January 1997
1184 --------------------------
1186 The FDIVP/FDIVRP and FSUBP/FSUBRP pairs had been inverted: this was
1187 fixed. This also affected the LCC driver.
1189 Fixed a bug regarding 32-bit effective addresses of the form
1190 [other_register+ESP].
1192 Documentary changes, notably documentation of the fact that Borland
1193 Win32 compilers use `obj' rather than `win32' object format.
1195 Fixed the COMENT record in OBJ files, which was formatted
1198 Fixed a bug causing segfaults in large RDF files.
1200 OBJ format now strips initial periods from segment and group
1201 definitions, in order to avoid complications with the local label
1204 Fixed a bug in disassembling far calls and jumps in NDISASM.
1206 Added support for user-defined sections in COFF and ELF files.
1208 Compiled the DOS binaries with a sensible amount of stack, to
1209 prevent stack overflows on any arithmetic expression containing
1212 Fixed a bug in handling of files that do not terminate in a newline.
1215 0.91 released November 1996
1216 ---------------------------
1219 Support for RDF added.
1220 Support for DBG debugging format added.
1221 Support for 32-bit extensions to Microsoft OBJ format added.
1222 Revised for Borland C: some variable names changed, makefile added.
1223 LCC support revised to actually work.
1224 JMP/CALL NEAR/FAR notation added.
1225 `a16', `o16', `a32' and `o32' prefixes added.
1226 Range checking on short jumps implemented.
1227 MMX instruction support added.
1228 Negative floating point constant support added.
1229 Memory handling improved to bypass 64K barrier under DOS.
1230 $ prefix to force treatment of reserved words as identifiers added.
1231 Default-size mechanism for object formats added.
1232 Compile-time configurability added.
1233 `#', `@', `~' and `?' are now valid characters in labels.
1234 `-e' and `-k' options in NDISASM added.
1237 0.90 released October 1996
1238 --------------------------
1240 First release version. First support for object file output. Other
1241 changes from previous version (0.3x) too numerous to document.