2 # gen-insn-attr-x86.awk: Instruction attribute table generator
3 # Written by Masami Hiramatsu <mhiramat@redhat.com>
5 # Usage: awk -f gen-insn-attr-x86.awk x86-opcode-map.txt > inat-tables.c
7 # Awk implementation sanity check
8 function check_awk_implement() {
9 if (sprintf("%x", 0) != "0")
10 return "Your awk has a printf-format problem."
15 function clear_vars() {
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"
35 # Setup generating tables
36 print "/* x86 opcode map generated from x86-opcode-map.txt */"
37 print "/* Do not change this code. */\n"
45 opnd_expr = "^[A-Za-z/]"
48 group_expr = "^Grp[0-9A-Za-z]+"
50 imm_expr = "^[IJAOL][a-z]"
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 imm_flag["Lx"] = "INAT_MAKE_IMM(INAT_IMM_BYTE)"
64 modrm_expr = "^([CDEGMNPQRSUVW/][a-z]+|NTA|T[012])"
65 force64_expr = "\\([df]64\\)"
66 rex_expr = "^REX(\\.[XRWB]+)*"
67 fpu_expr = "^ESC" # TODO
69 lprefix1_expr = "\\(66\\)"
70 lprefix2_expr = "\\(F3\\)"
71 lprefix3_expr = "\\(F2\\)"
74 # All opcodes starting with lower-case 'v' or with (v1) superscript
76 vexok_opcode_expr = "^v.*"
77 vexok_expr = "\\(v1\\)"
78 # All opcodes with (v) superscript supports *only* VEX prefix
79 vexonly_expr = "\\(v\\)"
81 prefix_expr = "\\(Prefix\\)"
82 prefix_num["Operand-Size"] = "INAT_PFX_OPNDSZ"
83 prefix_num["REPNE"] = "INAT_PFX_REPNE"
84 prefix_num["REP/REPE"] = "INAT_PFX_REPE"
85 prefix_num["LOCK"] = "INAT_PFX_LOCK"
86 prefix_num["SEG=CS"] = "INAT_PFX_CS"
87 prefix_num["SEG=DS"] = "INAT_PFX_DS"
88 prefix_num["SEG=ES"] = "INAT_PFX_ES"
89 prefix_num["SEG=FS"] = "INAT_PFX_FS"
90 prefix_num["SEG=GS"] = "INAT_PFX_GS"
91 prefix_num["SEG=SS"] = "INAT_PFX_SS"
92 prefix_num["Address-Size"] = "INAT_PFX_ADDRSZ"
93 prefix_num["VEX+1byte"] = "INAT_PFX_VEX2"
94 prefix_num["VEX+2byte"] = "INAT_PFX_VEX3"
99 function semantic_error(msg) {
100 print "Semantic error at " NR ": " msg > "/dev/stderr"
104 function debug(msg) {
108 function array_size(arr, i,c) {
118 semantic_error("Hit Table: before EndTable:.");
123 # escape opcode table
125 for (i = 2; i <= NF; i++)
128 tname = sprintf("inat_escape_table_%d", eid)
134 # AVX/escape opcode table
138 if (tname == "") # AVX only opcode table
139 tname = sprintf("inat_avx_table_%d", $2)
141 if (aid == -1 && eid == -1) # primary opcode table
142 tname = "inat_primary_table"
148 semantic_error("No group: " $2 )
150 tname = "inat_group_table_" gid
153 function print_table(tbl,name,fmt,n)
155 print "const insn_attr_t " name " = {"
156 for (i = 0; i < n; i++) {
159 print " [" id "] = " tbl[id] ","
167 if (array_size(table) != 0) {
168 print_table(table, tname "[INAT_GROUP_TABLE_SIZE]",
170 gtable[gid,0] = tname
172 if (array_size(lptable1) != 0) {
173 print_table(lptable1, tname "_1[INAT_GROUP_TABLE_SIZE]",
175 gtable[gid,1] = tname "_1"
177 if (array_size(lptable2) != 0) {
178 print_table(lptable2, tname "_2[INAT_GROUP_TABLE_SIZE]",
180 gtable[gid,2] = tname "_2"
182 if (array_size(lptable3) != 0) {
183 print_table(lptable3, tname "_3[INAT_GROUP_TABLE_SIZE]",
185 gtable[gid,3] = tname "_3"
188 # print primary/escaped tables
189 if (array_size(table) != 0) {
190 print_table(table, tname "[INAT_OPCODE_TABLE_SIZE]",
192 etable[eid,0] = tname
194 atable[aid,0] = tname
196 if (array_size(lptable1) != 0) {
197 print_table(lptable1,tname "_1[INAT_OPCODE_TABLE_SIZE]",
199 etable[eid,1] = tname "_1"
201 atable[aid,1] = tname "_1"
203 if (array_size(lptable2) != 0) {
204 print_table(lptable2,tname "_2[INAT_OPCODE_TABLE_SIZE]",
206 etable[eid,2] = tname "_2"
208 atable[aid,2] = tname "_2"
210 if (array_size(lptable3) != 0) {
211 print_table(lptable3,tname "_3[INAT_OPCODE_TABLE_SIZE]",
213 etable[eid,3] = tname "_3"
215 atable[aid,3] = tname "_3"
222 function add_flags(old,new) {
231 # convert operands to flags.
232 function convert_operands(count,opnd, i,j,imm,mod)
236 for (j = 1; j <= count; j++) {
238 if (match(i, imm_expr) == 1) {
240 semantic_error("Unknown imm opnd: " i)
243 semantic_error("Second IMM error")
244 imm = add_flags(imm, "INAT_SCNDIMM")
247 } else if (match(i, modrm_expr))
250 return add_flags(imm, mod)
257 idx = "0x" substr($1, 1, index($1,":") - 1)
259 semantic_error("Redefine " idx " in " tname)
261 # check if escaped opcode
262 if ("escape" == $2) {
264 semantic_error("No escaped name")
266 for (i = 4; i <= NF; i++)
269 semantic_error("Redefine escape (" ref ")")
272 table[idx] = "INAT_MAKE_ESCAPE(" escape[ref] ")"
286 if (match($i, opnd_expr)) {
288 count = split($(i++), opnds, ",")
289 flags = convert_operands(count, opnds)
291 if (match($i, ext_expr))
293 if (match($i, sep_expr))
296 semantic_error($i " is not a separator")
298 # check if group opcode
299 if (match(opcode, group_expr)) {
300 if (!(opcode in group)) {
304 flags = add_flags(flags, "INAT_MAKE_GROUP(" group[opcode] ")")
306 # check force(or default) 64bit
307 if (match(ext, force64_expr))
308 flags = add_flags(flags, "INAT_FORCE64")
311 if (match(opcode, rex_expr))
312 flags = add_flags(flags, "INAT_MAKE_PREFIX(INAT_PFX_REX)")
314 # check coprocessor escape : TODO
315 if (match(opcode, fpu_expr))
316 flags = add_flags(flags, "INAT_MODRM")
319 if (match(ext, vexonly_expr))
320 flags = add_flags(flags, "INAT_VEXOK | INAT_VEXONLY")
321 else if (match(ext, vexok_expr) || match(opcode, vexok_opcode_expr))
322 flags = add_flags(flags, "INAT_VEXOK")
325 if (match(ext, prefix_expr)) {
326 if (!prefix_num[opcode])
327 semantic_error("Unknown prefix: " opcode)
328 flags = add_flags(flags, "INAT_MAKE_PREFIX(" prefix_num[opcode] ")")
330 if (length(flags) == 0)
332 # check if last prefix
333 if (match(ext, lprefix1_expr)) {
334 lptable1[idx] = add_flags(lptable1[idx],flags)
335 variant = "INAT_VARIANT"
336 } else if (match(ext, lprefix2_expr)) {
337 lptable2[idx] = add_flags(lptable2[idx],flags)
338 variant = "INAT_VARIANT"
339 } else if (match(ext, lprefix3_expr)) {
340 lptable3[idx] = add_flags(lptable3[idx],flags)
341 variant = "INAT_VARIANT"
343 table[idx] = add_flags(table[idx],flags)
347 table[idx] = add_flags(table[idx],variant)
351 if (awkchecked != "")
353 # print escape opcode map's array
354 print "/* Escape opcode map array */"
355 print "const insn_attr_t const *inat_escape_tables[INAT_ESC_MAX + 1]" \
356 "[INAT_LSTPFX_MAX + 1] = {"
357 for (i = 0; i < geid; i++)
358 for (j = 0; j < max_lprefix; j++)
360 print " ["i"]["j"] = "etable[i,j]","
362 # print group opcode map's array
363 print "/* Group opcode map array */"
364 print "const insn_attr_t const *inat_group_tables[INAT_GRP_MAX + 1]"\
365 "[INAT_LSTPFX_MAX + 1] = {"
366 for (i = 0; i < ggid; i++)
367 for (j = 0; j < max_lprefix; j++)
369 print " ["i"]["j"] = "gtable[i,j]","
371 # print AVX opcode map's array
372 print "/* AVX opcode map array */"
373 print "const insn_attr_t const *inat_avx_tables[X86_VEX_M_MAX + 1]"\
374 "[INAT_LSTPFX_MAX + 1] = {"
375 for (i = 0; i < gaid; i++)
376 for (j = 0; j < max_lprefix; j++)
378 print " ["i"]["j"] = "atable[i,j]","