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 | ||
7 | BEGIN { | |
8 | print "/* x86 opcode map generated from x86-opcode-map.txt */" | |
9 | print "/* Do not change this code. */" | |
10 | ggid = 1 | |
11 | geid = 1 | |
12 | ||
13 | opnd_expr = "^[[:alpha:]]" | |
14 | ext_expr = "^\\(" | |
15 | sep_expr = "^\\|$" | |
16 | group_expr = "^Grp[[:alnum:]]+" | |
17 | ||
18 | imm_expr = "^[IJAO][[:lower:]]" | |
19 | imm_flag["Ib"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)" | |
20 | imm_flag["Jb"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)" | |
21 | imm_flag["Iw"] = "INAT_MAKE_IMM(INAT_IMM_WORD)" | |
22 | imm_flag["Id"] = "INAT_MAKE_IMM(INAT_IMM_DWORD)" | |
23 | imm_flag["Iq"] = "INAT_MAKE_IMM(INAT_IMM_QWORD)" | |
24 | imm_flag["Ap"] = "INAT_MAKE_IMM(INAT_IMM_PTR)" | |
25 | imm_flag["Iz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)" | |
26 | imm_flag["Jz"] = "INAT_MAKE_IMM(INAT_IMM_VWORD32)" | |
27 | imm_flag["Iv"] = "INAT_MAKE_IMM(INAT_IMM_VWORD)" | |
28 | imm_flag["Ob"] = "INAT_MOFFSET" | |
29 | imm_flag["Ov"] = "INAT_MOFFSET" | |
30 | ||
31 | modrm_expr = "^([CDEGMNPQRSUVW][[:lower:]]+|NTA|T[012])" | |
32 | force64_expr = "\\([df]64\\)" | |
33 | rex_expr = "^REX(\\.[XRWB]+)*" | |
34 | fpu_expr = "^ESC" # TODO | |
35 | ||
36 | lprefix1_expr = "\\(66\\)" | |
37 | delete lptable1 | |
38 | lprefix2_expr = "\\(F2\\)" | |
39 | delete lptable2 | |
40 | lprefix3_expr = "\\(F3\\)" | |
41 | delete lptable3 | |
42 | max_lprefix = 4 | |
43 | ||
44 | prefix_expr = "\\(Prefix\\)" | |
45 | prefix_num["Operand-Size"] = "INAT_PFX_OPNDSZ" | |
46 | prefix_num["REPNE"] = "INAT_PFX_REPNE" | |
47 | prefix_num["REP/REPE"] = "INAT_PFX_REPE" | |
48 | prefix_num["LOCK"] = "INAT_PFX_LOCK" | |
49 | prefix_num["SEG=CS"] = "INAT_PFX_CS" | |
50 | prefix_num["SEG=DS"] = "INAT_PFX_DS" | |
51 | prefix_num["SEG=ES"] = "INAT_PFX_ES" | |
52 | prefix_num["SEG=FS"] = "INAT_PFX_FS" | |
53 | prefix_num["SEG=GS"] = "INAT_PFX_GS" | |
54 | prefix_num["SEG=SS"] = "INAT_PFX_SS" | |
55 | prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ" | |
56 | ||
57 | delete table | |
58 | delete etable | |
59 | delete gtable | |
60 | eid = -1 | |
61 | gid = -1 | |
62 | } | |
63 | ||
64 | function semantic_error(msg) { | |
65 | print "Semantic error at " NR ": " msg > "/dev/stderr" | |
66 | exit 1 | |
67 | } | |
68 | ||
69 | function debug(msg) { | |
70 | print "DEBUG: " msg | |
71 | } | |
72 | ||
73 | function array_size(arr, i,c) { | |
74 | c = 0 | |
75 | for (i in arr) | |
76 | c++ | |
77 | return c | |
78 | } | |
79 | ||
80 | /^Table:/ { | |
81 | print "/* " $0 " */" | |
82 | } | |
83 | ||
84 | /^Referrer:/ { | |
85 | if (NF == 1) { | |
86 | # primary opcode table | |
87 | tname = "inat_primary_table" | |
88 | eid = -1 | |
89 | } else { | |
90 | # escape opcode table | |
91 | ref = "" | |
92 | for (i = 2; i <= NF; i++) | |
93 | ref = ref $i | |
94 | eid = escape[ref] | |
95 | tname = sprintf("inat_escape_table_%d", eid) | |
96 | } | |
97 | } | |
98 | ||
99 | /^GrpTable:/ { | |
100 | print "/* " $0 " */" | |
101 | if (!($2 in group)) | |
102 | semantic_error("No group: " $2 ) | |
103 | gid = group[$2] | |
104 | tname = "inat_group_table_" gid | |
105 | } | |
106 | ||
107 | function print_table(tbl,name,fmt,n) | |
108 | { | |
109 | print "const insn_attr_t " name " = {" | |
110 | for (i = 0; i < n; i++) { | |
111 | id = sprintf(fmt, i) | |
112 | if (tbl[id]) | |
113 | print " [" id "] = " tbl[id] "," | |
114 | } | |
115 | print "};" | |
116 | } | |
117 | ||
118 | /^EndTable/ { | |
119 | if (gid != -1) { | |
120 | # print group tables | |
121 | if (array_size(table) != 0) { | |
122 | print_table(table, tname "[INAT_GROUP_TABLE_SIZE]", | |
123 | "0x%x", 8) | |
124 | gtable[gid,0] = tname | |
125 | } | |
126 | if (array_size(lptable1) != 0) { | |
127 | print_table(lptable1, tname "_1[INAT_GROUP_TABLE_SIZE]", | |
128 | "0x%x", 8) | |
129 | gtable[gid,1] = tname "_1" | |
130 | } | |
131 | if (array_size(lptable2) != 0) { | |
132 | print_table(lptable2, tname "_2[INAT_GROUP_TABLE_SIZE]", | |
133 | "0x%x", 8) | |
134 | gtable[gid,2] = tname "_2" | |
135 | } | |
136 | if (array_size(lptable3) != 0) { | |
137 | print_table(lptable3, tname "_3[INAT_GROUP_TABLE_SIZE]", | |
138 | "0x%x", 8) | |
139 | gtable[gid,3] = tname "_3" | |
140 | } | |
141 | } else { | |
142 | # print primary/escaped tables | |
143 | if (array_size(table) != 0) { | |
144 | print_table(table, tname "[INAT_OPCODE_TABLE_SIZE]", | |
145 | "0x%02x", 256) | |
146 | etable[eid,0] = tname | |
147 | } | |
148 | if (array_size(lptable1) != 0) { | |
149 | print_table(lptable1,tname "_1[INAT_OPCODE_TABLE_SIZE]", | |
150 | "0x%02x", 256) | |
151 | etable[eid,1] = tname "_1" | |
152 | } | |
153 | if (array_size(lptable2) != 0) { | |
154 | print_table(lptable2,tname "_2[INAT_OPCODE_TABLE_SIZE]", | |
155 | "0x%02x", 256) | |
156 | etable[eid,2] = tname "_2" | |
157 | } | |
158 | if (array_size(lptable3) != 0) { | |
159 | print_table(lptable3,tname "_3[INAT_OPCODE_TABLE_SIZE]", | |
160 | "0x%02x", 256) | |
161 | etable[eid,3] = tname "_3" | |
162 | } | |
163 | } | |
164 | print "" | |
165 | delete table | |
166 | delete lptable1 | |
167 | delete lptable2 | |
168 | delete lptable3 | |
169 | gid = -1 | |
170 | eid = -1 | |
171 | } | |
172 | ||
173 | function add_flags(old,new) { | |
174 | if (old && new) | |
175 | return old " | " new | |
176 | else if (old) | |
177 | return old | |
178 | else | |
179 | return new | |
180 | } | |
181 | ||
182 | # convert operands to flags. | |
183 | function convert_operands(opnd, i,imm,mod) | |
184 | { | |
185 | imm = null | |
186 | mod = null | |
187 | for (i in opnd) { | |
188 | i = opnd[i] | |
189 | if (match(i, imm_expr) == 1) { | |
190 | if (!imm_flag[i]) | |
191 | semantic_error("Unknown imm opnd: " i) | |
192 | if (imm) { | |
193 | if (i != "Ib") | |
194 | semantic_error("Second IMM error") | |
195 | imm = add_flags(imm, "INAT_SCNDIMM") | |
196 | } else | |
197 | imm = imm_flag[i] | |
198 | } else if (match(i, modrm_expr)) | |
199 | mod = "INAT_MODRM" | |
200 | } | |
201 | return add_flags(imm, mod) | |
202 | } | |
203 | ||
204 | /^[0-9a-f]+\:/ { | |
205 | if (NR == 1) | |
206 | next | |
207 | # get index | |
208 | idx = "0x" substr($1, 1, index($1,":") - 1) | |
209 | if (idx in table) | |
210 | semantic_error("Redefine " idx " in " tname) | |
211 | ||
212 | # check if escaped opcode | |
213 | if ("escape" == $2) { | |
214 | if ($3 != "#") | |
215 | semantic_error("No escaped name") | |
216 | ref = "" | |
217 | for (i = 4; i <= NF; i++) | |
218 | ref = ref $i | |
219 | if (ref in escape) | |
220 | semantic_error("Redefine escape (" ref ")") | |
221 | escape[ref] = geid | |
222 | geid++ | |
223 | table[idx] = "INAT_MAKE_ESCAPE(" escape[ref] ")" | |
224 | next | |
225 | } | |
226 | ||
227 | variant = null | |
228 | # converts | |
229 | i = 2 | |
230 | while (i <= NF) { | |
231 | opcode = $(i++) | |
232 | delete opnds | |
233 | ext = null | |
234 | flags = null | |
235 | opnd = null | |
236 | # parse one opcode | |
237 | if (match($i, opnd_expr)) { | |
238 | opnd = $i | |
239 | split($(i++), opnds, ",") | |
240 | flags = convert_operands(opnds) | |
241 | } | |
242 | if (match($i, ext_expr)) | |
243 | ext = $(i++) | |
244 | if (match($i, sep_expr)) | |
245 | i++ | |
246 | else if (i < NF) | |
247 | semantic_error($i " is not a separator") | |
248 | ||
249 | # check if group opcode | |
250 | if (match(opcode, group_expr)) { | |
251 | if (!(opcode in group)) { | |
252 | group[opcode] = ggid | |
253 | ggid++ | |
254 | } | |
255 | flags = add_flags(flags, "INAT_MAKE_GROUP(" group[opcode] ")") | |
256 | } | |
257 | # check force(or default) 64bit | |
258 | if (match(ext, force64_expr)) | |
259 | flags = add_flags(flags, "INAT_FORCE64") | |
260 | ||
261 | # check REX prefix | |
262 | if (match(opcode, rex_expr)) | |
263 | flags = add_flags(flags, "INAT_REXPFX") | |
264 | ||
265 | # check coprocessor escape : TODO | |
266 | if (match(opcode, fpu_expr)) | |
267 | flags = add_flags(flags, "INAT_MODRM") | |
268 | ||
269 | # check prefixes | |
270 | if (match(ext, prefix_expr)) { | |
271 | if (!prefix_num[opcode]) | |
272 | semantic_error("Unknown prefix: " opcode) | |
273 | flags = add_flags(flags, "INAT_MAKE_PREFIX(" prefix_num[opcode] ")") | |
274 | } | |
275 | if (length(flags) == 0) | |
276 | continue | |
277 | # check if last prefix | |
278 | if (match(ext, lprefix1_expr)) { | |
279 | lptable1[idx] = add_flags(lptable1[idx],flags) | |
280 | variant = "INAT_VARIANT" | |
281 | } else if (match(ext, lprefix2_expr)) { | |
282 | lptable2[idx] = add_flags(lptable2[idx],flags) | |
283 | variant = "INAT_VARIANT" | |
284 | } else if (match(ext, lprefix3_expr)) { | |
285 | lptable3[idx] = add_flags(lptable3[idx],flags) | |
286 | variant = "INAT_VARIANT" | |
287 | } else { | |
288 | table[idx] = add_flags(table[idx],flags) | |
289 | } | |
290 | } | |
291 | if (variant) | |
292 | table[idx] = add_flags(table[idx],variant) | |
293 | } | |
294 | ||
295 | END { | |
296 | # print escape opcode map's array | |
297 | print "/* Escape opcode map array */" | |
298 | print "const insn_attr_t const *inat_escape_tables[INAT_ESC_MAX + 1]" \ | |
299 | "[INAT_LPREFIX_MAX + 1] = {" | |
300 | for (i = 0; i < geid; i++) | |
301 | for (j = 0; j < max_lprefix; j++) | |
302 | if (etable[i,j]) | |
303 | print " ["i"]["j"] = "etable[i,j]"," | |
304 | print "};\n" | |
305 | # print group opcode map's array | |
306 | print "/* Group opcode map array */" | |
307 | print "const insn_attr_t const *inat_group_tables[INAT_GRP_MAX + 1]"\ | |
308 | "[INAT_LPREFIX_MAX + 1] = {" | |
309 | for (i = 0; i < ggid; i++) | |
310 | for (j = 0; j < max_lprefix; j++) | |
311 | if (gtable[i,j]) | |
312 | print " ["i"]["j"] = "gtable[i,j]"," | |
313 | print "};" | |
314 | } |