1 /*
2 * Copyright (c) 2020, 2025, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24
25 #ifndef CPU_X86_C2_MACROASSEMBLER_X86_HPP
26 #define CPU_X86_C2_MACROASSEMBLER_X86_HPP
27
28 // C2_MacroAssembler contains high-level macros for C2
29
30 public:
31 // C2 compiled method's prolog code.
32 void verified_entry(Compile* C, int sp_inc = 0);
33
34 void entry_barrier();
35 Assembler::AvxVectorLen vector_length_encoding(int vlen_in_bytes);
36
37 // Code used by cmpFastLock and cmpFastUnlock mach instructions in .ad file.
38 // See full description in c2_MacroAssembler_x86.cpp.
39 void fast_lock_lightweight(Register obj, Register box, Register rax_reg,
40 Register t, Register thread);
41 void fast_unlock_lightweight(Register obj, Register reg_rax, Register t, Register thread);
42
43 void verify_int_in_range(uint idx, const TypeInt* t, Register val);
44 void verify_long_in_range(uint idx, const TypeLong* t, Register val, Register tmp);
45
46 // Generic instructions support for use in .ad files C2 code generation
47 void vabsnegd(int opcode, XMMRegister dst, XMMRegister src);
48 void vabsnegd(int opcode, XMMRegister dst, XMMRegister src, int vector_len);
49 void vabsnegf(int opcode, XMMRegister dst, XMMRegister src);
50 void vabsnegf(int opcode, XMMRegister dst, XMMRegister src, int vector_len);
51
52 void pminmax(int opcode, BasicType elem_bt, XMMRegister dst, XMMRegister src,
53 XMMRegister tmp = xnoreg);
54 void vpminmax(int opcode, BasicType elem_bt,
55 XMMRegister dst, XMMRegister src1, XMMRegister src2,
56 int vlen_enc);
57
58 void vpuminmax(int opcode, BasicType elem_bt,
59 XMMRegister dst, XMMRegister src1, XMMRegister src2,
60 int vlen_enc);
61
62 void vpuminmax(int opcode, BasicType elem_bt,
63 XMMRegister dst, XMMRegister src1, Address src2,
64 int vlen_enc);
65
66 void vminmax_fp(int opcode, BasicType elem_bt,
67 XMMRegister dst, XMMRegister a, XMMRegister b,
68 XMMRegister tmp, XMMRegister atmp, XMMRegister btmp,
69 int vlen_enc);
70
71 void vminmax_fp(int opc, BasicType elem_bt, XMMRegister dst, KRegister mask,
72 XMMRegister src1, XMMRegister src2, int vlen_enc);
73
74 void vpuminmaxq(int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2, XMMRegister xtmp1, XMMRegister xtmp2, int vlen_enc);
75
76 void evminmax_fp(int opcode, BasicType elem_bt,
77 XMMRegister dst, XMMRegister a, XMMRegister b,
78 KRegister ktmp, XMMRegister atmp, XMMRegister btmp,
79 int vlen_enc);
80
81 void signum_fp(int opcode, XMMRegister dst, XMMRegister zero, XMMRegister one);
82
83 void vector_compress_expand(int opcode, XMMRegister dst, XMMRegister src, KRegister mask,
84 bool merge, BasicType bt, int vec_enc);
85
86 void vector_mask_compress(KRegister dst, KRegister src, Register rtmp1, Register rtmp2, int mask_len);
87
88 void vextendbw(bool sign, XMMRegister dst, XMMRegister src, int vector_len);
89 void vextendbw(bool sign, XMMRegister dst, XMMRegister src);
90 void vextendbd(bool sign, XMMRegister dst, XMMRegister src, int vector_len);
91 void vextendwd(bool sign, XMMRegister dst, XMMRegister src, int vector_len);
92
93 void vshiftd(int opcode, XMMRegister dst, XMMRegister shift);
94 void vshiftd_imm(int opcode, XMMRegister dst, int shift);
95 void vshiftd(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
96 void vshiftd_imm(int opcode, XMMRegister dst, XMMRegister nds, int shift, int vector_len);
97 void vshiftw(int opcode, XMMRegister dst, XMMRegister shift);
98 void vshiftw(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
99 void vshiftq(int opcode, XMMRegister dst, XMMRegister shift);
100 void vshiftq_imm(int opcode, XMMRegister dst, int shift);
101 void vshiftq(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
102 void vshiftq_imm(int opcode, XMMRegister dst, XMMRegister nds, int shift, int vector_len);
103
104 void vprotate_imm(int opcode, BasicType etype, XMMRegister dst, XMMRegister src, int shift, int vector_len);
105 void vprotate_var(int opcode, BasicType etype, XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len);
106
107 void varshiftd(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
108 void varshiftw(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc);
109 void varshiftq(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vlen_enc, XMMRegister vtmp = xnoreg);
110 void varshiftbw(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len, XMMRegister vtmp);
111 void evarshiftb(int opcode, XMMRegister dst, XMMRegister src, XMMRegister shift, int vector_len, XMMRegister vtmp);
112
113 void insert(BasicType typ, XMMRegister dst, Register val, int idx);
114 void vinsert(BasicType typ, XMMRegister dst, XMMRegister src, Register val, int idx);
115 void vgather(BasicType typ, XMMRegister dst, Register base, XMMRegister idx, XMMRegister mask, int vector_len);
116 void evgather(BasicType typ, XMMRegister dst, KRegister mask, Register base, XMMRegister idx, int vector_len);
117 void evscatter(BasicType typ, Register base, XMMRegister idx, KRegister mask, XMMRegister src, int vector_len);
118
119 void evmovdqu(BasicType type, KRegister kmask, XMMRegister dst, Address src, bool merge, int vector_len);
120 void evmovdqu(BasicType type, KRegister kmask, Address dst, XMMRegister src, bool merge, int vector_len);
121 void evmovdqu(BasicType type, KRegister kmask, XMMRegister dst, XMMRegister src, bool merge, int vector_len);
122
123 // extract
124 void extract(BasicType typ, Register dst, XMMRegister src, int idx);
125 XMMRegister get_lane(BasicType typ, XMMRegister dst, XMMRegister src, int elemindex);
126 void get_elem(BasicType typ, Register dst, XMMRegister src, int elemindex);
127 void get_elem(BasicType typ, XMMRegister dst, XMMRegister src, int elemindex, XMMRegister vtmp = xnoreg);
128 void movsxl(BasicType typ, Register dst);
129
130 // vector test
131 void vectortest(BasicType bt, XMMRegister src1, XMMRegister src2, XMMRegister vtmp, int vlen_in_bytes);
132
133 // Covert B2X
134 void vconvert_b2x(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, int vlen_enc);
135 void vpbroadcast(BasicType elem_bt, XMMRegister dst, Register src, int vlen_enc);
136
137 // blend
138 void evpcmp(BasicType typ, KRegister kdmask, KRegister ksmask, XMMRegister src1, XMMRegister src2, int comparison, int vector_len);
139 void evpcmp(BasicType typ, KRegister kdmask, KRegister ksmask, XMMRegister src1, AddressLiteral src2, int comparison, int vector_len, Register rscratch = noreg);
140 void evpblend(BasicType typ, XMMRegister dst, KRegister kmask, XMMRegister src1, XMMRegister src2, bool merge, int vector_len);
141
142 void load_vector(BasicType bt, XMMRegister dst, Address src, int vlen_in_bytes);
143 void load_vector(BasicType bt, XMMRegister dst, AddressLiteral src, int vlen_in_bytes, Register rscratch = noreg);
144
145 void load_vector_mask(XMMRegister dst, XMMRegister src, int vlen_in_bytes, BasicType elem_bt, bool is_legacy);
146 void load_vector_mask(KRegister dst, XMMRegister src, XMMRegister xtmp, bool novlbwdq, int vlen_enc);
147
148 void load_constant_vector(BasicType bt, XMMRegister dst, InternalAddress src, int vlen);
149 void load_iota_indices(XMMRegister dst, int vlen_in_bytes, BasicType bt);
150
151 // Reductions for vectors of bytes, shorts, ints, longs, floats, and doubles.
152
153 // dst = src1 reduce(op, src2) using vtmp as temps
154 void reduceI(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
155 void reduceL(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
156 void genmask(KRegister dst, Register len, Register temp);
157
158 // dst = reduce(op, src2) using vtmp as temps
159 void reduce_fp(int opcode, int vlen,
160 XMMRegister dst, XMMRegister src,
161 XMMRegister vtmp1, XMMRegister vtmp2 = xnoreg);
162 void unordered_reduce_fp(int opcode, int vlen,
163 XMMRegister dst, XMMRegister src,
164 XMMRegister vtmp1 = xnoreg, XMMRegister vtmp2 = xnoreg);
165 void reduceB(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
166 void mulreduceB(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
167 void reduceS(int opcode, int vlen, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
168 void reduceFloatMinMax(int opcode, int vlen, bool is_dst_valid,
169 XMMRegister dst, XMMRegister src,
170 XMMRegister tmp, XMMRegister atmp, XMMRegister btmp, XMMRegister xmm_0, XMMRegister xmm_1 = xnoreg);
171 void reduceDoubleMinMax(int opcode, int vlen, bool is_dst_valid,
172 XMMRegister dst, XMMRegister src,
173 XMMRegister tmp, XMMRegister atmp, XMMRegister btmp, XMMRegister xmm_0, XMMRegister xmm_1 = xnoreg);
174 private:
175 void reduceF(int opcode, int vlen, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
176 void reduceD(int opcode, int vlen, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
177 void unorderedReduceF(int opcode, int vlen, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
178 void unorderedReduceD(int opcode, int vlen, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
179
180 // Int Reduction
181 void reduce2I (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
182 void reduce4I (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
183 void reduce8I (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
184 void reduce16I(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
185
186 // Byte Reduction
187 void reduce8B (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
188 void reduce16B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
189 void reduce32B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
190 void reduce64B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
191 void mulreduce8B (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
192 void mulreduce16B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
193 void mulreduce32B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
194 void mulreduce64B(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
195
196 // Short Reduction
197 void reduce4S (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
198 void reduce8S (int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
199 void reduce16S(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
200 void reduce32S(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
201
202 // Long Reduction
203 void reduce2L(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
204 void reduce4L(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
205 void reduce8L(int opcode, Register dst, Register src1, XMMRegister src2, XMMRegister vtmp1, XMMRegister vtmp2);
206
207 // Float Reduction
208 void reduce2F (int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
209 void reduce4F (int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
210 void reduce8F (int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
211 void reduce16F(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
212
213 // Unordered Float Reduction
214 void unorderedReduce2F(int opcode, XMMRegister dst, XMMRegister src);
215 void unorderedReduce4F(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
216 void unorderedReduce8F(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
217 void unorderedReduce16F(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
218
219 // Double Reduction
220 void reduce2D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
221 void reduce4D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
222 void reduce8D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
223
224 // Unordered Double Reduction
225 void unorderedReduce2D(int opcode, XMMRegister dst, XMMRegister src);
226 void unorderedReduce4D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp);
227 void unorderedReduce8D(int opcode, XMMRegister dst, XMMRegister src, XMMRegister vtmp1, XMMRegister vtmp2);
228
229 // Base reduction instruction
230 void reduce_operation_128(BasicType typ, int opcode, XMMRegister dst, XMMRegister src);
231 void reduce_operation_256(BasicType typ, int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2);
232 void unordered_reduce_operation_128(BasicType typ, int opcode, XMMRegister dst, XMMRegister src);
233 void unordered_reduce_operation_256(BasicType typ, int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2);
234
235 public:
236 void vector_mask_operation_helper(int opc, Register dst, Register tmp, int masklen);
237
238 void vector_mask_operation(int opc, Register dst, KRegister mask, Register tmp, int masklen, int masksize, int vec_enc);
239
240 void vector_mask_operation(int opc, Register dst, XMMRegister mask, XMMRegister xtmp,
241 Register tmp, int masklen, BasicType bt, int vec_enc);
242 void vector_long_to_maskvec(XMMRegister dst, Register src, Register rtmp1,
243 Register rtmp2, XMMRegister xtmp, int mask_len, int vec_enc);
244
245 void vector_maskall_operation(KRegister dst, Register src, int mask_len);
246
247 void string_indexof_char(Register str1, Register cnt1, Register ch, Register result,
248 XMMRegister vec1, XMMRegister vec2, XMMRegister vec3, Register tmp);
249
250 void stringL_indexof_char(Register str1, Register cnt1, Register ch, Register result,
251 XMMRegister vec1, XMMRegister vec2, XMMRegister vec3, Register tmp);
252
253 // IndexOf strings.
254 // Small strings are loaded through stack if they cross page boundary.
255 void string_indexof(Register str1, Register str2,
256 Register cnt1, Register cnt2,
257 int int_cnt2, Register result,
258 XMMRegister vec, Register tmp,
259 int ae);
260
261 // IndexOf for constant substrings with size >= 8 elements
262 // which don't need to be loaded through stack.
263 void string_indexofC8(Register str1, Register str2,
264 Register cnt1, Register cnt2,
265 int int_cnt2, Register result,
266 XMMRegister vec, Register tmp,
267 int ae);
268
269 // Smallest code: we don't need to load through stack,
270 // check string tail.
271
272 // helper function for string_compare
273 void load_next_elements(Register elem1, Register elem2, Register str1, Register str2,
274 Address::ScaleFactor scale, Address::ScaleFactor scale1,
275 Address::ScaleFactor scale2, Register index, int ae);
276 // Compare strings.
277 void string_compare(Register str1, Register str2,
278 Register cnt1, Register cnt2, Register result,
279 XMMRegister vec1, int ae, KRegister mask = knoreg);
280
281 // Search for Non-ASCII character (Negative byte value) in a byte array,
282 // return index of the first such character, otherwise len.
283 void count_positives(Register ary1, Register len,
284 Register result, Register tmp1,
285 XMMRegister vec1, XMMRegister vec2, KRegister mask1 = knoreg, KRegister mask2 = knoreg);
286
287 // Compare char[] or byte[] arrays.
288 void arrays_equals(bool is_array_equ, Register ary1, Register ary2, Register limit,
289 Register result, Register chr, XMMRegister vec1, XMMRegister vec2,
290 bool is_char, KRegister mask = knoreg, bool expand_ary2 = false);
291
292 void arrays_hashcode(Register str1, Register cnt1, Register result,
293 Register tmp1, Register tmp2, Register tmp3, XMMRegister vnext,
294 XMMRegister vcoef0, XMMRegister vcoef1, XMMRegister vcoef2, XMMRegister vcoef3,
295 XMMRegister vresult0, XMMRegister vresult1, XMMRegister vresult2, XMMRegister vresult3,
296 XMMRegister vtmp0, XMMRegister vtmp1, XMMRegister vtmp2, XMMRegister vtmp3,
297 BasicType eltype);
298
299 // helper functions for arrays_hashcode
300 int arrays_hashcode_elsize(BasicType eltype);
301 void arrays_hashcode_elload(Register dst, Address src, BasicType eltype);
302 void arrays_hashcode_elvload(XMMRegister dst, Address src, BasicType eltype);
303 void arrays_hashcode_elvload(XMMRegister dst, AddressLiteral src, BasicType eltype);
304 void arrays_hashcode_elvcast(XMMRegister dst, BasicType eltype);
305
306 void convertF2I(BasicType dst_bt, BasicType src_bt, Register dst, XMMRegister src);
307
308 void evmasked_op(int ideal_opc, BasicType eType, KRegister mask,
309 XMMRegister dst, XMMRegister src1, XMMRegister src2,
310 bool merge, int vlen_enc, bool is_varshift = false);
311
312 void evmasked_op(int ideal_opc, BasicType eType, KRegister mask,
313 XMMRegister dst, XMMRegister src1, Address src2,
314 bool merge, int vlen_enc);
315
316 void evmasked_op(int ideal_opc, BasicType eType, KRegister mask, XMMRegister dst,
317 XMMRegister src1, int imm8, bool merge, int vlen_enc);
318
319 void masked_op(int ideal_opc, int mask_len, KRegister dst,
320 KRegister src1, KRegister src2);
321
322 void vector_unsigned_cast(XMMRegister dst, XMMRegister src, int vlen_enc,
323 BasicType from_elem_bt, BasicType to_elem_bt);
324
325 void vector_signed_cast(XMMRegister dst, XMMRegister src, int vlen_enc,
326 BasicType from_elem_bt, BasicType to_elem_bt);
327
328 void vector_cast_int_to_subword(BasicType to_elem_bt, XMMRegister dst, XMMRegister zero,
329 XMMRegister xtmp, Register rscratch, int vec_enc);
330
331 void vector_castF2X_avx(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
332 XMMRegister xtmp2, XMMRegister xtmp3, XMMRegister xtmp4,
333 AddressLiteral float_sign_flip, Register rscratch, int vec_enc);
334
335 void vector_castF2X_evex(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
336 XMMRegister xtmp2, KRegister ktmp1, KRegister ktmp2, AddressLiteral float_sign_flip,
337 Register rscratch, int vec_enc);
338
339 void vector_castF2L_evex(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
340 KRegister ktmp1, KRegister ktmp2, AddressLiteral double_sign_flip,
341 Register rscratch, int vec_enc);
342
343 void vector_castD2X_evex(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
344 XMMRegister xtmp2, KRegister ktmp1, KRegister ktmp2, AddressLiteral sign_flip,
345 Register rscratch, int vec_enc);
346
347 void vector_castD2X_avx(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
348 XMMRegister xtmp2, XMMRegister xtmp3, XMMRegister xtmp4, XMMRegister xtmp5,
349 AddressLiteral float_sign_flip, Register rscratch, int vec_enc);
350
351 void vector_castF2X_avx10(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, int vec_enc);
352
353 void vector_castF2X_avx10(BasicType to_elem_bt, XMMRegister dst, Address src, int vec_enc);
354
355 void vector_castD2X_avx10(BasicType to_elem_bt, XMMRegister dst, XMMRegister src, int vec_enc);
356
357 void vector_castD2X_avx10(BasicType to_elem_bt, XMMRegister dst, Address src, int vec_enc);
358
359 void vector_cast_double_to_int_special_cases_avx(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
360 XMMRegister xtmp3, XMMRegister xtmp4, XMMRegister xtmp5, Register rscratch,
361 AddressLiteral float_sign_flip, int vec_enc);
362
363 void vector_cast_double_to_int_special_cases_evex(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
364 KRegister ktmp1, KRegister ktmp2, Register rscratch, AddressLiteral float_sign_flip,
365 int vec_enc);
366
367 void vector_cast_double_to_long_special_cases_evex(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
368 KRegister ktmp1, KRegister ktmp2, Register rscratch, AddressLiteral double_sign_flip,
369 int vec_enc);
370
371 void vector_cast_float_to_int_special_cases_evex(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
372 KRegister ktmp1, KRegister ktmp2, Register rscratch, AddressLiteral float_sign_flip,
373 int vec_enc);
374
375 void vector_cast_float_to_long_special_cases_evex(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
376 KRegister ktmp1, KRegister ktmp2, Register rscratch, AddressLiteral double_sign_flip,
377 int vec_enc);
378
379 void vector_cast_float_to_int_special_cases_avx(XMMRegister dst, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2, XMMRegister xtmp3,
380 XMMRegister xtmp4, Register rscratch, AddressLiteral float_sign_flip,
381 int vec_enc);
382
383 void vector_crosslane_doubleword_pack_avx(XMMRegister dst, XMMRegister src, XMMRegister zero,
384 XMMRegister xtmp, int index, int vec_enc);
385
386 void vector_mask_cast(XMMRegister dst, XMMRegister src, BasicType dst_bt, BasicType src_bt, int vlen);
387
388 void vector_round_double_evex(XMMRegister dst, XMMRegister src, AddressLiteral double_sign_flip, AddressLiteral new_mxcsr, int vec_enc,
389 Register tmp, XMMRegister xtmp1, XMMRegister xtmp2, KRegister ktmp1, KRegister ktmp2);
390
391 void vector_round_float_evex(XMMRegister dst, XMMRegister src, AddressLiteral double_sign_flip, AddressLiteral new_mxcsr, int vec_enc,
392 Register tmp, XMMRegister xtmp1, XMMRegister xtmp2, KRegister ktmp1, KRegister ktmp2);
393
394 void vector_round_float_avx(XMMRegister dst, XMMRegister src, AddressLiteral float_sign_flip, AddressLiteral new_mxcsr, int vec_enc,
395 Register tmp, XMMRegister xtmp1, XMMRegister xtmp2, XMMRegister xtmp3, XMMRegister xtmp4);
396
397 void vector_compress_expand_avx2(int opcode, XMMRegister dst, XMMRegister src, XMMRegister mask,
398 Register rtmp, Register rscratch, XMMRegister permv, XMMRegister xtmp,
399 BasicType bt, int vec_enc);
400
401 void udivI(Register rax, Register divisor, Register rdx);
402 void umodI(Register rax, Register divisor, Register rdx);
403 void udivmodI(Register rax, Register divisor, Register rdx, Register tmp);
404
405 void reverseI(Register dst, Register src, XMMRegister xtmp1,
406 XMMRegister xtmp2, Register rtmp);
407 void reverseL(Register dst, Register src, XMMRegister xtmp1,
408 XMMRegister xtmp2, Register rtmp1, Register rtmp2);
409 void udivL(Register rax, Register divisor, Register rdx);
410 void umodL(Register rax, Register divisor, Register rdx);
411 void udivmodL(Register rax, Register divisor, Register rdx, Register tmp);
412
413 void evpternlog(XMMRegister dst, int func, KRegister mask, XMMRegister src2, XMMRegister src3,
414 bool merge, BasicType bt, int vlen_enc);
415
416 void evpternlog(XMMRegister dst, int func, KRegister mask, XMMRegister src2, Address src3,
417 bool merge, BasicType bt, int vlen_enc);
418
419 void vector_reverse_bit(BasicType bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
420 XMMRegister xtmp2, Register rtmp, int vec_enc);
421
422 void vector_reverse_bit_gfni(BasicType bt, XMMRegister dst, XMMRegister src, AddressLiteral mask, int vec_enc,
423 XMMRegister xtmp, Register rscratch = noreg);
424
425 void vector_reverse_byte(BasicType bt, XMMRegister dst, XMMRegister src, int vec_enc);
426
427 void vector_popcount_int(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
428 XMMRegister xtmp2, Register rtmp, int vec_enc);
429
430 void vector_popcount_long(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
431 XMMRegister xtmp2, Register rtmp, int vec_enc);
432
433 void vector_popcount_short(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
434 XMMRegister xtmp2, Register rtmp, int vec_enc);
435
436 void vector_popcount_byte(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
437 XMMRegister xtmp2, Register rtmp, int vec_enc);
438
439 void vector_popcount_integral(BasicType bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
440 XMMRegister xtmp2, Register rtmp, int vec_enc);
441
442 void vector_popcount_integral_evex(BasicType bt, XMMRegister dst, XMMRegister src,
443 KRegister mask, bool merge, int vec_enc);
444
445 void vbroadcast(BasicType bt, XMMRegister dst, int imm32, Register rtmp, int vec_enc);
446
447 void vector_reverse_byte64(BasicType bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
448 XMMRegister xtmp2, Register rtmp, int vec_enc);
449
450 void vector_count_leading_zeros_evex(BasicType bt, XMMRegister dst, XMMRegister src,
451 XMMRegister xtmp1, XMMRegister xtmp2, XMMRegister xtmp3,
452 KRegister ktmp, Register rtmp, bool merge, int vec_enc);
453
454 void vector_count_leading_zeros_byte_avx(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
455 XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp, int vec_enc);
456
457 void vector_count_leading_zeros_short_avx(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
458 XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp, int vec_enc);
459
460 void vector_count_leading_zeros_int_avx(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
461 XMMRegister xtmp2, XMMRegister xtmp3, int vec_enc);
462
463 void vector_count_leading_zeros_long_avx(XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
464 XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp, int vec_enc);
465
466 void vector_count_leading_zeros_avx(BasicType bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
467 XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp, int vec_enc);
468
469 void vpadd(BasicType bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vec_enc);
470
471 void vpsub(BasicType bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vec_enc);
472
473 void vector_count_trailing_zeros_evex(BasicType bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
474 XMMRegister xtmp2, XMMRegister xtmp3, XMMRegister xtmp4, KRegister ktmp,
475 Register rtmp, int vec_enc);
476
477 void vector_swap_nbits(int nbits, int bitmask, XMMRegister dst, XMMRegister src,
478 XMMRegister xtmp1, Register rtmp, int vec_enc);
479
480 void vector_count_trailing_zeros_avx(BasicType bt, XMMRegister dst, XMMRegister src, XMMRegister xtmp1,
481 XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp, int vec_enc);
482
483 void vector_signum_avx(int opcode, XMMRegister dst, XMMRegister src, XMMRegister zero, XMMRegister one,
484 XMMRegister xtmp1, int vec_enc);
485
486 void vector_signum_evex(int opcode, XMMRegister dst, XMMRegister src, XMMRegister zero, XMMRegister one,
487 KRegister ktmp1, int vec_enc);
488
489 void vmovmask(BasicType elem_bt, XMMRegister dst, Address src, XMMRegister mask, int vec_enc);
490
491 void vmovmask(BasicType elem_bt, Address dst, XMMRegister src, XMMRegister mask, int vec_enc);
492
493 void rearrange_bytes(XMMRegister dst, XMMRegister shuffle, XMMRegister src, XMMRegister xtmp1,
494 XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp, KRegister ktmp, int vlen_enc);
495
496 void vector_rearrange_int_float(BasicType bt, XMMRegister dst, XMMRegister shuffle,
497 XMMRegister src, int vlen_enc);
498
499 void efp16sh(int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2);
500
501 void vgather_subword(BasicType elem_ty, XMMRegister dst, Register base, Register idx_base, Register mask,
502 XMMRegister xtmp1, XMMRegister xtmp2, XMMRegister xtmp3, Register rtmp,
503 Register midx, Register length, int vector_len, int vlen_enc);
504
505 void vgather8b_masked(BasicType elem_bt, XMMRegister dst, Register base, Register idx_base,
506 Register mask, Register midx, Register rtmp, int vlen_enc);
507 void vgather8b(BasicType elem_bt, XMMRegister dst, Register base, Register idx_base,
508 Register rtmp, int vlen_enc);
509
510 void vector_saturating_op(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, bool is_unsigned, int vlen_enc);
511
512 void vector_saturating_op(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, Address src2, bool is_unsigned, int vlen_enc);
513
514 void vector_saturating_op(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vlen_enc);
515
516 void vector_saturating_op(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, Address src2, int vlen_enc);
517
518 void vector_saturating_unsigned_op(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vlen_enc);
519
520 void vector_saturating_unsigned_op(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, Address src2, int vlen_enc);
521
522 void vector_sub_dq_saturating_unsigned_evex(BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, KRegister ktmp, int vlen_enc);
523
524 void vector_sub_dq_saturating_unsigned_avx(BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2,
525 XMMRegister xtmp1, XMMRegister xtmp2, int vlen_enc);
526
527 void vector_add_dq_saturating_unsigned_evex(BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2,
528 XMMRegister xtmp1, XMMRegister xtmp2, KRegister ktmp, int vlen_enc);
529
530 void vector_add_dq_saturating_unsigned_avx(BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2,
531 XMMRegister xtmp1, XMMRegister xtmp2, XMMRegister xtmp3, int vlen_enc);
532
533 void vector_addsub_dq_saturating_avx(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2,
534 XMMRegister xtmp1, XMMRegister xtmp2, XMMRegister xtmp3, XMMRegister xtmp4, int vlen_enc);
535
536 void vector_addsub_dq_saturating_evex(int opc, BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2,
537 XMMRegister xtmp1, XMMRegister xtmp2, KRegister ktmp1, KRegister ktmp2, int vlen_enc);
538
539 void evpmovd2m_emu(KRegister ktmp, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2, int vlen_enc, bool xtmp2_hold_M1 = false);
540
541 void evpmovq2m_emu(KRegister ktmp, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2, int vlen_enc, bool xtmp2_hold_M1 = false);
542
543 void vpsign_extend_dq(BasicType etype, XMMRegister dst, XMMRegister src, int vlen_enc);
544
545 void vpgenmin_value(BasicType etype, XMMRegister dst, XMMRegister allones, int vlen_enc, bool compute_allones = false);
546
547 void vpgenmax_value(BasicType etype, XMMRegister dst, XMMRegister allones, int vlen_enc, bool compute_allones = false);
548
549 void evpcmpu(BasicType etype, KRegister kmask, XMMRegister src1, XMMRegister src2, Assembler::ComparisonPredicate cond, int vlen_enc);
550
551 void vpcmpgt(BasicType etype, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vlen_enc);
552
553 void evpmov_vec_to_mask(BasicType etype, KRegister ktmp, XMMRegister src, XMMRegister xtmp1, XMMRegister xtmp2,
554 int vlen_enc, bool xtmp2_hold_M1 = false);
555
556 void evmasked_saturating_op(int ideal_opc, BasicType elem_bt, KRegister mask, XMMRegister dst, XMMRegister src1, XMMRegister src2,
557 bool is_unsigned, bool merge, int vlen_enc);
558
559 void evmasked_saturating_op(int ideal_opc, BasicType elem_bt, KRegister mask, XMMRegister dst, XMMRegister src1, Address src2,
560 bool is_unsigned, bool merge, int vlen_enc);
561
562 void evmasked_saturating_signed_op(int ideal_opc, BasicType elem_bt, KRegister mask, XMMRegister dst, XMMRegister src1, XMMRegister src2,
563 bool merge, int vlen_enc);
564
565 void evmasked_saturating_signed_op(int ideal_opc, BasicType elem_bt, KRegister mask, XMMRegister dst, XMMRegister src1, Address src2,
566 bool merge, int vlen_enc);
567
568 void evmasked_saturating_unsigned_op(int ideal_opc, BasicType elem_bt, KRegister mask, XMMRegister dst, XMMRegister src1,
569 XMMRegister src2, bool merge, int vlen_enc);
570
571 void evmasked_saturating_unsigned_op(int ideal_opc, BasicType elem_bt, KRegister mask, XMMRegister dst, XMMRegister src1,
572 Address src2, bool merge, int vlen_enc);
573
574 void select_from_two_vectors_evex(BasicType elem_bt, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vlen_enc);
575
576 void evfp16ph(int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2, int vlen_enc);
577
578 void evfp16ph(int opcode, XMMRegister dst, XMMRegister src1, Address src2, int vlen_enc);
579
580 void vector_max_min_fp16(int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2,
581 KRegister ktmp, XMMRegister xtmp1, XMMRegister xtmp2, int vlen_enc);
582
583 void scalar_max_min_fp16(int opcode, XMMRegister dst, XMMRegister src1, XMMRegister src2,
584 KRegister ktmp, XMMRegister xtmp1, XMMRegister xtmp2);
585
586 void reconstruct_frame_pointer(Register rtmp);
587
588 #endif // CPU_X86_C2_MACROASSEMBLER_X86_HPP