| blob | ccf7760d37dabd6a56c693353882f8b9c2f7aa12 |
1 NASM Wishlist
2 =============
4 Numbers on right hand side are version numbers that it would be nice to
5 have this done by. ? means I haven't looked at it yet.
7 - Check misc/ide.cfg into RCS as Watcom IDE enhancement thingy. 0.98
8 (nop@dlc.fi)
10 - Package the Linux Assembler HOWTO. 0.98
12 - 3DNow!, SSE and other extensions need documenting. 0.98
13 hpa: Does it really make sense to have a whole instruction set
14 reference packaged with the assembler?
16 - prototypes of lrotate don't match in test/*. Fix. 0.98
18 - Build djgpp binaries for 0.98 onwards. Look into PMODE/W as a stub 0.98
19 - it might be a lot better than CWSDPMI. It's in PMW133.ZIP.
21 - %undef operator that goes along with %define DONE
23 - Fix `%error' giving error messages twice. 0.99
24 Not especially important, as changes planned for 1.1x below will make
25 the preprocessor be only called once.
27 - Sort out problems with OBJ: 0.99
28 * TLINK32 doesn't seem to like SEGDEF32 et al. So for that, we
29 should avoid xxx32 records wherever we can.
30 * However, didn't we change _to_ using xxx32 at some stage? Try
31 to remember why and when.
32 * Apparently Delphi's linker has trouble with two or more
33 globals being defined inside a PUBDEF32. Don't even know if it
34 _can_ cope with a PUBDEF16.
35 * Might need extra flags. *sigh*
37 - Symbol table output may possibly be useful. 0.99
38 Ken Martwick (kenm@efn.org) wants the following format:
39 labelname type offset(hex) repetition count
40 Possibly include xref addresses after repetition count?
42 - There are various other bugs in outelf.c that make certain kinds 0.99
43 of relocation not work. See zbrown.asm. Looks like we may have to do
44 a major rewrite of parts of it. Compare some NASM code output with
45 equivalent GAS code output. Look at the ELF spec. Generally fix things.
47 - NASM is currently using a kludge in ELF that involves defining 0.99
48 a symbol at a zero absolute offset. This isn't needed, as the
49 documented solution to the problem that this solves is to use
50 SHN_UNDEF.
52 - Debug information, in all formats it can be usefully done in. 0.99
53 * including line-number record support.
54 * "George C. Lindauer" <gclind01@starbase.spd.louisville.edu>
55 wants to have some say in how this goes through.
56 * Andrew Crabtree <andrewc@rosemail.rose.hp.com> wants to help out.
58 - Think about a line-continuation character. 0.99
60 - Consider allowing declaration of two labels on the same line,
61 syntax 'label1[:] label2[:] ... instruction'. Need to investigate
62 feasibility. 0.99
64 - Quoting of quotes by doubling them, in string and char constants. 0.99
66 - Two-operand syntax for SEGMENT/SECTION macro to avoid warnings 0.99
67 of ignored section parameters on reissue of __SECT__.
68 Or maybe skip the warning if the given parameters are identical to
69 what was actually stored. Investigate.
71 - Apparently we are not missing a PSRAQ instruction, because it
72 doesn't exist. Check that it doesn't exist as an undocumented
73 instruction, or something stupid like that. 0.99
75 - Any assembled form starting 0x80 can also start 0x82. ndisasm 1.00
76 should know this. New special code in instruction encodings,
77 probably.
79 - Pointing an EQU at an external symbol now generates an error. There 1.05
80 may be a better way of handling this; we should look into it.
81 Ideally, the label mechanism should be changed to cope with one
82 label being declared relative to another - that may work, but could be
83 a pain to implement (or is it? it may be easy enough that you just
84 need to declare a new offset in the same segment...) This should be done
85 before v1.0 is released. There is a comment regarding this in labels.c,
86 towards the end of the file, which discusses ways of fixing this.
88 - nested %rep used to cause a panic. Now a more informative error 1.10
89 message is produced. This problem whould be fixed before v1.0.
90 See comment in switch() statement block for PP_REP in do_directive()
91 in preproc.c (line 1585, or thereabouts)
93 - Contribution: zgraeme.tar contains improved hash table routines ?
94 contributed by Graeme Defty <graeme@HK.Super.NET> for use in the
95 label manager.
97 - Contribution: zsyntax.zip contains a syntax-highlighting mode for ?
98 NASM, for use with the Aurora text editor (??).
100 - Contribution: zvim.zip contains a syntax-highlighting mode for ?
101 NASM, for use with vim.
103 - Contribution: zkendal1.zip and zkendal2.zip contain Kendall ?
104 Bennett's (<KendallB@scitechsoft.com>) alternative syntax stuff,
105 providing an alternative syntax mode for NASM which allows a macro
106 set to be written that allows the same source files to be
107 assembled with NASM and TASM.
109 - Add the UD2 instruction. DONE
111 - Add the four instructions documented in 24368901.pdf (Intel's own DONE
112 document).
114 - Some means of avoiding MOV memoffs,EAX which apparently the 1.10?
115 Pentium pairing detector thinks modifies EAX. Similar means of
116 choosing instruction encodings where necessary.
118 - The example of ..@ makes it clear that a ..@ label isn't just ?
119 local, but doesn't make it clear that it isn't just global either.
121 - hpa wants an evaluator operator for ceil(log2(x)). ?
123 - Extra reloc types in ELF: R_386_16 type 20, PC16 is 21, 8 is 22, PC8 is 23.
124 Add support for the 16s at least. ?
127 - Lazy section creation or selective section output, in COFF/win32 ?
128 at least and probably other formats: don't bother to emit a section
129 if it contains no data. Particularly the default auto-created
130 section. We believe zero-length sections crash at least WLINK (in
131 win32).
133 - Make the flags field in `struct itemplate' in insns.h a long ?
134 instead of an int.
136 - Implement %ifref to check whether a single-line macro has ever been ?
137 expanded since (last re) definition. Or maybe not. We'll see.
139 - add pointer to \k{insLEAVE} and \k{insENTER} in chapters about ?
140 mixed-language programming.
142 - Some equivalent to TASM's GLOBAL directive, ie something which ?
143 defines a symbol as external if it doesn't end up being defined
144 but defines it as public if it does end up being defined.
146 - Documentation doesn't explain about C++ name mangling. ?
148 - see if BITS can be made to do anything sensible in obj (eg set the ?
149 default new-segment property to Use32).
151 - OBJ: coalesce consecutive offset and segment fixups for the same ?
152 location into full-32bit-pointer fixups. This is apparently
153 necessary because some twazzock in the PowerBASIC development
154 team didn't deign to support the OMF spec the way the rest of the
155 world sees it.
157 - Allow % to be separated from the rest of a preproc directive, for ?
158 alternative directive indentation styles.
160 - __DATE__, __TIME__, and text variants of __NASM_MAJOR__ and ?
161 __NASM_MINOR__.
163 - Warn on TIMES combined with multi-line macros. TIMES gets applied 1.00
164 to first line only - should bring to users' attention.
166 - Re-work the evaluator, again, with a per-object-format fixup 1.10
167 routine, so as to be able to cope with section offsets "really"
168 being pure numbers; should be able to allow at _least_ the two
169 common idioms
170 TIMES 510-$ DB 0 ; bootsector
171 MOV AX,(PROG_END-100H)/16 ; .COM TSR
172 Would need to call the fixup throughout the evaluator, and the
173 fixup would have to be allowed to return UNKNOWN on pass one if it
174 had to. (_Always_ returning UNKNOWN on pass one, though a lovely
175 clean design, breaks the first of the above examples.)
177 - Preprocessor identifier concatenation? 1.10
179 - Arbitrary section names in `bin'. ?
180 Is this necessary? Is it even desirable?
181 hpa: Desirable, yes. Necessary? Probably not, but there are
182 definitely cases where it becomes quite useful.
184 - Ability to read from a pipe. Obviously not useful under dos, so 1.10
185 memory problems with storing entire input file aren't a problem
186 either.
188 Related topic: file caching under DOS/32 bit... 1.10?
189 maybe even implement discardable buffers that get thrown away
190 when we get a NULL returned from malloc(). Only really useful under
191 DOS. Think about it.
193 Another related topic: possibly spool out the pre-processed 1.10?
194 stuff to a file, to avoid having to re-process it. Possible problems
195 with preprocessor values not known on pass 1? Have a look...
197 Or maybe we can spool out a pre-parsed version...? 1.10
198 Need to investigate feasibility. Does the results from the parser
199 change from pass 1 to pass 2? Would it be feasible to alter it so that
200 the parser returns an invariant result, and this is then processed
201 afterwards to resolve label references, etc?
203 - Subsection support? ?
205 - A good ALIGN mechanism, similar to GAS's. GAS pads out space by 1.10?
206 means of the following (32-bit) instructions:
207 8DB42600000000 lea esi,[esi+0x0]
208 8DB600000000 lea esi,[esi+0x0]
209 8D742600 lea esi,[esi+0x0]
210 8D7600 lea esi,[esi+0x0]
211 8D36 lea esi,[esi]
212 90 nop
213 It uses up to two of these instructions to do up to 14-byte pads;
214 when more than 14 bytes are needed, it issues a (short) jump to
215 the end of the padded section and then NOPs the rest. Come up with
216 a similar scheme for 16 bit mode, and also come up with a way to
217 use it - internal to the assembler, so that programs using ALIGN
218 don't knock over preprocess-only mode.
219 Also re-work the macro form so that when given one argument in a
220 code section it calls this feature.
222 - Possibly a means whereby FP constants can be specified as ?
223 immediate operands to non-FP instructions.
224 * Possible syntax: MOV EAX,FLOAT 1.2 to get a single-precision FP
225 constant. Then maybe MOV EAX,HI_FLOAT 1.2 and MOV EAX,LO_FLOAT
226 1.2 to get the two halves of a double-precision one. Best to
227 ignore extended-precision in case it bites.
228 * Alternatively, maybe MOV EAX,FLOAT(4,0-4,1.2) to get bytes 0-4
229 (ie 0-3) of a 4-byte constant. Then HI_FLOAT is FLOAT(8,4-8,x)
230 and LO_FLOAT is FLOAT(8,0-4,x). But this version allows two-byte
231 chunks, one-byte chunks, even stranger chunks, and pieces of
232 ten-byte reals to be bandied around as well.
234 - A UNION macro might be quite cool, now that ABSOLUTE is sane ?
235 enough to be able to handle it.
237 - An equivalent to gcc's ## stringify operator, plus string ?
238 concatenation, somehow implemented without undue ugliness, so as
239 to be able to do `%include "/my/path/%1"' in a macro, or something
240 similar...
242 - Actually _do_ something with the processor, privileged and 1.10
243 undocumented flags in the instruction table. When this happens,
244 consider allowing PMULHRW to map to either of the Cyrix or AMD
245 versions?
246 hpa: The -p option to ndisasm now uses this to some extent.
248 - Maybe NEC V20/V30 instructions? ?
249 hpa: What are they? Should be trivial to implement.
251 - Yet more object formats.
252 * Possibly direct support for .EXE files? 1.10
254 - Symbol map in binary format. Format-specific options... 1.10?
256 - REDESIGN: Think about EQU dependency, and about start-point 1.20?
257 specification in OBJ. Possibly re-think directive support.
259 - Think about a wrapper program like gcc? Possibly invent a _patch_ 2.00?
260 for gcc so that it can take .asm files on the command line?
262 - If a wrapper happens, think about adding an option to cause the ?
263 resulting executable file to be executed immediately, thus
264 allowing NASM source files to have #!... (probably silly)
266 - Multi-platform support? If so: definitely Alpha; possibly Java ?
267 byte code; probably ARM/StrongARM; maybe Sparc; maybe Mips; maybe
268 Vax. Perhaps Z80 and 6502, just for a laugh?
270 - Consider a 'verbose' option that prints information about the ?
271 resulting object file onto stdout.
273 - Line numbers in the .lst file don't match the line numbers in the ?
274 input. They probably should, rather than the current matching
275 of the post-preprocessor line numbers.