Commit | Line | Data |
---|---|---|
eb13296c MH |
1 | #!/bin/awk -f |
2 | # gen-insn-attr-x86.awk: Instruction attribute table generator | |
3 | # Written by Masami Hiramatsu <mhiramat@redhat.com> | |
4 | # | |
5 | # Usage: awk -f gen-insn-attr-x86.awk x86-opcode-map.txt > inat-tables.c | |
6 | ||
69d991f3 MH |
7 | # Awk implementation sanity check |
8 | function check_awk_implement() { | |
69d991f3 MH |
9 | if (sprintf("%x", 0) != "0") |
10 | return "Your awk has a printf-format problem." | |
11 | return "" | |
12 | } | |
13 | ||
e0e492e9 MH |
14 | # Clear working vars |
15 | function clear_vars() { | |
16 | delete table | |
17 | delete lptable2 | |
18 | delete lptable1 | |
19 | delete lptable3 | |
20 | eid = -1 # escape id | |
21 | gid = -1 # group id | |
22 | aid = -1 # AVX id | |
23 | tname = "" | |
24 | } | |
25 | ||
eb13296c | 26 | BEGIN { |
69d991f3 MH |
27 | # Implementation error checking |
28 | awkchecked = check_awk_implement() | |
29 | if (awkchecked != "") { | |
30 | print "Error: " awkchecked > "/dev/stderr" | |
31 | print "Please try to use gawk." > "/dev/stderr" | |
32 | exit 1 | |
33 | } | |
34 | ||
35 | # Setup generating tables | |
eb13296c | 36 | print "/* x86 opcode map generated from x86-opcode-map.txt */" |
e0e492e9 | 37 | print "/* Do not change this code. */\n" |
eb13296c MH |
38 | ggid = 1 |
39 | geid = 1 | |
e0e492e9 MH |
40 | gaid = 0 |
41 | delete etable | |
42 | delete gtable | |
43 | delete atable | |
eb13296c | 44 | |
4beb3d6d | 45 | opnd_expr = "^[A-Za-z/]" |
eb13296c MH |
46 | ext_expr = "^\\(" |
47 | sep_expr = "^\\|$" | |
4beb3d6d | 48 | group_expr = "^Grp[0-9A-Za-z]+" |
eb13296c | 49 | |
4beb3d6d | 50 | imm_expr = "^[IJAO][a-z]" |
eb13296c MH |
51 | imm_flag["Ib"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)" |
52 | imm_flag["Jb"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)" | |
53 | imm_flag["Iw"] = "INAT_MAKE_IMM(INAT_IMM_WORD)" | |
54 | imm_flag["Id"] = "INAT_MAKE_IMM(INAT_IMM_DWORD)" | |
55 | imm_flag["Iq"] = "INAT_MAKE_IMM(INAT_IMM_QWORD)" | |
56 | imm_flag["Ap"] = "INAT_MAKE_IMM(INAT_IMM_PTR)" | |
57 | imm_flag["Iz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)" | |
58 | imm_flag["Jz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)" | |
59 | imm_flag["Iv"] = "INAT_MAKE_IMM(INAT_IMM_VWORD)" | |
60 | imm_flag["Ob"] = "INAT_MOFFSET" | |
61 | imm_flag["Ov"] = "INAT_MOFFSET" | |
62 | ||
4beb3d6d | 63 | modrm_expr = "^([CDEGMNPQRSUVW/][a-z]+|NTA|T[012])" |
eb13296c MH |
64 | force64_expr = "\\([df]64\\)" |
65 | rex_expr = "^REX(\\.[XRWB]+)*" | |
66 | fpu_expr = "^ESC" # TODO | |
67 | ||
68 | lprefix1_expr = "\\(66\\)" | |
e0e492e9 MH |
69 | lprefix2_expr = "\\(F3\\)" |
70 | lprefix3_expr = "\\(F2\\)" | |
eb13296c MH |
71 | max_lprefix = 4 |
72 | ||
e0e492e9 MH |
73 | vexok_expr = "\\(VEX\\)" |
74 | vexonly_expr = "\\(oVEX\\)" | |
75 | ||
eb13296c MH |
76 | prefix_expr = "\\(Prefix\\)" |
77 | prefix_num["Operand-Size"] = "INAT_PFX_OPNDSZ" | |
78 | prefix_num["REPNE"] = "INAT_PFX_REPNE" | |
79 | prefix_num["REP/REPE"] = "INAT_PFX_REPE" | |
80 | prefix_num["LOCK"] = "INAT_PFX_LOCK" | |
81 | prefix_num["SEG=CS"] = "INAT_PFX_CS" | |
82 | prefix_num["SEG=DS"] = "INAT_PFX_DS" | |
83 | prefix_num["SEG=ES"] = "INAT_PFX_ES" | |
84 | prefix_num["SEG=FS"] = "INAT_PFX_FS" | |
85 | prefix_num["SEG=GS"] = "INAT_PFX_GS" | |
86 | prefix_num["SEG=SS"] = "INAT_PFX_SS" | |
87 | prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ" | |
e0e492e9 MH |
88 | prefix_num["2bytes-VEX"] = "INAT_PFX_VEX2" |
89 | prefix_num["3bytes-VEX"] = "INAT_PFX_VEX3" | |
eb13296c | 90 | |
e0e492e9 | 91 | clear_vars() |
eb13296c MH |
92 | } |
93 | ||
94 | function semantic_error(msg) { | |
95 | print "Semantic error at " NR ": " msg > "/dev/stderr" | |
96 | exit 1 | |
97 | } | |
98 | ||
99 | function debug(msg) { | |
100 | print "DEBUG: " msg | |
101 | } | |
102 | ||
103 | function array_size(arr, i,c) { | |
104 | c = 0 | |
105 | for (i in arr) | |
106 | c++ | |
107 | return c | |
108 | } | |
109 | ||
110 | /^Table:/ { | |
111 | print "/* " $0 " */" | |
e0e492e9 MH |
112 | if (tname != "") |
113 | semantic_error("Hit Table: before EndTable:."); | |
eb13296c MH |
114 | } |
115 | ||
116 | /^Referrer:/ { | |
e0e492e9 | 117 | if (NF != 1) { |
eb13296c MH |
118 | # escape opcode table |
119 | ref = "" | |
120 | for (i = 2; i <= NF; i++) | |
121 | ref = ref $i | |
122 | eid = escape[ref] | |
123 | tname = sprintf("inat_escape_table_%d", eid) | |
124 | } | |
125 | } | |
126 | ||
e0e492e9 MH |
127 | /^AVXcode:/ { |
128 | if (NF != 1) { | |
129 | # AVX/escape opcode table | |
130 | aid = $2 | |
131 | if (gaid <= aid) | |
132 | gaid = aid + 1 | |
133 | if (tname == "") # AVX only opcode table | |
134 | tname = sprintf("inat_avx_table_%d", $2) | |
135 | } | |
136 | if (aid == -1 && eid == -1) # primary opcode table | |
137 | tname = "inat_primary_table" | |
138 | } | |
139 | ||
eb13296c MH |
140 | /^GrpTable:/ { |
141 | print "/* " $0 " */" | |
142 | if (!($2 in group)) | |
143 | semantic_error("No group: " $2 ) | |
144 | gid = group[$2] | |
145 | tname = "inat_group_table_" gid | |
146 | } | |
147 | ||
148 | function print_table(tbl,name,fmt,n) | |
149 | { | |
150 | print "const insn_attr_t " name " = {" | |
151 | for (i = 0; i < n; i++) { | |
152 | id = sprintf(fmt, i) | |
153 | if (tbl[id]) | |
154 | print " [" id "] = " tbl[id] "," | |
155 | } | |
156 | print "};" | |
157 | } | |
158 | ||
159 | /^EndTable/ { | |
160 | if (gid != -1) { | |
161 | # print group tables | |
162 | if (array_size(table) != 0) { | |
163 | print_table(table, tname "[INAT_GROUP_TABLE_SIZE]", | |
164 | "0x%x", 8) | |
165 | gtable[gid,0] = tname | |
166 | } | |
167 | if (array_size(lptable1) != 0) { | |
168 | print_table(lptable1, tname "_1[INAT_GROUP_TABLE_SIZE]", | |
169 | "0x%x", 8) | |
170 | gtable[gid,1] = tname "_1" | |
171 | } | |
172 | if (array_size(lptable2) != 0) { | |
173 | print_table(lptable2, tname "_2[INAT_GROUP_TABLE_SIZE]", | |
174 | "0x%x", 8) | |
175 | gtable[gid,2] = tname "_2" | |
176 | } | |
177 | if (array_size(lptable3) != 0) { | |
178 | print_table(lptable3, tname "_3[INAT_GROUP_TABLE_SIZE]", | |
179 | "0x%x", 8) | |
180 | gtable[gid,3] = tname "_3" | |
181 | } | |
182 | } else { | |
183 | # print primary/escaped tables | |
184 | if (array_size(table) != 0) { | |
185 | print_table(table, tname "[INAT_OPCODE_TABLE_SIZE]", | |
186 | "0x%02x", 256) | |
187 | etable[eid,0] = tname | |
e0e492e9 MH |
188 | if (aid >= 0) |
189 | atable[aid,0] = tname | |
eb13296c MH |
190 | } |
191 | if (array_size(lptable1) != 0) { | |
192 | print_table(lptable1,tname "_1[INAT_OPCODE_TABLE_SIZE]", | |
193 | "0x%02x", 256) | |
194 | etable[eid,1] = tname "_1" | |
e0e492e9 MH |
195 | if (aid >= 0) |
196 | atable[aid,1] = tname "_1" | |
eb13296c MH |
197 | } |
198 | if (array_size(lptable2) != 0) { | |
199 | print_table(lptable2,tname "_2[INAT_OPCODE_TABLE_SIZE]", | |
200 | "0x%02x", 256) | |
201 | etable[eid,2] = tname "_2" | |
e0e492e9 MH |
202 | if (aid >= 0) |
203 | atable[aid,2] = tname "_2" | |
eb13296c MH |
204 | } |
205 | if (array_size(lptable3) != 0) { | |
206 | print_table(lptable3,tname "_3[INAT_OPCODE_TABLE_SIZE]", | |
207 | "0x%02x", 256) | |
208 | etable[eid,3] = tname "_3" | |
e0e492e9 MH |
209 | if (aid >= 0) |
210 | atable[aid,3] = tname "_3" | |
eb13296c MH |
211 | } |
212 | } | |
213 | print "" | |
e0e492e9 | 214 | clear_vars() |
eb13296c MH |
215 | } |
216 | ||
217 | function add_flags(old,new) { | |
218 | if (old && new) | |
219 | return old " | " new | |
220 | else if (old) | |
221 | return old | |
222 | else | |
223 | return new | |
224 | } | |
225 | ||
226 | # convert operands to flags. | |
23637568 | 227 | function convert_operands(count,opnd, i,j,imm,mod) |
eb13296c MH |
228 | { |
229 | imm = null | |
230 | mod = null | |
23637568 JN |
231 | for (j = 1; j <= count; j++) { |
232 | i = opnd[j] | |
eb13296c MH |
233 | if (match(i, imm_expr) == 1) { |
234 | if (!imm_flag[i]) | |
235 | semantic_error("Unknown imm opnd: " i) | |
236 | if (imm) { | |
237 | if (i != "Ib") | |
238 | semantic_error("Second IMM error") | |
239 | imm = add_flags(imm, "INAT_SCNDIMM") | |
240 | } else | |
241 | imm = imm_flag[i] | |
242 | } else if (match(i, modrm_expr)) | |
243 | mod = "INAT_MODRM" | |
244 | } | |
245 | return add_flags(imm, mod) | |
246 | } | |
247 | ||
248 | /^[0-9a-f]+\:/ { | |
249 | if (NR == 1) | |
250 | next | |
251 | # get index | |
252 | idx = "0x" substr($1, 1, index($1,":") - 1) | |
253 | if (idx in table) | |
254 | semantic_error("Redefine " idx " in " tname) | |
255 | ||
256 | # check if escaped opcode | |
257 | if ("escape" == $2) { | |
258 | if ($3 != "#") | |
259 | semantic_error("No escaped name") | |
260 | ref = "" | |
261 | for (i = 4; i <= NF; i++) | |
262 | ref = ref $i | |
263 | if (ref in escape) | |
264 | semantic_error("Redefine escape (" ref ")") | |
265 | escape[ref] = geid | |
266 | geid++ | |
267 | table[idx] = "INAT_MAKE_ESCAPE(" escape[ref] ")" | |
268 | next | |
269 | } | |
270 | ||
271 | variant = null | |
272 | # converts | |
273 | i = 2 | |
274 | while (i <= NF) { | |
275 | opcode = $(i++) | |
276 | delete opnds | |
277 | ext = null | |
278 | flags = null | |
279 | opnd = null | |
280 | # parse one opcode | |
281 | if (match($i, opnd_expr)) { | |
282 | opnd = $i | |
23637568 JN |
283 | count = split($(i++), opnds, ",") |
284 | flags = convert_operands(count, opnds) | |
eb13296c MH |
285 | } |
286 | if (match($i, ext_expr)) | |
287 | ext = $(i++) | |
288 | if (match($i, sep_expr)) | |
289 | i++ | |
290 | else if (i < NF) | |
291 | semantic_error($i " is not a separator") | |
292 | ||
293 | # check if group opcode | |
294 | if (match(opcode, group_expr)) { | |
295 | if (!(opcode in group)) { | |
296 | group[opcode] = ggid | |
297 | ggid++ | |
298 | } | |
299 | flags = add_flags(flags, "INAT_MAKE_GROUP(" group[opcode] ")") | |
300 | } | |
301 | # check force(or default) 64bit | |
302 | if (match(ext, force64_expr)) | |
303 | flags = add_flags(flags, "INAT_FORCE64") | |
304 | ||
305 | # check REX prefix | |
306 | if (match(opcode, rex_expr)) | |
04d46c1b | 307 | flags = add_flags(flags, "INAT_MAKE_PREFIX(INAT_PFX_REX)") |
eb13296c MH |
308 | |
309 | # check coprocessor escape : TODO | |
310 | if (match(opcode, fpu_expr)) | |
311 | flags = add_flags(flags, "INAT_MODRM") | |
312 | ||
e0e492e9 MH |
313 | # check VEX only code |
314 | if (match(ext, vexonly_expr)) | |
315 | flags = add_flags(flags, "INAT_VEXOK | INAT_VEXONLY") | |
316 | ||
317 | # check VEX only code | |
318 | if (match(ext, vexok_expr)) | |
319 | flags = add_flags(flags, "INAT_VEXOK") | |
320 | ||
eb13296c MH |
321 | # check prefixes |
322 | if (match(ext, prefix_expr)) { | |
323 | if (!prefix_num[opcode]) | |
324 | semantic_error("Unknown prefix: " opcode) | |
325 | flags = add_flags(flags, "INAT_MAKE_PREFIX(" prefix_num[opcode] ")") | |
326 | } | |
327 | if (length(flags) == 0) | |
328 | continue | |
329 | # check if last prefix | |
330 | if (match(ext, lprefix1_expr)) { | |
331 | lptable1[idx] = add_flags(lptable1[idx],flags) | |
332 | variant = "INAT_VARIANT" | |
333 | } else if (match(ext, lprefix2_expr)) { | |
334 | lptable2[idx] = add_flags(lptable2[idx],flags) | |
335 | variant = "INAT_VARIANT" | |
336 | } else if (match(ext, lprefix3_expr)) { | |
337 | lptable3[idx] = add_flags(lptable3[idx],flags) | |
338 | variant = "INAT_VARIANT" | |
339 | } else { | |
340 | table[idx] = add_flags(table[idx],flags) | |
341 | } | |
342 | } | |
343 | if (variant) | |
344 | table[idx] = add_flags(table[idx],variant) | |
345 | } | |
346 | ||
347 | END { | |
69d991f3 MH |
348 | if (awkchecked != "") |
349 | exit 1 | |
eb13296c MH |
350 | # print escape opcode map's array |
351 | print "/* Escape opcode map array */" | |
352 | print "const insn_attr_t const *inat_escape_tables[INAT_ESC_MAX + 1]" \ | |
04d46c1b | 353 | "[INAT_LSTPFX_MAX + 1] = {" |
eb13296c MH |
354 | for (i = 0; i < geid; i++) |
355 | for (j = 0; j < max_lprefix; j++) | |
356 | if (etable[i,j]) | |
357 | print " ["i"]["j"] = "etable[i,j]"," | |
358 | print "};\n" | |
359 | # print group opcode map's array | |
360 | print "/* Group opcode map array */" | |
361 | print "const insn_attr_t const *inat_group_tables[INAT_GRP_MAX + 1]"\ | |
04d46c1b | 362 | "[INAT_LSTPFX_MAX + 1] = {" |
eb13296c MH |
363 | for (i = 0; i < ggid; i++) |
364 | for (j = 0; j < max_lprefix; j++) | |
365 | if (gtable[i,j]) | |
366 | print " ["i"]["j"] = "gtable[i,j]"," | |
e0e492e9 MH |
367 | print "};\n" |
368 | # print AVX opcode map's array | |
369 | print "/* AVX opcode map array */" | |
370 | print "const insn_attr_t const *inat_avx_tables[X86_VEX_M_MAX + 1]"\ | |
371 | "[INAT_LSTPFX_MAX + 1] = {" | |
372 | for (i = 0; i < gaid; i++) | |
373 | for (j = 0; j < max_lprefix; j++) | |
374 | if (atable[i,j]) | |
375 | print " ["i"]["j"] = "atable[i,j]"," | |
eb13296c MH |
376 | print "};" |
377 | } | |
e0e492e9 | 378 |