1 #!/bin/awk -f
2 # SPDX-License-Identifier: GPL-2.0
3 # gen-insn-attr-x86.awk: Instruction attribute table generator
4 # Written by Masami Hiramatsu <mhiramat@redhat.com>
5 #
6 # Usage: awk -f gen-insn-attr-x86.awk x86-opcode-map.txt > inat-tables.c
7
8 # Awk implementation sanity check
9 function check_awk_implement() {
10 if (sprintf("%x", 0) != "0")
11 return "Your awk has a printf-format problem."
12 return ""
13 }
14
15 # Clear working vars
16 function clear_vars() {
17 delete table
18 delete lptable2
19 delete lptable1
20 delete lptable3
21 eid = -1 # escape id
22 gid = -1 # group id
23 aid = -1 # AVX id
24 tname = ""
25 }
26
27 BEGIN {
28 # Implementation error checking
29 awkchecked = check_awk_implement()
30 if (awkchecked != "") {
31 print "Error: " awkchecked > "/dev/stderr"
32 print "Please try to use gawk." > "/dev/stderr"
33 exit 1
34 }
35
36 # Setup generating tables
37 print "/* x86 opcode map generated from x86-opcode-map.txt */"
38 print "/* Do not change this code. */\n"
39 ggid = 1
40 geid = 1
41 gaid = 0
42 delete etable
43 delete gtable
44 delete atable
45
46 opnd_expr = "^[A-Za-z/]"
47 ext_expr = "^\\("
48 sep_expr = "^\\|$"
49 group_expr = "^Grp[0-9A-Za-z]+"
50
51 imm_expr = "^[IJAOL][a-z]"
52 imm_flag["Ib"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
53 imm_flag["Jb"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
54 imm_flag["Iw"] = "INAT_MAKE_IMM(INAT_IMM_WORD)"
55 imm_flag["Id"] = "INAT_MAKE_IMM(INAT_IMM_DWORD)"
56 imm_flag["Iq"] = "INAT_MAKE_IMM(INAT_IMM_QWORD)"
57 imm_flag["Ap"] = "INAT_MAKE_IMM(INAT_IMM_PTR)"
58 imm_flag["Iz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)"
59 imm_flag["Jz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)"
60 imm_flag["Iv"] = "INAT_MAKE_IMM(INAT_IMM_VWORD)"
61 imm_flag["Ob"] = "INAT_MOFFSET"
62 imm_flag["Ov"] = "INAT_MOFFSET"
63 imm_flag["Lx"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
64
65 modrm_expr = "^([CDEGMNPQRSUVW/][a-z]+|NTA|T[012])"
66 force64_expr = "\\([df]64\\)"
67 rex_expr = "^((REX(\\.[XRWB]+)+)|(REX$))"
68 rex2_expr = "\\(REX2\\)"
69 no_rex2_expr = "\\(!REX2\\)"
70 fpu_expr = "^ESC" # TODO
71
72 lprefix1_expr = "\\((66|!F3)\\)"
73 lprefix2_expr = "\\(F3\\)"
74 lprefix3_expr = "\\((F2|!F3|66&F2)\\)"
75 lprefix_expr = "\\((66|F2|F3)\\)"
76 max_lprefix = 4
77
78 # All opcodes starting with lower-case 'v', 'k' or with (v1) superscript
79 # accepts VEX prefix
80 vexok_opcode_expr = "^[vk].*"
81 vexok_expr = "\\(v1\\)"
82 # All opcodes with (v) superscript supports *only* VEX prefix
83 vexonly_expr = "\\(v\\)"
84 # All opcodes with (ev) superscript supports *only* EVEX prefix
85 evexonly_expr = "\\(ev\\)"
86 # (es) is the same as (ev) but also "SCALABLE" i.e. W and pp determine operand size
87 evex_scalable_expr = "\\(es\\)"
88
89 prefix_expr = "\\(Prefix\\)"
90 prefix_num["Operand-Size"] = "INAT_PFX_OPNDSZ"
91 prefix_num["REPNE"] = "INAT_PFX_REPNE"
92 prefix_num["REP/REPE"] = "INAT_PFX_REPE"
93 prefix_num["XACQUIRE"] = "INAT_PFX_REPNE"
94 prefix_num["XRELEASE"] = "INAT_PFX_REPE"
95 prefix_num["LOCK"] = "INAT_PFX_LOCK"
96 prefix_num["SEG=CS"] = "INAT_PFX_CS"
97 prefix_num["SEG=DS"] = "INAT_PFX_DS"
98 prefix_num["SEG=ES"] = "INAT_PFX_ES"
99 prefix_num["SEG=FS"] = "INAT_PFX_FS"
100 prefix_num["SEG=GS"] = "INAT_PFX_GS"
101 prefix_num["SEG=SS"] = "INAT_PFX_SS"
102 prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ"
103 prefix_num["VEX+1byte"] = "INAT_PFX_VEX2"
104 prefix_num["VEX+2byte"] = "INAT_PFX_VEX3"
105 prefix_num["EVEX"] = "INAT_PFX_EVEX"
106 prefix_num["REX2"] = "INAT_PFX_REX2"
107
108 clear_vars()
109 }
110
111 function semantic_error(msg) {
112 print "Semantic error at " NR ": " msg > "/dev/stderr"
113 exit 1
114 }
115
116 function debug(msg) {
117 print "DEBUG: " msg
118 }
119
120 function array_size(arr, i,c) {
121 c = 0
122 for (i in arr)
123 c++
124 return c
125 }
126
127 /^Table:/ {
128 print "/* " $0 " */"
129 if (tname != "")
130 semantic_error("Hit Table: before EndTable:.");
131 }
132
133 /^Referrer:/ {
134 if (NF != 1) {
135 # escape opcode table
136 ref = ""
137 for (i = 2; i <= NF; i++)
138 ref = ref $i
139 eid = escape[ref]
140 tname = sprintf("inat_escape_table_%d", eid)
141 }
142 }
143
144 /^AVXcode:/ {
145 if (NF != 1) {
146 # AVX/escape opcode table
147 aid = $2
148 if (gaid <= aid)
149 gaid = aid + 1
150 if (tname == "") # AVX only opcode table
151 tname = sprintf("inat_avx_table_%d", $2)
152 }
153 if (aid == -1 && eid == -1) # primary opcode table
154 tname = "inat_primary_table"
155 }
156
157 /^GrpTable:/ {
158 print "/* " $0 " */"
159 if (!($2 in group))
160 semantic_error("No group: " $2 )
161 gid = group[$2]
162 tname = "inat_group_table_" gid
163 }
164
165 function print_table(tbl,name,fmt,n)
166 {
167 print "const insn_attr_t " name " = {"
168 for (i = 0; i < n; i++) {
169 id = sprintf(fmt, i)
170 if (tbl[id])
171 print " [" id "] = " tbl[id] ","
172 }
173 print "};"
174 }
175
176 /^EndTable/ {
177 if (gid != -1) {
178 # print group tables
179 if (array_size(table) != 0) {
180 print_table(table, tname "[INAT_GROUP_TABLE_SIZE]",
181 "0x%x", 8)
182 gtable[gid,0] = tname
183 }
184 if (array_size(lptable1) != 0) {
185 print_table(lptable1, tname "_1[INAT_GROUP_TABLE_SIZE]",
186 "0x%x", 8)
187 gtable[gid,1] = tname "_1"
188 }
189 if (array_size(lptable2) != 0) {
190 print_table(lptable2, tname "_2[INAT_GROUP_TABLE_SIZE]",
191 "0x%x", 8)
192 gtable[gid,2] = tname "_2"
193 }
194 if (array_size(lptable3) != 0) {
195 print_table(lptable3, tname "_3[INAT_GROUP_TABLE_SIZE]",
196 "0x%x", 8)
197 gtable[gid,3] = tname "_3"
198 }
199 } else {
200 # print primary/escaped tables
201 if (array_size(table) != 0) {
202 print_table(table, tname "[INAT_OPCODE_TABLE_SIZE]",
203 "0x%02x", 256)
204 etable[eid,0] = tname
205 if (aid >= 0)
206 atable[aid,0] = tname
207 }
208 if (array_size(lptable1) != 0) {
209 print_table(lptable1,tname "_1[INAT_OPCODE_TABLE_SIZE]",
210 "0x%02x", 256)
211 etable[eid,1] = tname "_1"
212 if (aid >= 0)
213 atable[aid,1] = tname "_1"
214 }
215 if (array_size(lptable2) != 0) {
216 print_table(lptable2,tname "_2[INAT_OPCODE_TABLE_SIZE]",
217 "0x%02x", 256)
218 etable[eid,2] = tname "_2"
219 if (aid >= 0)
220 atable[aid,2] = tname "_2"
221 }
222 if (array_size(lptable3) != 0) {
223 print_table(lptable3,tname "_3[INAT_OPCODE_TABLE_SIZE]",
224 "0x%02x", 256)
225 etable[eid,3] = tname "_3"
226 if (aid >= 0)
227 atable[aid,3] = tname "_3"
228 }
229 }
230 print ""
231 clear_vars()
232 }
233
234 function add_flags(old,new) {
235 if (old && new)
236 return old " | " new
237 else if (old)
238 return old
239 else
240 return new
241 }
242
243 # convert operands to flags.
244 function convert_operands(count,opnd, i,j,imm,mod)
245 {
246 imm = null
247 mod = null
248 for (j = 1; j <= count; j++) {
249 i = opnd[j]
250 if (match(i, imm_expr) == 1) {
251 if (!imm_flag[i])
252 semantic_error("Unknown imm opnd: " i)
253 if (imm) {
254 if (i != "Ib")
255 semantic_error("Second IMM error")
256 imm = add_flags(imm, "INAT_SCNDIMM")
257 } else
258 imm = imm_flag[i]
259 } else if (match(i, modrm_expr))
260 mod = "INAT_MODRM"
261 }
262 return add_flags(imm, mod)
263 }
264
265 /^[0-9a-f]+:/ {
266 if (NR == 1)
267 next
268 # get index
269 idx = "0x" substr($1, 1, index($1,":") - 1)
270 if (idx in table)
271 semantic_error("Redefine " idx " in " tname)
272
273 # check if escaped opcode
274 if ("escape" == $2) {
275 if ($3 != "#")
276 semantic_error("No escaped name")
277 ref = ""
278 for (i = 4; i <= NF; i++)
279 ref = ref $i
280 if (ref in escape)
281 semantic_error("Redefine escape (" ref ")")
282 escape[ref] = geid
283 geid++
284 table[idx] = "INAT_MAKE_ESCAPE(" escape[ref] ")"
285 next
286 }
287
288 variant = null
289 # converts
290 i = 2
291 while (i <= NF) {
292 opcode = $(i++)
293 delete opnds
294 ext = null
295 flags = null
296 opnd = null
297 # parse one opcode
298 if (match($i, opnd_expr)) {
299 opnd = $i
300 count = split($(i++), opnds, ",")
301 flags = convert_operands(count, opnds)
302 }
303 if (match($i, ext_expr))
304 ext = $(i++)
305 if (match($i, sep_expr))
306 i++
307 else if (i < NF)
308 semantic_error($i " is not a separator")
309
310 # check if group opcode
311 if (match(opcode, group_expr)) {
312 if (!(opcode in group)) {
313 group[opcode] = ggid
314 ggid++
315 }
316 flags = add_flags(flags, "INAT_MAKE_GROUP(" group[opcode] ")")
317 }
318 # check force(or default) 64bit
319 if (match(ext, force64_expr))
320 flags = add_flags(flags, "INAT_FORCE64")
321
322 # check REX2 not allowed
323 if (match(ext, no_rex2_expr))
324 flags = add_flags(flags, "INAT_NO_REX2")
325
326 # check REX prefix
327 if (match(opcode, rex_expr))
328 flags = add_flags(flags, "INAT_MAKE_PREFIX(INAT_PFX_REX)")
329
330 # check coprocessor escape : TODO
331 if (match(opcode, fpu_expr))
332 flags = add_flags(flags, "INAT_MODRM")
333
334 # check VEX codes
335 if (match(ext, evexonly_expr))
336 flags = add_flags(flags, "INAT_VEXOK | INAT_EVEXONLY")
337 else if (match(ext, evex_scalable_expr))
338 flags = add_flags(flags, "INAT_VEXOK | INAT_EVEXONLY | INAT_EVEX_SCALABLE")
339 else if (match(ext, vexonly_expr))
340 flags = add_flags(flags, "INAT_VEXOK | INAT_VEXONLY")
341 else if (match(ext, vexok_expr) || match(opcode, vexok_opcode_expr))
342 flags = add_flags(flags, "INAT_VEXOK")
343
344 # check prefixes
345 if (match(ext, prefix_expr)) {
346 if (!prefix_num[opcode])
347 semantic_error("Unknown prefix: " opcode)
348 flags = add_flags(flags, "INAT_MAKE_PREFIX(" prefix_num[opcode] ")")
349 }
350 if (length(flags) == 0)
351 continue
352 # check if last prefix
353 if (match(ext, lprefix1_expr)) {
354 lptable1[idx] = add_flags(lptable1[idx],flags)
355 variant = "INAT_VARIANT"
356 }
357 if (match(ext, lprefix2_expr)) {
358 lptable2[idx] = add_flags(lptable2[idx],flags)
359 variant = "INAT_VARIANT"
360 }
361 if (match(ext, lprefix3_expr)) {
362 lptable3[idx] = add_flags(lptable3[idx],flags)
363 variant = "INAT_VARIANT"
364 }
365 if (match(ext, rex2_expr))
366 table[idx] = add_flags(table[idx], "INAT_REX2_VARIANT")
367 if (!match(ext, lprefix_expr)){
368 table[idx] = add_flags(table[idx],flags)
369 }
370 }
371 if (variant)
372 table[idx] = add_flags(table[idx],variant)
373 }
374
375 END {
376 if (awkchecked != "")
377 exit 1
378
379 print "#ifndef __BOOT_COMPRESSED\n"
380
381 # print escape opcode map's array
382 print "/* Escape opcode map array */"
383 print "const insn_attr_t * const inat_escape_tables[INAT_ESC_MAX + 1]" \
384 "[INAT_LSTPFX_MAX + 1] = {"
385 for (i = 0; i < geid; i++)
386 for (j = 0; j < max_lprefix; j++)
387 if (etable[i,j])
388 print " ["i"]["j"] = "etable[i,j]","
389 print "};\n"
390 # print group opcode map's array
391 print "/* Group opcode map array */"
392 print "const insn_attr_t * const inat_group_tables[INAT_GRP_MAX + 1]"\
393 "[INAT_LSTPFX_MAX + 1] = {"
394 for (i = 0; i < ggid; i++)
395 for (j = 0; j < max_lprefix; j++)
396 if (gtable[i,j])
397 print " ["i"]["j"] = "gtable[i,j]","
398 print "};\n"
399 # print AVX opcode map's array
400 print "/* AVX opcode map array */"
401 print "const insn_attr_t * const inat_avx_tables[X86_VEX_M_MAX + 1]"\
402 "[INAT_LSTPFX_MAX + 1] = {"
403 for (i = 0; i < gaid; i++)
404 for (j = 0; j < max_lprefix; j++)
405 if (atable[i,j])
406 print " ["i"]["j"] = "atable[i,j]","
407 print "};\n"
408
409 print "#else /* !__BOOT_COMPRESSED */\n"
410
411 print "/* Escape opcode map array */"
412 print "static const insn_attr_t *inat_escape_tables[INAT_ESC_MAX + 1]" \
413 "[INAT_LSTPFX_MAX + 1];"
414 print ""
415
416 print "/* Group opcode map array */"
417 print "static const insn_attr_t *inat_group_tables[INAT_GRP_MAX + 1]"\
418 "[INAT_LSTPFX_MAX + 1];"
419 print ""
420
421 print "/* AVX opcode map array */"
422 print "static const insn_attr_t *inat_avx_tables[X86_VEX_M_MAX + 1]"\
423 "[INAT_LSTPFX_MAX + 1];"
424 print ""
425
426 print "static void inat_init_tables(void)"
427 print "{"
428
429 # print escape opcode map's array
430 print "\t/* Print Escape opcode map array */"
431 for (i = 0; i < geid; i++)
432 for (j = 0; j < max_lprefix; j++)
433 if (etable[i,j])
434 print "\tinat_escape_tables["i"]["j"] = "etable[i,j]";"
435 print ""
436
437 # print group opcode map's array
438 print "\t/* Print Group opcode map array */"
439 for (i = 0; i < ggid; i++)
440 for (j = 0; j < max_lprefix; j++)
441 if (gtable[i,j])
442 print "\tinat_group_tables["i"]["j"] = "gtable[i,j]";"
443 print ""
444 # print AVX opcode map's array
445 print "\t/* Print AVX opcode map array */"
446 for (i = 0; i < gaid; i++)
447 for (j = 0; j < max_lprefix; j++)
448 if (atable[i,j])
449 print "\tinat_avx_tables["i"]["j"] = "atable[i,j]";"
450
451 print "}"
452 print "#endif"
453 }
454
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.