1/* 2 * Tiny Code Generator for QEMU 3 * 4 * Copyright (c) 2009 Ulrich Hecht <uli@suse.de> 5 * Copyright (c) 2009 Alexander Graf <agraf@suse.de> 6 * Copyright (c) 2010 Richard Henderson <rth@twiddle.net> 7 * 8 * Permission is hereby granted, free of charge, to any person obtaining a copy 9 * of this software and associated documentation files (the "Software"), to deal 10 * in the Software without restriction, including without limitation the rights 11 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 12 * copies of the Software, and to permit persons to whom the Software is 13 * furnished to do so, subject to the following conditions: 14 * 15 * The above copyright notice and this permission notice shall be included in 16 * all copies or substantial portions of the Software. 17 * 18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 19 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 20 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 21 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 22 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 23 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN 24 * THE SOFTWARE. 25 */ 26 27#include "../tcg-ldst.c.inc" 28#include "../tcg-pool.c.inc" 29#include "elf.h" 30 31#define TCG_CT_CONST_S16 (1 << 8) 32#define TCG_CT_CONST_S32 (1 << 9) 33#define TCG_CT_CONST_S33 (1 << 10) 34#define TCG_CT_CONST_ZERO (1 << 11) 35#define TCG_CT_CONST_P32 (1 << 12) 36#define TCG_CT_CONST_INV (1 << 13) 37#define TCG_CT_CONST_INVRISBG (1 << 14) 38 39#define ALL_GENERAL_REGS MAKE_64BIT_MASK(0, 16) 40#define ALL_VECTOR_REGS MAKE_64BIT_MASK(32, 32) 41 42/* Several places within the instruction set 0 means "no register" 43 rather than TCG_REG_R0. */ 44#define TCG_REG_NONE 0 45 46/* A scratch register that may be be used throughout the backend. */ 47#define TCG_TMP0 TCG_REG_R1 48 49#ifndef CONFIG_SOFTMMU 50#define TCG_GUEST_BASE_REG TCG_REG_R13 51#endif 52 53/* All of the following instructions are prefixed with their instruction 54 format, and are defined as 8- or 16-bit quantities, even when the two 55 halves of the 16-bit quantity may appear 32 bits apart in the insn. 56 This makes it easy to copy the values from the tables in Appendix B. */ 57typedef enum S390Opcode { 58 RIL_AFI = 0xc209, 59 RIL_AGFI = 0xc208, 60 RIL_ALFI = 0xc20b, 61 RIL_ALGFI = 0xc20a, 62 RIL_BRASL = 0xc005, 63 RIL_BRCL = 0xc004, 64 RIL_CFI = 0xc20d, 65 RIL_CGFI = 0xc20c, 66 RIL_CLFI = 0xc20f, 67 RIL_CLGFI = 0xc20e, 68 RIL_CLRL = 0xc60f, 69 RIL_CLGRL = 0xc60a, 70 RIL_CRL = 0xc60d, 71 RIL_CGRL = 0xc608, 72 RIL_IIHF = 0xc008, 73 RIL_IILF = 0xc009, 74 RIL_LARL = 0xc000, 75 RIL_LGFI = 0xc001, 76 RIL_LGRL = 0xc408, 77 RIL_LLIHF = 0xc00e, 78 RIL_LLILF = 0xc00f, 79 RIL_LRL = 0xc40d, 80 RIL_MSFI = 0xc201, 81 RIL_MSGFI = 0xc200, 82 RIL_NIHF = 0xc00a, 83 RIL_NILF = 0xc00b, 84 RIL_OIHF = 0xc00c, 85 RIL_OILF = 0xc00d, 86 RIL_SLFI = 0xc205, 87 RIL_SLGFI = 0xc204, 88 RIL_XIHF = 0xc006, 89 RIL_XILF = 0xc007, 90 91 RI_AGHI = 0xa70b, 92 RI_AHI = 0xa70a, 93 RI_BRC = 0xa704, 94 RI_CHI = 0xa70e, 95 RI_CGHI = 0xa70f, 96 RI_IIHH = 0xa500, 97 RI_IIHL = 0xa501, 98 RI_IILH = 0xa502, 99 RI_IILL = 0xa503, 100 RI_LGHI = 0xa709, 101 RI_LLIHH = 0xa50c, 102 RI_LLIHL = 0xa50d, 103 RI_LLILH = 0xa50e, 104 RI_LLILL = 0xa50f, 105 RI_MGHI = 0xa70d, 106 RI_MHI = 0xa70c, 107 RI_NIHH = 0xa504, 108 RI_NIHL = 0xa505, 109 RI_NILH = 0xa506, 110 RI_NILL = 0xa507, 111 RI_OIHH = 0xa508, 112 RI_OIHL = 0xa509, 113 RI_OILH = 0xa50a, 114 RI_OILL = 0xa50b, 115 RI_TMLL = 0xa701, 116 117 RIEb_CGRJ = 0xec64, 118 RIEb_CLGRJ = 0xec65, 119 RIEb_CLRJ = 0xec77, 120 RIEb_CRJ = 0xec76, 121 122 RIEc_CGIJ = 0xec7c, 123 RIEc_CIJ = 0xec7e, 124 RIEc_CLGIJ = 0xec7d, 125 RIEc_CLIJ = 0xec7f, 126 127 RIEf_RISBG = 0xec55, 128 129 RIEg_LOCGHI = 0xec46, 130 131 RRE_AGR = 0xb908, 132 RRE_ALGR = 0xb90a, 133 RRE_ALCR = 0xb998, 134 RRE_ALCGR = 0xb988, 135 RRE_ALGFR = 0xb91a, 136 RRE_CGR = 0xb920, 137 RRE_CLGR = 0xb921, 138 RRE_DLGR = 0xb987, 139 RRE_DLR = 0xb997, 140 RRE_DSGFR = 0xb91d, 141 RRE_DSGR = 0xb90d, 142 RRE_FLOGR = 0xb983, 143 RRE_LGBR = 0xb906, 144 RRE_LCGR = 0xb903, 145 RRE_LGFR = 0xb914, 146 RRE_LGHR = 0xb907, 147 RRE_LGR = 0xb904, 148 RRE_LLGCR = 0xb984, 149 RRE_LLGFR = 0xb916, 150 RRE_LLGHR = 0xb985, 151 RRE_LRVR = 0xb91f, 152 RRE_LRVGR = 0xb90f, 153 RRE_LTGR = 0xb902, 154 RRE_MLGR = 0xb986, 155 RRE_MSGR = 0xb90c, 156 RRE_MSR = 0xb252, 157 RRE_NGR = 0xb980, 158 RRE_OGR = 0xb981, 159 RRE_SGR = 0xb909, 160 RRE_SLGR = 0xb90b, 161 RRE_SLBR = 0xb999, 162 RRE_SLBGR = 0xb989, 163 RRE_XGR = 0xb982, 164 165 RRFa_MGRK = 0xb9ec, 166 RRFa_MSRKC = 0xb9fd, 167 RRFa_MSGRKC = 0xb9ed, 168 RRFa_NCRK = 0xb9f5, 169 RRFa_NCGRK = 0xb9e5, 170 RRFa_NNRK = 0xb974, 171 RRFa_NNGRK = 0xb964, 172 RRFa_NORK = 0xb976, 173 RRFa_NOGRK = 0xb966, 174 RRFa_NRK = 0xb9f4, 175 RRFa_NGRK = 0xb9e4, 176 RRFa_NXRK = 0xb977, 177 RRFa_NXGRK = 0xb967, 178 RRFa_OCRK = 0xb975, 179 RRFa_OCGRK = 0xb965, 180 RRFa_ORK = 0xb9f6, 181 RRFa_OGRK = 0xb9e6, 182 RRFa_SRK = 0xb9f9, 183 RRFa_SGRK = 0xb9e9, 184 RRFa_SLRK = 0xb9fb, 185 RRFa_SLGRK = 0xb9eb, 186 RRFa_XRK = 0xb9f7, 187 RRFa_XGRK = 0xb9e7, 188 189 RRFam_SELGR = 0xb9e3, 190 191 RRFc_LOCR = 0xb9f2, 192 RRFc_LOCGR = 0xb9e2, 193 RRFc_POPCNT = 0xb9e1, 194 195 RR_AR = 0x1a, 196 RR_ALR = 0x1e, 197 RR_BASR = 0x0d, 198 RR_BCR = 0x07, 199 RR_CLR = 0x15, 200 RR_CR = 0x19, 201 RR_DR = 0x1d, 202 RR_LCR = 0x13, 203 RR_LR = 0x18, 204 RR_LTR = 0x12, 205 RR_NR = 0x14, 206 RR_OR = 0x16, 207 RR_SR = 0x1b, 208 RR_SLR = 0x1f, 209 RR_XR = 0x17, 210 211 RSY_RLL = 0xeb1d, 212 RSY_RLLG = 0xeb1c, 213 RSY_SLLG = 0xeb0d, 214 RSY_SLLK = 0xebdf, 215 RSY_SRAG = 0xeb0a, 216 RSY_SRAK = 0xebdc, 217 RSY_SRLG = 0xeb0c, 218 RSY_SRLK = 0xebde, 219 220 RS_SLL = 0x89, 221 RS_SRA = 0x8a, 222 RS_SRL = 0x88, 223 224 RXY_AG = 0xe308, 225 RXY_AY = 0xe35a, 226 RXY_CG = 0xe320, 227 RXY_CLG = 0xe321, 228 RXY_CLY = 0xe355, 229 RXY_CY = 0xe359, 230 RXY_LAY = 0xe371, 231 RXY_LB = 0xe376, 232 RXY_LG = 0xe304, 233 RXY_LGB = 0xe377, 234 RXY_LGF = 0xe314, 235 RXY_LGH = 0xe315, 236 RXY_LHY = 0xe378, 237 RXY_LLGC = 0xe390, 238 RXY_LLGF = 0xe316, 239 RXY_LLGH = 0xe391, 240 RXY_LMG = 0xeb04, 241 RXY_LPQ = 0xe38f, 242 RXY_LRV = 0xe31e, 243 RXY_LRVG = 0xe30f, 244 RXY_LRVH = 0xe31f, 245 RXY_LY = 0xe358, 246 RXY_NG = 0xe380, 247 RXY_OG = 0xe381, 248 RXY_STCY = 0xe372, 249 RXY_STG = 0xe324, 250 RXY_STHY = 0xe370, 251 RXY_STMG = 0xeb24, 252 RXY_STPQ = 0xe38e, 253 RXY_STRV = 0xe33e, 254 RXY_STRVG = 0xe32f, 255 RXY_STRVH = 0xe33f, 256 RXY_STY = 0xe350, 257 RXY_XG = 0xe382, 258 259 RX_A = 0x5a, 260 RX_C = 0x59, 261 RX_L = 0x58, 262 RX_LA = 0x41, 263 RX_LH = 0x48, 264 RX_ST = 0x50, 265 RX_STC = 0x42, 266 RX_STH = 0x40, 267 268 VRIa_VGBM = 0xe744, 269 VRIa_VREPI = 0xe745, 270 VRIb_VGM = 0xe746, 271 VRIc_VREP = 0xe74d, 272 273 VRRa_VLC = 0xe7de, 274 VRRa_VLP = 0xe7df, 275 VRRa_VLR = 0xe756, 276 VRRc_VA = 0xe7f3, 277 VRRc_VCEQ = 0xe7f8, /* we leave the m5 cs field 0 */ 278 VRRc_VCH = 0xe7fb, /* " */ 279 VRRc_VCHL = 0xe7f9, /* " */ 280 VRRc_VERLLV = 0xe773, 281 VRRc_VESLV = 0xe770, 282 VRRc_VESRAV = 0xe77a, 283 VRRc_VESRLV = 0xe778, 284 VRRc_VML = 0xe7a2, 285 VRRc_VMN = 0xe7fe, 286 VRRc_VMNL = 0xe7fc, 287 VRRc_VMX = 0xe7ff, 288 VRRc_VMXL = 0xe7fd, 289 VRRc_VN = 0xe768, 290 VRRc_VNC = 0xe769, 291 VRRc_VNN = 0xe76e, 292 VRRc_VNO = 0xe76b, 293 VRRc_VNX = 0xe76c, 294 VRRc_VO = 0xe76a, 295 VRRc_VOC = 0xe76f, 296 VRRc_VPKS = 0xe797, /* we leave the m5 cs field 0 */ 297 VRRc_VS = 0xe7f7, 298 VRRa_VUPH = 0xe7d7, 299 VRRa_VUPL = 0xe7d6, 300 VRRc_VX = 0xe76d, 301 VRRe_VSEL = 0xe78d, 302 VRRf_VLVGP = 0xe762, 303 304 VRSa_VERLL = 0xe733, 305 VRSa_VESL = 0xe730, 306 VRSa_VESRA = 0xe73a, 307 VRSa_VESRL = 0xe738, 308 VRSb_VLVG = 0xe722, 309 VRSc_VLGV = 0xe721, 310 311 VRX_VL = 0xe706, 312 VRX_VLLEZ = 0xe704, 313 VRX_VLREP = 0xe705, 314 VRX_VST = 0xe70e, 315 VRX_VSTEF = 0xe70b, 316 VRX_VSTEG = 0xe70a, 317 318 NOP = 0x0707, 319} S390Opcode; 320 321#ifdef CONFIG_DEBUG_TCG 322static const char * const tcg_target_reg_names[TCG_TARGET_NB_REGS] = { 323 "%r0", "%r1", "%r2", "%r3", "%r4", "%r5", "%r6", "%r7", 324 "%r8", "%r9", "%r10", "%r11", "%r12", "%r13", "%r14", "%r15", 325 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 326 "%v0", "%v1", "%v2", "%v3", "%v4", "%v5", "%v6", "%v7", 327 "%v8", "%v9", "%v10", "%v11", "%v12", "%v13", "%v14", "%v15", 328 "%v16", "%v17", "%v18", "%v19", "%v20", "%v21", "%v22", "%v23", 329 "%v24", "%v25", "%v26", "%v27", "%v28", "%v29", "%v30", "%v31", 330}; 331#endif 332 333/* Since R6 is a potential argument register, choose it last of the 334 call-saved registers. Likewise prefer the call-clobbered registers 335 in reverse order to maximize the chance of avoiding the arguments. */ 336static const int tcg_target_reg_alloc_order[] = { 337 /* Call saved registers. */ 338 TCG_REG_R13, 339 TCG_REG_R12, 340 TCG_REG_R11, 341 TCG_REG_R10, 342 TCG_REG_R9, 343 TCG_REG_R8, 344 TCG_REG_R7, 345 TCG_REG_R6, 346 /* Call clobbered registers. */ 347 TCG_REG_R14, 348 TCG_REG_R0, 349 TCG_REG_R1, 350 /* Argument registers, in reverse order of allocation. */ 351 TCG_REG_R5, 352 TCG_REG_R4, 353 TCG_REG_R3, 354 TCG_REG_R2, 355 356 /* V8-V15 are call saved, and omitted. */ 357 TCG_REG_V0, 358 TCG_REG_V1, 359 TCG_REG_V2, 360 TCG_REG_V3, 361 TCG_REG_V4, 362 TCG_REG_V5, 363 TCG_REG_V6, 364 TCG_REG_V7, 365 TCG_REG_V16, 366 TCG_REG_V17, 367 TCG_REG_V18, 368 TCG_REG_V19, 369 TCG_REG_V20, 370 TCG_REG_V21, 371 TCG_REG_V22, 372 TCG_REG_V23, 373 TCG_REG_V24, 374 TCG_REG_V25, 375 TCG_REG_V26, 376 TCG_REG_V27, 377 TCG_REG_V28, 378 TCG_REG_V29, 379 TCG_REG_V30, 380 TCG_REG_V31, 381}; 382 383static const int tcg_target_call_iarg_regs[] = { 384 TCG_REG_R2, 385 TCG_REG_R3, 386 TCG_REG_R4, 387 TCG_REG_R5, 388 TCG_REG_R6, 389}; 390 391static TCGReg tcg_target_call_oarg_reg(TCGCallReturnKind kind, int slot) 392{ 393 tcg_debug_assert(kind == TCG_CALL_RET_NORMAL); 394 tcg_debug_assert(slot == 0); 395 return TCG_REG_R2; 396} 397 398#define S390_CC_EQ 8 399#define S390_CC_LT 4 400#define S390_CC_GT 2 401#define S390_CC_OV 1 402#define S390_CC_NE (S390_CC_LT | S390_CC_GT) 403#define S390_CC_LE (S390_CC_LT | S390_CC_EQ) 404#define S390_CC_GE (S390_CC_GT | S390_CC_EQ) 405#define S390_CC_NEVER 0 406#define S390_CC_ALWAYS 15 407 408/* Condition codes that result from a COMPARE and COMPARE LOGICAL. */ 409static const uint8_t tcg_cond_to_s390_cond[] = { 410 [TCG_COND_EQ] = S390_CC_EQ, 411 [TCG_COND_NE] = S390_CC_NE, 412 [TCG_COND_LT] = S390_CC_LT, 413 [TCG_COND_LE] = S390_CC_LE, 414 [TCG_COND_GT] = S390_CC_GT, 415 [TCG_COND_GE] = S390_CC_GE, 416 [TCG_COND_LTU] = S390_CC_LT, 417 [TCG_COND_LEU] = S390_CC_LE, 418 [TCG_COND_GTU] = S390_CC_GT, 419 [TCG_COND_GEU] = S390_CC_GE, 420}; 421 422/* Condition codes that result from a LOAD AND TEST. Here, we have no 423 unsigned instruction variation, however since the test is vs zero we 424 can re-map the outcomes appropriately. */ 425static const uint8_t tcg_cond_to_ltr_cond[] = { 426 [TCG_COND_EQ] = S390_CC_EQ, 427 [TCG_COND_NE] = S390_CC_NE, 428 [TCG_COND_LT] = S390_CC_LT, 429 [TCG_COND_LE] = S390_CC_LE, 430 [TCG_COND_GT] = S390_CC_GT, 431 [TCG_COND_GE] = S390_CC_GE, 432 [TCG_COND_LTU] = S390_CC_NEVER, 433 [TCG_COND_LEU] = S390_CC_EQ, 434 [TCG_COND_GTU] = S390_CC_NE, 435 [TCG_COND_GEU] = S390_CC_ALWAYS, 436}; 437 438static const tcg_insn_unit *tb_ret_addr; 439uint64_t s390_facilities[3]; 440 441static inline bool is_general_reg(TCGReg r) 442{ 443 return r <= TCG_REG_R15; 444} 445 446static inline bool is_vector_reg(TCGReg r) 447{ 448 return r >= TCG_REG_V0 && r <= TCG_REG_V31; 449} 450 451static bool patch_reloc(tcg_insn_unit *src_rw, int type, 452 intptr_t value, intptr_t addend) 453{ 454 const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw); 455 intptr_t pcrel2; 456 uint32_t old; 457 458 value += addend; 459 pcrel2 = (tcg_insn_unit *)value - src_rx; 460 461 switch (type) { 462 case R_390_PC16DBL: 463 if (pcrel2 == (int16_t)pcrel2) { 464 tcg_patch16(src_rw, pcrel2); 465 return true; 466 } 467 break; 468 case R_390_PC32DBL: 469 if (pcrel2 == (int32_t)pcrel2) { 470 tcg_patch32(src_rw, pcrel2); 471 return true; 472 } 473 break; 474 case R_390_20: 475 if (value == sextract64(value, 0, 20)) { 476 old = *(uint32_t *)src_rw & 0xf00000ff; 477 old |= ((value & 0xfff) << 16) | ((value & 0xff000) >> 4); 478 tcg_patch32(src_rw, old); 479 return true; 480 } 481 break; 482 default: 483 g_assert_not_reached(); 484 } 485 return false; 486} 487 488static int is_const_p16(uint64_t val) 489{ 490 for (int i = 0; i < 4; ++i) { 491 uint64_t mask = 0xffffull << (i * 16); 492 if ((val & ~mask) == 0) { 493 return i; 494 } 495 } 496 return -1; 497} 498 499static int is_const_p32(uint64_t val) 500{ 501 if ((val & 0xffffffff00000000ull) == 0) { 502 return 0; 503 } 504 if ((val & 0x00000000ffffffffull) == 0) { 505 return 1; 506 } 507 return -1; 508} 509 510/* 511 * Accept bit patterns like these: 512 * 0....01....1 513 * 1....10....0 514 * 1..10..01..1 515 * 0..01..10..0 516 * Copied from gcc sources. 517 */ 518static bool risbg_mask(uint64_t c) 519{ 520 uint64_t lsb; 521 /* We don't change the number of transitions by inverting, 522 so make sure we start with the LSB zero. */ 523 if (c & 1) { 524 c = ~c; 525 } 526 /* Reject all zeros or all ones. */ 527 if (c == 0) { 528 return false; 529 } 530 /* Find the first transition. */ 531 lsb = c & -c; 532 /* Invert to look for a second transition. */ 533 c = ~c; 534 /* Erase the first transition. */ 535 c &= -lsb; 536 /* Find the second transition, if any. */ 537 lsb = c & -c; 538 /* Match if all the bits are 1's, or if c is zero. */ 539 return c == -lsb; 540} 541 542/* Test if a constant matches the constraint. */ 543static bool tcg_target_const_match(int64_t val, TCGType type, int ct) 544{ 545 if (ct & TCG_CT_CONST) { 546 return 1; 547 } 548 549 if (type == TCG_TYPE_I32) { 550 val = (int32_t)val; 551 } 552 553 /* The following are mutually exclusive. */ 554 if (ct & TCG_CT_CONST_S16) { 555 return val == (int16_t)val; 556 } else if (ct & TCG_CT_CONST_S32) { 557 return val == (int32_t)val; 558 } else if (ct & TCG_CT_CONST_S33) { 559 return val >= -0xffffffffll && val <= 0xffffffffll; 560 } else if (ct & TCG_CT_CONST_ZERO) { 561 return val == 0; 562 } 563 564 if (ct & TCG_CT_CONST_INV) { 565 val = ~val; 566 } 567 /* 568 * Note that is_const_p16 is a subset of is_const_p32, 569 * so we don't need both constraints. 570 */ 571 if ((ct & TCG_CT_CONST_P32) && is_const_p32(val) >= 0) { 572 return true; 573 } 574 if ((ct & TCG_CT_CONST_INVRISBG) && risbg_mask(~val)) { 575 return true; 576 } 577 578 return 0; 579} 580 581/* Emit instructions according to the given instruction format. */ 582 583static void tcg_out_insn_RR(TCGContext *s, S390Opcode op, TCGReg r1, TCGReg r2) 584{ 585 tcg_out16(s, (op << 8) | (r1 << 4) | r2); 586} 587 588static void tcg_out_insn_RRE(TCGContext *s, S390Opcode op, 589 TCGReg r1, TCGReg r2) 590{ 591 tcg_out32(s, (op << 16) | (r1 << 4) | r2); 592} 593 594/* RRF-a without the m4 field */ 595static void tcg_out_insn_RRFa(TCGContext *s, S390Opcode op, 596 TCGReg r1, TCGReg r2, TCGReg r3) 597{ 598 tcg_out32(s, (op << 16) | (r3 << 12) | (r1 << 4) | r2); 599} 600 601/* RRF-a with the m4 field */ 602static void tcg_out_insn_RRFam(TCGContext *s, S390Opcode op, 603 TCGReg r1, TCGReg r2, TCGReg r3, int m4) 604{ 605 tcg_out32(s, (op << 16) | (r3 << 12) | (m4 << 8) | (r1 << 4) | r2); 606} 607 608static void tcg_out_insn_RRFc(TCGContext *s, S390Opcode op, 609 TCGReg r1, TCGReg r2, int m3) 610{ 611 tcg_out32(s, (op << 16) | (m3 << 12) | (r1 << 4) | r2); 612} 613 614static void tcg_out_insn_RI(TCGContext *s, S390Opcode op, TCGReg r1, int i2) 615{ 616 tcg_out32(s, (op << 16) | (r1 << 20) | (i2 & 0xffff)); 617} 618 619static void tcg_out_insn_RIEg(TCGContext *s, S390Opcode op, TCGReg r1, 620 int i2, int m3) 621{ 622 tcg_out16(s, (op & 0xff00) | (r1 << 4) | m3); 623 tcg_out32(s, (i2 << 16) | (op & 0xff)); 624} 625 626static void tcg_out_insn_RIL(TCGContext *s, S390Opcode op, TCGReg r1, int i2) 627{ 628 tcg_out16(s, op | (r1 << 4)); 629 tcg_out32(s, i2); 630} 631 632static void tcg_out_insn_RS(TCGContext *s, S390Opcode op, TCGReg r1, 633 TCGReg b2, TCGReg r3, int disp) 634{ 635 tcg_out32(s, (op << 24) | (r1 << 20) | (r3 << 16) | (b2 << 12) 636 | (disp & 0xfff)); 637} 638 639static void tcg_out_insn_RSY(TCGContext *s, S390Opcode op, TCGReg r1, 640 TCGReg b2, TCGReg r3, int disp) 641{ 642 tcg_out16(s, (op & 0xff00) | (r1 << 4) | r3); 643 tcg_out32(s, (op & 0xff) | (b2 << 28) 644 | ((disp & 0xfff) << 16) | ((disp & 0xff000) >> 4)); 645} 646 647#define tcg_out_insn_RX tcg_out_insn_RS 648#define tcg_out_insn_RXY tcg_out_insn_RSY 649 650static int RXB(TCGReg v1, TCGReg v2, TCGReg v3, TCGReg v4) 651{ 652 /* 653 * Shift bit 4 of each regno to its corresponding bit of RXB. 654 * RXB itself begins at bit 8 of the instruction so 8 - 4 = 4 655 * is the left-shift of the 4th operand. 656 */ 657 return ((v1 & 0x10) << (4 + 3)) 658 | ((v2 & 0x10) << (4 + 2)) 659 | ((v3 & 0x10) << (4 + 1)) 660 | ((v4 & 0x10) << (4 + 0)); 661} 662 663static void tcg_out_insn_VRIa(TCGContext *s, S390Opcode op, 664 TCGReg v1, uint16_t i2, int m3) 665{ 666 tcg_debug_assert(is_vector_reg(v1)); 667 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4)); 668 tcg_out16(s, i2); 669 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, 0, 0) | (m3 << 12)); 670} 671 672static void tcg_out_insn_VRIb(TCGContext *s, S390Opcode op, 673 TCGReg v1, uint8_t i2, uint8_t i3, int m4) 674{ 675 tcg_debug_assert(is_vector_reg(v1)); 676 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4)); 677 tcg_out16(s, (i2 << 8) | (i3 & 0xff)); 678 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, 0, 0) | (m4 << 12)); 679} 680 681static void tcg_out_insn_VRIc(TCGContext *s, S390Opcode op, 682 TCGReg v1, uint16_t i2, TCGReg v3, int m4) 683{ 684 tcg_debug_assert(is_vector_reg(v1)); 685 tcg_debug_assert(is_vector_reg(v3)); 686 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | (v3 & 0xf)); 687 tcg_out16(s, i2); 688 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, v3, 0) | (m4 << 12)); 689} 690 691static void tcg_out_insn_VRRa(TCGContext *s, S390Opcode op, 692 TCGReg v1, TCGReg v2, int m3) 693{ 694 tcg_debug_assert(is_vector_reg(v1)); 695 tcg_debug_assert(is_vector_reg(v2)); 696 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | (v2 & 0xf)); 697 tcg_out32(s, (op & 0x00ff) | RXB(v1, v2, 0, 0) | (m3 << 12)); 698} 699 700static void tcg_out_insn_VRRc(TCGContext *s, S390Opcode op, 701 TCGReg v1, TCGReg v2, TCGReg v3, int m4) 702{ 703 tcg_debug_assert(is_vector_reg(v1)); 704 tcg_debug_assert(is_vector_reg(v2)); 705 tcg_debug_assert(is_vector_reg(v3)); 706 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | (v2 & 0xf)); 707 tcg_out16(s, v3 << 12); 708 tcg_out16(s, (op & 0x00ff) | RXB(v1, v2, v3, 0) | (m4 << 12)); 709} 710 711static void tcg_out_insn_VRRe(TCGContext *s, S390Opcode op, 712 TCGReg v1, TCGReg v2, TCGReg v3, TCGReg v4) 713{ 714 tcg_debug_assert(is_vector_reg(v1)); 715 tcg_debug_assert(is_vector_reg(v2)); 716 tcg_debug_assert(is_vector_reg(v3)); 717 tcg_debug_assert(is_vector_reg(v4)); 718 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | (v2 & 0xf)); 719 tcg_out16(s, v3 << 12); 720 tcg_out16(s, (op & 0x00ff) | RXB(v1, v2, v3, v4) | (v4 << 12)); 721} 722 723static void tcg_out_insn_VRRf(TCGContext *s, S390Opcode op, 724 TCGReg v1, TCGReg r2, TCGReg r3) 725{ 726 tcg_debug_assert(is_vector_reg(v1)); 727 tcg_debug_assert(is_general_reg(r2)); 728 tcg_debug_assert(is_general_reg(r3)); 729 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | r2); 730 tcg_out16(s, r3 << 12); 731 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, 0, 0)); 732} 733 734static void tcg_out_insn_VRSa(TCGContext *s, S390Opcode op, TCGReg v1, 735 intptr_t d2, TCGReg b2, TCGReg v3, int m4) 736{ 737 tcg_debug_assert(is_vector_reg(v1)); 738 tcg_debug_assert(d2 >= 0 && d2 <= 0xfff); 739 tcg_debug_assert(is_general_reg(b2)); 740 tcg_debug_assert(is_vector_reg(v3)); 741 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | (v3 & 0xf)); 742 tcg_out16(s, b2 << 12 | d2); 743 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, v3, 0) | (m4 << 12)); 744} 745 746static void tcg_out_insn_VRSb(TCGContext *s, S390Opcode op, TCGReg v1, 747 intptr_t d2, TCGReg b2, TCGReg r3, int m4) 748{ 749 tcg_debug_assert(is_vector_reg(v1)); 750 tcg_debug_assert(d2 >= 0 && d2 <= 0xfff); 751 tcg_debug_assert(is_general_reg(b2)); 752 tcg_debug_assert(is_general_reg(r3)); 753 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | r3); 754 tcg_out16(s, b2 << 12 | d2); 755 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, 0, 0) | (m4 << 12)); 756} 757 758static void tcg_out_insn_VRSc(TCGContext *s, S390Opcode op, TCGReg r1, 759 intptr_t d2, TCGReg b2, TCGReg v3, int m4) 760{ 761 tcg_debug_assert(is_general_reg(r1)); 762 tcg_debug_assert(d2 >= 0 && d2 <= 0xfff); 763 tcg_debug_assert(is_general_reg(b2)); 764 tcg_debug_assert(is_vector_reg(v3)); 765 tcg_out16(s, (op & 0xff00) | (r1 << 4) | (v3 & 0xf)); 766 tcg_out16(s, b2 << 12 | d2); 767 tcg_out16(s, (op & 0x00ff) | RXB(0, 0, v3, 0) | (m4 << 12)); 768} 769 770static void tcg_out_insn_VRX(TCGContext *s, S390Opcode op, TCGReg v1, 771 TCGReg b2, TCGReg x2, intptr_t d2, int m3) 772{ 773 tcg_debug_assert(is_vector_reg(v1)); 774 tcg_debug_assert(d2 >= 0 && d2 <= 0xfff); 775 tcg_debug_assert(is_general_reg(x2)); 776 tcg_debug_assert(is_general_reg(b2)); 777 tcg_out16(s, (op & 0xff00) | ((v1 & 0xf) << 4) | x2); 778 tcg_out16(s, (b2 << 12) | d2); 779 tcg_out16(s, (op & 0x00ff) | RXB(v1, 0, 0, 0) | (m3 << 12)); 780} 781 782/* Emit an opcode with "type-checking" of the format. */ 783#define tcg_out_insn(S, FMT, OP, ...) \ 784 glue(tcg_out_insn_,FMT)(S, glue(glue(FMT,_),OP), ## __VA_ARGS__) 785 786 787/* emit 64-bit shifts */ 788static void tcg_out_sh64(TCGContext* s, S390Opcode op, TCGReg dest, 789 TCGReg src, TCGReg sh_reg, int sh_imm) 790{ 791 tcg_out_insn_RSY(s, op, dest, sh_reg, src, sh_imm); 792} 793 794/* emit 32-bit shifts */ 795static void tcg_out_sh32(TCGContext* s, S390Opcode op, TCGReg dest, 796 TCGReg sh_reg, int sh_imm) 797{ 798 tcg_out_insn_RS(s, op, dest, sh_reg, 0, sh_imm); 799} 800 801static bool tcg_out_mov(TCGContext *s, TCGType type, TCGReg dst, TCGReg src) 802{ 803 if (src == dst) { 804 return true; 805 } 806 switch (type) { 807 case TCG_TYPE_I32: 808 if (likely(is_general_reg(dst) && is_general_reg(src))) { 809 tcg_out_insn(s, RR, LR, dst, src); 810 break; 811 } 812 /* fallthru */ 813 814 case TCG_TYPE_I64: 815 if (likely(is_general_reg(dst))) { 816 if (likely(is_general_reg(src))) { 817 tcg_out_insn(s, RRE, LGR, dst, src); 818 } else { 819 tcg_out_insn(s, VRSc, VLGV, dst, 0, 0, src, 3); 820 } 821 break; 822 } else if (is_general_reg(src)) { 823 tcg_out_insn(s, VRSb, VLVG, dst, 0, 0, src, 3); 824 break; 825 } 826 /* fallthru */ 827 828 case TCG_TYPE_V64: 829 case TCG_TYPE_V128: 830 tcg_out_insn(s, VRRa, VLR, dst, src, 0); 831 break; 832 833 default: 834 g_assert_not_reached(); 835 } 836 return true; 837} 838 839static const S390Opcode li_insns[4] = { 840 RI_LLILL, RI_LLILH, RI_LLIHL, RI_LLIHH 841}; 842static const S390Opcode oi_insns[4] = { 843 RI_OILL, RI_OILH, RI_OIHL, RI_OIHH 844}; 845static const S390Opcode lif_insns[2] = { 846 RIL_LLILF, RIL_LLIHF, 847}; 848 849/* load a register with an immediate value */ 850static void tcg_out_movi(TCGContext *s, TCGType type, 851 TCGReg ret, tcg_target_long sval) 852{ 853 tcg_target_ulong uval = sval; 854 ptrdiff_t pc_off; 855 int i; 856 857 if (type == TCG_TYPE_I32) { 858 uval = (uint32_t)sval; 859 sval = (int32_t)sval; 860 } 861 862 /* Try all 32-bit insns that can load it in one go. */ 863 if (sval >= -0x8000 && sval < 0x8000) { 864 tcg_out_insn(s, RI, LGHI, ret, sval); 865 return; 866 } 867 868 i = is_const_p16(uval); 869 if (i >= 0) { 870 tcg_out_insn_RI(s, li_insns[i], ret, uval >> (i * 16)); 871 return; 872 } 873 874 /* Try all 48-bit insns that can load it in one go. */ 875 if (sval == (int32_t)sval) { 876 tcg_out_insn(s, RIL, LGFI, ret, sval); 877 return; 878 } 879 880 i = is_const_p32(uval); 881 if (i >= 0) { 882 tcg_out_insn_RIL(s, lif_insns[i], ret, uval >> (i * 32)); 883 return; 884 } 885 886 /* Try for PC-relative address load. For odd addresses, add one. */ 887 pc_off = tcg_pcrel_diff(s, (void *)sval) >> 1; 888 if (pc_off == (int32_t)pc_off) { 889 tcg_out_insn(s, RIL, LARL, ret, pc_off); 890 if (sval & 1) { 891 tcg_out_insn(s, RI, AGHI, ret, 1); 892 } 893 return; 894 } 895 896 /* Otherwise, load it by parts. */ 897 i = is_const_p16((uint32_t)uval); 898 if (i >= 0) { 899 tcg_out_insn_RI(s, li_insns[i], ret, uval >> (i * 16)); 900 } else { 901 tcg_out_insn(s, RIL, LLILF, ret, uval); 902 } 903 uval >>= 32; 904 i = is_const_p16(uval); 905 if (i >= 0) { 906 tcg_out_insn_RI(s, oi_insns[i + 2], ret, uval >> (i * 16)); 907 } else { 908 tcg_out_insn(s, RIL, OIHF, ret, uval); 909 } 910} 911 912/* Emit a load/store type instruction. Inputs are: 913 DATA: The register to be loaded or stored. 914 BASE+OFS: The effective address. 915 OPC_RX: If the operation has an RX format opcode (e.g. STC), otherwise 0. 916 OPC_RXY: The RXY format opcode for the operation (e.g. STCY). */ 917 918static void tcg_out_mem(TCGContext *s, S390Opcode opc_rx, S390Opcode opc_rxy, 919 TCGReg data, TCGReg base, TCGReg index, 920 tcg_target_long ofs) 921{ 922 if (ofs < -0x80000 || ofs >= 0x80000) { 923 /* Combine the low 20 bits of the offset with the actual load insn; 924 the high 44 bits must come from an immediate load. */ 925 tcg_target_long low = ((ofs & 0xfffff) ^ 0x80000) - 0x80000; 926 tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, ofs - low); 927 ofs = low; 928 929 /* If we were already given an index register, add it in. */ 930 if (index != TCG_REG_NONE) { 931 tcg_out_insn(s, RRE, AGR, TCG_TMP0, index); 932 } 933 index = TCG_TMP0; 934 } 935 936 if (opc_rx && ofs >= 0 && ofs < 0x1000) { 937 tcg_out_insn_RX(s, opc_rx, data, base, index, ofs); 938 } else { 939 tcg_out_insn_RXY(s, opc_rxy, data, base, index, ofs); 940 } 941} 942 943static void tcg_out_vrx_mem(TCGContext *s, S390Opcode opc_vrx, 944 TCGReg data, TCGReg base, TCGReg index, 945 tcg_target_long ofs, int m3) 946{ 947 if (ofs < 0 || ofs >= 0x1000) { 948 if (ofs >= -0x80000 && ofs < 0x80000) { 949 tcg_out_insn(s, RXY, LAY, TCG_TMP0, base, index, ofs); 950 base = TCG_TMP0; 951 index = TCG_REG_NONE; 952 ofs = 0; 953 } else { 954 tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, ofs); 955 if (index != TCG_REG_NONE) { 956 tcg_out_insn(s, RRE, AGR, TCG_TMP0, index); 957 } 958 index = TCG_TMP0; 959 ofs = 0; 960 } 961 } 962 tcg_out_insn_VRX(s, opc_vrx, data, base, index, ofs, m3); 963} 964 965/* load data without address translation or endianness conversion */ 966static void tcg_out_ld(TCGContext *s, TCGType type, TCGReg data, 967 TCGReg base, intptr_t ofs) 968{ 969 switch (type) { 970 case TCG_TYPE_I32: 971 if (likely(is_general_reg(data))) { 972 tcg_out_mem(s, RX_L, RXY_LY, data, base, TCG_REG_NONE, ofs); 973 break; 974 } 975 tcg_out_vrx_mem(s, VRX_VLLEZ, data, base, TCG_REG_NONE, ofs, MO_32); 976 break; 977 978 case TCG_TYPE_I64: 979 if (likely(is_general_reg(data))) { 980 tcg_out_mem(s, 0, RXY_LG, data, base, TCG_REG_NONE, ofs); 981 break; 982 } 983 /* fallthru */ 984 985 case TCG_TYPE_V64: 986 tcg_out_vrx_mem(s, VRX_VLLEZ, data, base, TCG_REG_NONE, ofs, MO_64); 987 break; 988 989 case TCG_TYPE_V128: 990 /* Hint quadword aligned. */ 991 tcg_out_vrx_mem(s, VRX_VL, data, base, TCG_REG_NONE, ofs, 4); 992 break; 993 994 default: 995 g_assert_not_reached(); 996 } 997} 998 999static void tcg_out_st(TCGContext *s, TCGType type, TCGReg data, 1000 TCGReg base, intptr_t ofs) 1001{ 1002 switch (type) { 1003 case TCG_TYPE_I32: 1004 if (likely(is_general_reg(data))) { 1005 tcg_out_mem(s, RX_ST, RXY_STY, data, base, TCG_REG_NONE, ofs); 1006 } else { 1007 tcg_out_vrx_mem(s, VRX_VSTEF, data, base, TCG_REG_NONE, ofs, 1); 1008 } 1009 break; 1010 1011 case TCG_TYPE_I64: 1012 if (likely(is_general_reg(data))) { 1013 tcg_out_mem(s, 0, RXY_STG, data, base, TCG_REG_NONE, ofs); 1014 break; 1015 } 1016 /* fallthru */ 1017 1018 case TCG_TYPE_V64: 1019 tcg_out_vrx_mem(s, VRX_VSTEG, data, base, TCG_REG_NONE, ofs, 0); 1020 break; 1021 1022 case TCG_TYPE_V128: 1023 /* Hint quadword aligned. */ 1024 tcg_out_vrx_mem(s, VRX_VST, data, base, TCG_REG_NONE, ofs, 4); 1025 break; 1026 1027 default: 1028 g_assert_not_reached(); 1029 } 1030} 1031 1032static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val, 1033 TCGReg base, intptr_t ofs) 1034{ 1035 return false; 1036} 1037 1038static bool tcg_out_xchg(TCGContext *s, TCGType type, TCGReg r1, TCGReg r2) 1039{ 1040 return false; 1041} 1042 1043static void tcg_out_addi_ptr(TCGContext *s, TCGReg rd, TCGReg rs, 1044 tcg_target_long imm) 1045{ 1046 /* This function is only used for passing structs by reference. */ 1047 tcg_out_mem(s, RX_LA, RXY_LAY, rd, rs, TCG_REG_NONE, imm); 1048} 1049 1050static inline void tcg_out_risbg(TCGContext *s, TCGReg dest, TCGReg src, 1051 int msb, int lsb, int ofs, int z) 1052{ 1053 /* Format RIE-f */ 1054 tcg_out16(s, (RIEf_RISBG & 0xff00) | (dest << 4) | src); 1055 tcg_out16(s, (msb << 8) | (z << 7) | lsb); 1056 tcg_out16(s, (ofs << 8) | (RIEf_RISBG & 0xff)); 1057} 1058 1059static void tcg_out_ext8s(TCGContext *s, TCGType type, TCGReg dest, TCGReg src) 1060{ 1061 tcg_out_insn(s, RRE, LGBR, dest, src); 1062} 1063 1064static void tcg_out_ext8u(TCGContext *s, TCGReg dest, TCGReg src) 1065{ 1066 tcg_out_insn(s, RRE, LLGCR, dest, src); 1067} 1068 1069static void tcg_out_ext16s(TCGContext *s, TCGType type, TCGReg dest, TCGReg src) 1070{ 1071 tcg_out_insn(s, RRE, LGHR, dest, src); 1072} 1073 1074static void tcg_out_ext16u(TCGContext *s, TCGReg dest, TCGReg src) 1075{ 1076 tcg_out_insn(s, RRE, LLGHR, dest, src); 1077} 1078 1079static void tcg_out_ext32s(TCGContext *s, TCGReg dest, TCGReg src) 1080{ 1081 tcg_out_insn(s, RRE, LGFR, dest, src); 1082} 1083 1084static void tcg_out_ext32u(TCGContext *s, TCGReg dest, TCGReg src) 1085{ 1086 tcg_out_insn(s, RRE, LLGFR, dest, src); 1087} 1088 1089static void tcg_out_exts_i32_i64(TCGContext *s, TCGReg dest, TCGReg src) 1090{ 1091 tcg_out_ext32s(s, dest, src); 1092} 1093 1094static void tcg_out_extu_i32_i64(TCGContext *s, TCGReg dest, TCGReg src) 1095{ 1096 tcg_out_ext32u(s, dest, src); 1097} 1098 1099static void tcg_out_extrl_i64_i32(TCGContext *s, TCGReg dest, TCGReg src) 1100{ 1101 tcg_out_mov(s, TCG_TYPE_I32, dest, src); 1102} 1103 1104static void tgen_andi_risbg(TCGContext *s, TCGReg out, TCGReg in, uint64_t val) 1105{ 1106 int msb, lsb; 1107 if ((val & 0x8000000000000001ull) == 0x8000000000000001ull) { 1108 /* Achieve wraparound by swapping msb and lsb. */ 1109 msb = 64 - ctz64(~val); 1110 lsb = clz64(~val) - 1; 1111 } else { 1112 msb = clz64(val); 1113 lsb = 63 - ctz64(val); 1114 } 1115 tcg_out_risbg(s, out, in, msb, lsb, 0, 1); 1116} 1117 1118static void tgen_andi(TCGContext *s, TCGType type, TCGReg dest, uint64_t val) 1119{ 1120 static const S390Opcode ni_insns[4] = { 1121 RI_NILL, RI_NILH, RI_NIHL, RI_NIHH 1122 }; 1123 static const S390Opcode nif_insns[2] = { 1124 RIL_NILF, RIL_NIHF 1125 }; 1126 uint64_t valid = (type == TCG_TYPE_I32 ? 0xffffffffull : -1ull); 1127 int i; 1128 1129 /* Look for the zero-extensions. */ 1130 if ((val & valid) == 0xffffffff) { 1131 tcg_out_ext32u(s, dest, dest); 1132 return; 1133 } 1134 if ((val & valid) == 0xff) { 1135 tcg_out_ext8u(s, dest, dest); 1136 return; 1137 } 1138 if ((val & valid) == 0xffff) { 1139 tcg_out_ext16u(s, dest, dest); 1140 return; 1141 } 1142 1143 i = is_const_p16(~val & valid); 1144 if (i >= 0) { 1145 tcg_out_insn_RI(s, ni_insns[i], dest, val >> (i * 16)); 1146 return; 1147 } 1148 1149 i = is_const_p32(~val & valid); 1150 tcg_debug_assert(i == 0 || type != TCG_TYPE_I32); 1151 if (i >= 0) { 1152 tcg_out_insn_RIL(s, nif_insns[i], dest, val >> (i * 32)); 1153 return; 1154 } 1155 1156 if (risbg_mask(val)) { 1157 tgen_andi_risbg(s, dest, dest, val); 1158 return; 1159 } 1160 1161 g_assert_not_reached(); 1162} 1163 1164static void tgen_ori(TCGContext *s, TCGReg dest, uint64_t val) 1165{ 1166 static const S390Opcode oif_insns[2] = { 1167 RIL_OILF, RIL_OIHF 1168 }; 1169 1170 int i; 1171 1172 i = is_const_p16(val); 1173 if (i >= 0) { 1174 tcg_out_insn_RI(s, oi_insns[i], dest, val >> (i * 16)); 1175 return; 1176 } 1177 1178 i = is_const_p32(val); 1179 if (i >= 0) { 1180 tcg_out_insn_RIL(s, oif_insns[i], dest, val >> (i * 32)); 1181 return; 1182 } 1183 1184 g_assert_not_reached(); 1185} 1186 1187static void tgen_xori(TCGContext *s, TCGReg dest, uint64_t val) 1188{ 1189 switch (is_const_p32(val)) { 1190 case 0: 1191 tcg_out_insn(s, RIL, XILF, dest, val); 1192 break; 1193 case 1: 1194 tcg_out_insn(s, RIL, XIHF, dest, val >> 32); 1195 break; 1196 default: 1197 g_assert_not_reached(); 1198 } 1199} 1200 1201static int tgen_cmp2(TCGContext *s, TCGType type, TCGCond c, TCGReg r1, 1202 TCGArg c2, bool c2const, bool need_carry, int *inv_cc) 1203{ 1204 bool is_unsigned = is_unsigned_cond(c); 1205 TCGCond inv_c = tcg_invert_cond(c); 1206 S390Opcode op; 1207 1208 if (c2const) { 1209 if (c2 == 0) { 1210 if (!(is_unsigned && need_carry)) { 1211 if (type == TCG_TYPE_I32) { 1212 tcg_out_insn(s, RR, LTR, r1, r1); 1213 } else { 1214 tcg_out_insn(s, RRE, LTGR, r1, r1); 1215 } 1216 *inv_cc = tcg_cond_to_ltr_cond[inv_c]; 1217 return tcg_cond_to_ltr_cond[c]; 1218 } 1219 } 1220 1221 if (!is_unsigned && c2 == (int16_t)c2) { 1222 op = (type == TCG_TYPE_I32 ? RI_CHI : RI_CGHI); 1223 tcg_out_insn_RI(s, op, r1, c2); 1224 goto exit; 1225 } 1226 1227 if (type == TCG_TYPE_I32) { 1228 op = (is_unsigned ? RIL_CLFI : RIL_CFI); 1229 tcg_out_insn_RIL(s, op, r1, c2); 1230 goto exit; 1231 } 1232 1233 /* 1234 * Constraints are for a signed 33-bit operand, which is a 1235 * convenient superset of this signed/unsigned test. 1236 */ 1237 if (c2 == (is_unsigned ? (TCGArg)(uint32_t)c2 : (TCGArg)(int32_t)c2)) { 1238 op = (is_unsigned ? RIL_CLGFI : RIL_CGFI); 1239 tcg_out_insn_RIL(s, op, r1, c2); 1240 goto exit; 1241 } 1242 1243 /* Load everything else into a register. */ 1244 tcg_out_movi(s, TCG_TYPE_I64, TCG_TMP0, c2); 1245 c2 = TCG_TMP0; 1246 } 1247 1248 if (type == TCG_TYPE_I32) { 1249 op = (is_unsigned ? RR_CLR : RR_CR); 1250 tcg_out_insn_RR(s, op, r1, c2); 1251 } else { 1252 op = (is_unsigned ? RRE_CLGR : RRE_CGR); 1253 tcg_out_insn_RRE(s, op, r1, c2); 1254 } 1255 1256 exit: 1257 *inv_cc = tcg_cond_to_s390_cond[inv_c]; 1258 return tcg_cond_to_s390_cond[c]; 1259} 1260 1261static int tgen_cmp(TCGContext *s, TCGType type, TCGCond c, TCGReg r1, 1262 TCGArg c2, bool c2const, bool need_carry) 1263{ 1264 int inv_cc; 1265 return tgen_cmp2(s, type, c, r1, c2, c2const, need_carry, &inv_cc); 1266} 1267 1268static void tgen_setcond(TCGContext *s, TCGType type, TCGCond cond, 1269 TCGReg dest, TCGReg c1, TCGArg c2, int c2const) 1270{ 1271 int cc; 1272 1273 /* With LOC2, we can always emit the minimum 3 insns. */ 1274 if (HAVE_FACILITY(LOAD_ON_COND2)) { 1275 /* Emit: d = 0, d = (cc ? 1 : d). */ 1276 cc = tgen_cmp(s, type, cond, c1, c2, c2const, false); 1277 tcg_out_movi(s, TCG_TYPE_I64, dest, 0); 1278 tcg_out_insn(s, RIEg, LOCGHI, dest, 1, cc); 1279 return; 1280 } 1281 1282 restart: 1283 switch (cond) { 1284 case TCG_COND_NE: 1285 /* X != 0 is X > 0. */ 1286 if (c2const && c2 == 0) { 1287 cond = TCG_COND_GTU; 1288 } else { 1289 break; 1290 } 1291 /* fallthru */ 1292 1293 case TCG_COND_GTU: 1294 case TCG_COND_GT: 1295 /* The result of a compare has CC=2 for GT and CC=3 unused. 1296 ADD LOGICAL WITH CARRY considers (CC & 2) the carry bit. */ 1297 tgen_cmp(s, type, cond, c1, c2, c2const, true); 1298 tcg_out_movi(s, type, dest, 0); 1299 tcg_out_insn(s, RRE, ALCGR, dest, dest); 1300 return; 1301 1302 case TCG_COND_EQ: 1303 /* X == 0 is X <= 0. */ 1304 if (c2const && c2 == 0) { 1305 cond = TCG_COND_LEU; 1306 } else { 1307 break; 1308 } 1309 /* fallthru */ 1310 1311 case TCG_COND_LEU: 1312 case TCG_COND_LE: 1313 /* As above, but we're looking for borrow, or !carry. 1314 The second insn computes d - d - borrow, or -1 for true 1315 and 0 for false. So we must mask to 1 bit afterward. */ 1316 tgen_cmp(s, type, cond, c1, c2, c2const, true); 1317 tcg_out_insn(s, RRE, SLBGR, dest, dest); 1318 tgen_andi(s, type, dest, 1); 1319 return; 1320 1321 case TCG_COND_GEU: 1322 case TCG_COND_LTU: 1323 case TCG_COND_LT: 1324 case TCG_COND_GE: 1325 /* Swap operands so that we can use LEU/GTU/GT/LE. */ 1326 if (!c2const) { 1327 TCGReg t = c1; 1328 c1 = c2; 1329 c2 = t; 1330 cond = tcg_swap_cond(cond); 1331 goto restart; 1332 } 1333 break; 1334 1335 default: 1336 g_assert_not_reached(); 1337 } 1338 1339 cc = tgen_cmp(s, type, cond, c1, c2, c2const, false); 1340 /* Emit: d = 0, t = 1, d = (cc ? t : d). */ 1341 tcg_out_movi(s, TCG_TYPE_I64, dest, 0); 1342 tcg_out_movi(s, TCG_TYPE_I64, TCG_TMP0, 1); 1343 tcg_out_insn(s, RRFc, LOCGR, dest, TCG_TMP0, cc); 1344} 1345 1346static void tgen_movcond_int(TCGContext *s, TCGType type, TCGReg dest, 1347 TCGArg v3, int v3const, TCGReg v4, 1348 int cc, int inv_cc) 1349{ 1350 TCGReg src; 1351 1352 if (v3const) { 1353 if (dest == v4) { 1354 if (HAVE_FACILITY(LOAD_ON_COND2)) { 1355 /* Emit: if (cc) dest = v3. */ 1356 tcg_out_insn(s, RIEg, LOCGHI, dest, v3, cc); 1357 return; 1358 } 1359 tcg_out_insn(s, RI, LGHI, TCG_TMP0, v3); 1360 src = TCG_TMP0; 1361 } else { 1362 /* LGR+LOCGHI is larger than LGHI+LOCGR. */ 1363 tcg_out_insn(s, RI, LGHI, dest, v3); 1364 cc = inv_cc; 1365 src = v4; 1366 } 1367 } else { 1368 if (HAVE_FACILITY(MISC_INSN_EXT3)) { 1369 /* Emit: dest = cc ? v3 : v4. */ 1370 tcg_out_insn(s, RRFam, SELGR, dest, v3, v4, cc); 1371 return; 1372 } 1373 if (dest == v4) { 1374 src = v3; 1375 } else { 1376 tcg_out_mov(s, type, dest, v3); 1377 cc = inv_cc; 1378 src = v4; 1379 } 1380 } 1381 1382 /* Emit: if (cc) dest = src. */ 1383 tcg_out_insn(s, RRFc, LOCGR, dest, src, cc); 1384} 1385 1386static void tgen_movcond(TCGContext *s, TCGType type, TCGCond c, TCGReg dest, 1387 TCGReg c1, TCGArg c2, int c2const, 1388 TCGArg v3, int v3const, TCGReg v4) 1389{ 1390 int cc, inv_cc; 1391 1392 cc = tgen_cmp2(s, type, c, c1, c2, c2const, false, &inv_cc); 1393 tgen_movcond_int(s, type, dest, v3, v3const, v4, cc, inv_cc); 1394} 1395 1396static void tgen_clz(TCGContext *s, TCGReg dest, TCGReg a1, 1397 TCGArg a2, int a2const) 1398{ 1399 /* Since this sets both R and R+1, we have no choice but to store the 1400 result into R0, allowing R1 == TCG_TMP0 to be clobbered as well. */ 1401 QEMU_BUILD_BUG_ON(TCG_TMP0 != TCG_REG_R1); 1402 tcg_out_insn(s, RRE, FLOGR, TCG_REG_R0, a1); 1403 1404 if (a2const && a2 == 64) { 1405 tcg_out_mov(s, TCG_TYPE_I64, dest, TCG_REG_R0); 1406 return; 1407 } 1408 1409 /* 1410 * Conditions from FLOGR are: 1411 * 2 -> one bit found 1412 * 8 -> no one bit found 1413 */ 1414 tgen_movcond_int(s, TCG_TYPE_I64, dest, a2, a2const, TCG_REG_R0, 8, 2); 1415} 1416 1417static void tgen_ctpop(TCGContext *s, TCGType type, TCGReg dest, TCGReg src) 1418{ 1419 /* With MIE3, and bit 0 of m4 set, we get the complete result. */ 1420 if (HAVE_FACILITY(MISC_INSN_EXT3)) { 1421 if (type == TCG_TYPE_I32) { 1422 tcg_out_ext32u(s, dest, src); 1423 src = dest; 1424 } 1425 tcg_out_insn(s, RRFc, POPCNT, dest, src, 8); 1426 return; 1427 } 1428 1429 /* Without MIE3, each byte gets the count of bits for the byte. */ 1430 tcg_out_insn(s, RRFc, POPCNT, dest, src, 0); 1431 1432 /* Multiply to sum each byte at the top of the word. */ 1433 if (type == TCG_TYPE_I32) { 1434 tcg_out_insn(s, RIL, MSFI, dest, 0x01010101); 1435 tcg_out_sh32(s, RS_SRL, dest, TCG_REG_NONE, 24); 1436 } else { 1437 tcg_out_movi(s, TCG_TYPE_I64, TCG_TMP0, 0x0101010101010101ull); 1438 tcg_out_insn(s, RRE, MSGR, dest, TCG_TMP0); 1439 tcg_out_sh64(s, RSY_SRLG, dest, dest, TCG_REG_NONE, 56); 1440 } 1441} 1442 1443static void tgen_deposit(TCGContext *s, TCGReg dest, TCGReg src, 1444 int ofs, int len, int z) 1445{ 1446 int lsb = (63 - ofs); 1447 int msb = lsb - (len - 1); 1448 tcg_out_risbg(s, dest, src, msb, lsb, ofs, z); 1449} 1450 1451static void tgen_extract(TCGContext *s, TCGReg dest, TCGReg src, 1452 int ofs, int len) 1453{ 1454 tcg_out_risbg(s, dest, src, 64 - len, 63, 64 - ofs, 1); 1455} 1456 1457static void tgen_gotoi(TCGContext *s, int cc, const tcg_insn_unit *dest) 1458{ 1459 ptrdiff_t off = tcg_pcrel_diff(s, dest) >> 1; 1460 if (off == (int16_t)off) { 1461 tcg_out_insn(s, RI, BRC, cc, off); 1462 } else if (off == (int32_t)off) { 1463 tcg_out_insn(s, RIL, BRCL, cc, off); 1464 } else { 1465 tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, (uintptr_t)dest); 1466 tcg_out_insn(s, RR, BCR, cc, TCG_TMP0); 1467 } 1468} 1469 1470static void tgen_branch(TCGContext *s, int cc, TCGLabel *l) 1471{ 1472 if (l->has_value) { 1473 tgen_gotoi(s, cc, l->u.value_ptr); 1474 } else { 1475 tcg_out16(s, RI_BRC | (cc << 4)); 1476 tcg_out_reloc(s, s->code_ptr, R_390_PC16DBL, l, 2); 1477 s->code_ptr += 1; 1478 } 1479} 1480 1481static void tgen_compare_branch(TCGContext *s, S390Opcode opc, int cc, 1482 TCGReg r1, TCGReg r2, TCGLabel *l) 1483{ 1484 tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, l, 2); 1485 /* Format RIE-b */ 1486 tcg_out16(s, (opc & 0xff00) | (r1 << 4) | r2); 1487 tcg_out16(s, 0); 1488 tcg_out16(s, cc << 12 | (opc & 0xff)); 1489} 1490 1491static void tgen_compare_imm_branch(TCGContext *s, S390Opcode opc, int cc, 1492 TCGReg r1, int i2, TCGLabel *l) 1493{ 1494 tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, l, 2); 1495 /* Format RIE-c */ 1496 tcg_out16(s, (opc & 0xff00) | (r1 << 4) | cc); 1497 tcg_out16(s, 0); 1498 tcg_out16(s, (i2 << 8) | (opc & 0xff)); 1499} 1500 1501static void tgen_brcond(TCGContext *s, TCGType type, TCGCond c, 1502 TCGReg r1, TCGArg c2, int c2const, TCGLabel *l) 1503{ 1504 int cc; 1505 bool is_unsigned = is_unsigned_cond(c); 1506 bool in_range; 1507 S390Opcode opc; 1508 1509 cc = tcg_cond_to_s390_cond[c]; 1510 1511 if (!c2const) { 1512 opc = (type == TCG_TYPE_I32 1513 ? (is_unsigned ? RIEb_CLRJ : RIEb_CRJ) 1514 : (is_unsigned ? RIEb_CLGRJ : RIEb_CGRJ)); 1515 tgen_compare_branch(s, opc, cc, r1, c2, l); 1516 return; 1517 } 1518 1519 /* 1520 * COMPARE IMMEDIATE AND BRANCH RELATIVE has an 8-bit immediate field. 1521 * If the immediate we've been given does not fit that range, we'll 1522 * fall back to separate compare and branch instructions using the 1523 * larger comparison range afforded by COMPARE IMMEDIATE. 1524 */ 1525 if (type == TCG_TYPE_I32) { 1526 if (is_unsigned) { 1527 opc = RIEc_CLIJ; 1528 in_range = (uint32_t)c2 == (uint8_t)c2; 1529 } else { 1530 opc = RIEc_CIJ; 1531 in_range = (int32_t)c2 == (int8_t)c2; 1532 } 1533 } else { 1534 if (is_unsigned) { 1535 opc = RIEc_CLGIJ; 1536 in_range = (uint64_t)c2 == (uint8_t)c2; 1537 } else { 1538 opc = RIEc_CGIJ; 1539 in_range = (int64_t)c2 == (int8_t)c2; 1540 } 1541 } 1542 if (in_range) { 1543 tgen_compare_imm_branch(s, opc, cc, r1, c2, l); 1544 return; 1545 } 1546 1547 cc = tgen_cmp(s, type, c, r1, c2, c2const, false); 1548 tgen_branch(s, cc, l); 1549} 1550 1551static void tcg_out_call_int(TCGContext *s, const tcg_insn_unit *dest) 1552{ 1553 ptrdiff_t off = tcg_pcrel_diff(s, dest) >> 1; 1554 if (off == (int32_t)off) { 1555 tcg_out_insn(s, RIL, BRASL, TCG_REG_R14, off); 1556 } else { 1557 tcg_out_movi(s, TCG_TYPE_PTR, TCG_TMP0, (uintptr_t)dest); 1558 tcg_out_insn(s, RR, BASR, TCG_REG_R14, TCG_TMP0); 1559 } 1560} 1561 1562static void tcg_out_call(TCGContext *s, const tcg_insn_unit *dest, 1563 const TCGHelperInfo *info) 1564{ 1565 tcg_out_call_int(s, dest); 1566} 1567 1568typedef struct { 1569 TCGReg base; 1570 TCGReg index; 1571 int disp; 1572 TCGAtomAlign aa; 1573} HostAddress; 1574 1575bool tcg_target_has_memory_bswap(MemOp memop) 1576{ 1577 TCGAtomAlign aa; 1578 1579 if ((memop & MO_SIZE) <= MO_64) { 1580 return true; 1581 } 1582 1583 /* 1584 * Reject 16-byte memop with 16-byte atomicity, 1585 * but do allow a pair of 64-bit operations. 1586 */ 1587 aa = atom_and_align_for_opc(tcg_ctx, memop, MO_ATOM_IFALIGN, true); 1588 return aa.atom <= MO_64; 1589} 1590 1591static void tcg_out_qemu_ld_direct(TCGContext *s, MemOp opc, TCGReg data, 1592 HostAddress h) 1593{ 1594 switch (opc & (MO_SSIZE | MO_BSWAP)) { 1595 case MO_UB: 1596 tcg_out_insn(s, RXY, LLGC, data, h.base, h.index, h.disp); 1597 break; 1598 case MO_SB: 1599 tcg_out_insn(s, RXY, LGB, data, h.base, h.index, h.disp); 1600 break; 1601 1602 case MO_UW | MO_BSWAP: 1603 /* swapped unsigned halfword load with upper bits zeroed */ 1604 tcg_out_insn(s, RXY, LRVH, data, h.base, h.index, h.disp); 1605 tcg_out_ext16u(s, data, data); 1606 break; 1607 case MO_UW: 1608 tcg_out_insn(s, RXY, LLGH, data, h.base, h.index, h.disp); 1609 break; 1610 1611 case MO_SW | MO_BSWAP: 1612 /* swapped sign-extended halfword load */ 1613 tcg_out_insn(s, RXY, LRVH, data, h.base, h.index, h.disp); 1614 tcg_out_ext16s(s, TCG_TYPE_REG, data, data); 1615 break; 1616 case MO_SW: 1617 tcg_out_insn(s, RXY, LGH, data, h.base, h.index, h.disp); 1618 break; 1619 1620 case MO_UL | MO_BSWAP: 1621 /* swapped unsigned int load with upper bits zeroed */ 1622 tcg_out_insn(s, RXY, LRV, data, h.base, h.index, h.disp); 1623 tcg_out_ext32u(s, data, data); 1624 break; 1625 case MO_UL: 1626 tcg_out_insn(s, RXY, LLGF, data, h.base, h.index, h.disp); 1627 break; 1628 1629 case MO_SL | MO_BSWAP: 1630 /* swapped sign-extended int load */ 1631 tcg_out_insn(s, RXY, LRV, data, h.base, h.index, h.disp); 1632 tcg_out_ext32s(s, data, data); 1633 break; 1634 case MO_SL: 1635 tcg_out_insn(s, RXY, LGF, data, h.base, h.index, h.disp); 1636 break; 1637 1638 case MO_UQ | MO_BSWAP: 1639 tcg_out_insn(s, RXY, LRVG, data, h.base, h.index, h.disp); 1640 break; 1641 case MO_UQ: 1642 tcg_out_insn(s, RXY, LG, data, h.base, h.index, h.disp); 1643 break; 1644 1645 default: 1646 g_assert_not_reached(); 1647 } 1648} 1649 1650static void tcg_out_qemu_st_direct(TCGContext *s, MemOp opc, TCGReg data, 1651 HostAddress h) 1652{ 1653 switch (opc & (MO_SIZE | MO_BSWAP)) { 1654 case MO_UB: 1655 if (h.disp >= 0 && h.disp < 0x1000) { 1656 tcg_out_insn(s, RX, STC, data, h.base, h.index, h.disp); 1657 } else { 1658 tcg_out_insn(s, RXY, STCY, data, h.base, h.index, h.disp); 1659 } 1660 break; 1661 1662 case MO_UW | MO_BSWAP: 1663 tcg_out_insn(s, RXY, STRVH, data, h.base, h.index, h.disp); 1664 break; 1665 case MO_UW: 1666 if (h.disp >= 0 && h.disp < 0x1000) { 1667 tcg_out_insn(s, RX, STH, data, h.base, h.index, h.disp); 1668 } else { 1669 tcg_out_insn(s, RXY, STHY, data, h.base, h.index, h.disp); 1670 } 1671 break; 1672 1673 case MO_UL | MO_BSWAP: 1674 tcg_out_insn(s, RXY, STRV, data, h.base, h.index, h.disp); 1675 break; 1676 case MO_UL: 1677 if (h.disp >= 0 && h.disp < 0x1000) { 1678 tcg_out_insn(s, RX, ST, data, h.base, h.index, h.disp); 1679 } else { 1680 tcg_out_insn(s, RXY, STY, data, h.base, h.index, h.disp); 1681 } 1682 break; 1683 1684 case MO_UQ | MO_BSWAP: 1685 tcg_out_insn(s, RXY, STRVG, data, h.base, h.index, h.disp); 1686 break; 1687 case MO_UQ: 1688 tcg_out_insn(s, RXY, STG, data, h.base, h.index, h.disp); 1689 break; 1690 1691 default: 1692 g_assert_not_reached(); 1693 } 1694} 1695 1696static const TCGLdstHelperParam ldst_helper_param = { 1697 .ntmp = 1, .tmp = { TCG_TMP0 } 1698}; 1699 1700static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) 1701{ 1702 MemOp opc = get_memop(lb->oi); 1703 1704 if (!patch_reloc(lb->label_ptr[0], R_390_PC16DBL, 1705 (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 2)) { 1706 return false; 1707 } 1708 1709 tcg_out_ld_helper_args(s, lb, &ldst_helper_param); 1710 tcg_out_call_int(s, qemu_ld_helpers[opc & MO_SIZE]); 1711 tcg_out_ld_helper_ret(s, lb, false, &ldst_helper_param); 1712 1713 tgen_gotoi(s, S390_CC_ALWAYS, lb->raddr); 1714 return true; 1715} 1716 1717static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb) 1718{ 1719 MemOp opc = get_memop(lb->oi); 1720 1721 if (!patch_reloc(lb->label_ptr[0], R_390_PC16DBL, 1722 (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 2)) { 1723 return false; 1724 } 1725 1726 tcg_out_st_helper_args(s, lb, &ldst_helper_param); 1727 tcg_out_call_int(s, qemu_st_helpers[opc & MO_SIZE]); 1728 1729 tgen_gotoi(s, S390_CC_ALWAYS, lb->raddr); 1730 return true; 1731} 1732 1733/* We're expecting to use a 20-bit negative offset on the tlb memory ops. */ 1734#define MIN_TLB_MASK_TABLE_OFS -(1 << 19) 1735 1736/* 1737 * For softmmu, perform the TLB load and compare. 1738 * For useronly, perform any required alignment tests. 1739 * In both cases, return a TCGLabelQemuLdst structure if the slow path 1740 * is required and fill in @h with the host address for the fast path. 1741 */ 1742static TCGLabelQemuLdst *prepare_host_addr(TCGContext *s, HostAddress *h, 1743 TCGReg addr_reg, MemOpIdx oi, 1744 bool is_ld) 1745{ 1746 TCGType addr_type = s->addr_type; 1747 TCGLabelQemuLdst *ldst = NULL; 1748 MemOp opc = get_memop(oi); 1749 MemOp s_bits = opc & MO_SIZE; 1750 unsigned a_mask; 1751 1752 h->aa = atom_and_align_for_opc(s, opc, MO_ATOM_IFALIGN, s_bits == MO_128); 1753 a_mask = (1 << h->aa.align) - 1; 1754 1755#ifdef CONFIG_SOFTMMU 1756 unsigned s_mask = (1 << s_bits) - 1; 1757 int mem_index = get_mmuidx(oi); 1758 int fast_off = tlb_mask_table_ofs(s, mem_index); 1759 int mask_off = fast_off + offsetof(CPUTLBDescFast, mask); 1760 int table_off = fast_off + offsetof(CPUTLBDescFast, table); 1761 int ofs, a_off; 1762 uint64_t tlb_mask; 1763 1764 ldst = new_ldst_label(s); 1765 ldst->is_ld = is_ld; 1766 ldst->oi = oi; 1767 ldst->addrlo_reg = addr_reg; 1768 1769 tcg_out_sh64(s, RSY_SRLG, TCG_TMP0, addr_reg, TCG_REG_NONE, 1770 s->page_bits - CPU_TLB_ENTRY_BITS); 1771 1772 tcg_out_insn(s, RXY, NG, TCG_TMP0, TCG_AREG0, TCG_REG_NONE, mask_off); 1773 tcg_out_insn(s, RXY, AG, TCG_TMP0, TCG_AREG0, TCG_REG_NONE, table_off); 1774 1775 /* 1776 * For aligned accesses, we check the first byte and include the alignment 1777 * bits within the address. For unaligned access, we check that we don't 1778 * cross pages using the address of the last byte of the access. 1779 */ 1780 a_off = (a_mask >= s_mask ? 0 : s_mask - a_mask); 1781 tlb_mask = (uint64_t)s->page_mask | a_mask; 1782 if (a_off == 0) { 1783 tgen_andi_risbg(s, TCG_REG_R0, addr_reg, tlb_mask); 1784 } else { 1785 tcg_out_insn(s, RX, LA, TCG_REG_R0, addr_reg, TCG_REG_NONE, a_off); 1786 tgen_andi(s, addr_type, TCG_REG_R0, tlb_mask); 1787 } 1788 1789 if (is_ld) { 1790 ofs = offsetof(CPUTLBEntry, addr_read); 1791 } else { 1792 ofs = offsetof(CPUTLBEntry, addr_write); 1793 } 1794 if (addr_type == TCG_TYPE_I32) { 1795 ofs += HOST_BIG_ENDIAN * 4; 1796 tcg_out_insn(s, RX, C, TCG_REG_R0, TCG_TMP0, TCG_REG_NONE, ofs); 1797 } else { 1798 tcg_out_insn(s, RXY, CG, TCG_REG_R0, TCG_TMP0, TCG_REG_NONE, ofs); 1799 } 1800 1801 tcg_out16(s, RI_BRC | (S390_CC_NE << 4)); 1802 ldst->label_ptr[0] = s->code_ptr++; 1803 1804 h->index = TCG_TMP0; 1805 tcg_out_insn(s, RXY, LG, h->index, TCG_TMP0, TCG_REG_NONE, 1806 offsetof(CPUTLBEntry, addend)); 1807 1808 if (addr_type == TCG_TYPE_I32) { 1809 tcg_out_insn(s, RRE, ALGFR, h->index, addr_reg); 1810 h->base = TCG_REG_NONE; 1811 } else { 1812 h->base = addr_reg; 1813 } 1814 h->disp = 0; 1815#else 1816 if (a_mask) { 1817 ldst = new_ldst_label(s); 1818 ldst->is_ld = is_ld; 1819 ldst->oi = oi; 1820 ldst->addrlo_reg = addr_reg; 1821 1822 /* We are expecting a_bits to max out at 7, much lower than TMLL. */ 1823 tcg_debug_assert(a_mask <= 0xffff); 1824 tcg_out_insn(s, RI, TMLL, addr_reg, a_mask); 1825 1826 tcg_out16(s, RI_BRC | (7 << 4)); /* CC in {1,2,3} */ 1827 ldst->label_ptr[0] = s->code_ptr++; 1828 } 1829 1830 h->base = addr_reg; 1831 if (addr_type == TCG_TYPE_I32) { 1832 tcg_out_ext32u(s, TCG_TMP0, addr_reg); 1833 h->base = TCG_TMP0; 1834 } 1835 if (guest_base < 0x80000) { 1836 h->index = TCG_REG_NONE; 1837 h->disp = guest_base; 1838 } else { 1839 h->index = TCG_GUEST_BASE_REG; 1840 h->disp = 0; 1841 } 1842#endif 1843 1844 return ldst; 1845} 1846 1847static void tcg_out_qemu_ld(TCGContext* s, TCGReg data_reg, TCGReg addr_reg, 1848 MemOpIdx oi, TCGType data_type) 1849{ 1850 TCGLabelQemuLdst *ldst; 1851 HostAddress h; 1852 1853 ldst = prepare_host_addr(s, &h, addr_reg, oi, true); 1854 tcg_out_qemu_ld_direct(s, get_memop(oi), data_reg, h); 1855 1856 if (ldst) { 1857 ldst->type = data_type; 1858 ldst->datalo_reg = data_reg; 1859 ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); 1860 } 1861} 1862 1863static void tcg_out_qemu_st(TCGContext* s, TCGReg data_reg, TCGReg addr_reg, 1864 MemOpIdx oi, TCGType data_type) 1865{ 1866 TCGLabelQemuLdst *ldst; 1867 HostAddress h; 1868 1869 ldst = prepare_host_addr(s, &h, addr_reg, oi, false); 1870 tcg_out_qemu_st_direct(s, get_memop(oi), data_reg, h); 1871 1872 if (ldst) { 1873 ldst->type = data_type; 1874 ldst->datalo_reg = data_reg; 1875 ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); 1876 } 1877} 1878 1879static void tcg_out_qemu_ldst_i128(TCGContext *s, TCGReg datalo, TCGReg datahi, 1880 TCGReg addr_reg, MemOpIdx oi, bool is_ld) 1881{ 1882 TCGLabel *l1 = NULL, *l2 = NULL; 1883 TCGLabelQemuLdst *ldst; 1884 HostAddress h; 1885 bool need_bswap; 1886 bool use_pair; 1887 S390Opcode insn; 1888 1889 ldst = prepare_host_addr(s, &h, addr_reg, oi, is_ld); 1890 1891 use_pair = h.aa.atom < MO_128; 1892 need_bswap = get_memop(oi) & MO_BSWAP; 1893 1894 if (!use_pair) { 1895 /* 1896 * Atomicity requires we use LPQ. If we've already checked for 1897 * 16-byte alignment, that's all we need. If we arrive with 1898 * lesser alignment, we have determined that less than 16-byte 1899 * alignment can be satisfied with two 8-byte loads. 1900 */ 1901 if (h.aa.align < MO_128) { 1902 use_pair = true; 1903 l1 = gen_new_label(); 1904 l2 = gen_new_label(); 1905 1906 tcg_out_insn(s, RI, TMLL, addr_reg, 15); 1907 tgen_branch(s, 7, l1); /* CC in {1,2,3} */ 1908 } 1909 1910 tcg_debug_assert(!need_bswap); 1911 tcg_debug_assert(datalo & 1); 1912 tcg_debug_assert(datahi == datalo - 1); 1913 insn = is_ld ? RXY_LPQ : RXY_STPQ; 1914 tcg_out_insn_RXY(s, insn, datahi, h.base, h.index, h.disp); 1915 1916 if (use_pair) { 1917 tgen_branch(s, S390_CC_ALWAYS, l2); 1918 tcg_out_label(s, l1); 1919 } 1920 } 1921 if (use_pair) { 1922 TCGReg d1, d2; 1923 1924 if (need_bswap) { 1925 d1 = datalo, d2 = datahi; 1926 insn = is_ld ? RXY_LRVG : RXY_STRVG; 1927 } else { 1928 d1 = datahi, d2 = datalo; 1929 insn = is_ld ? RXY_LG : RXY_STG; 1930 } 1931 1932 if (h.base == d1 || h.index == d1) { 1933 tcg_out_insn(s, RXY, LAY, TCG_TMP0, h.base, h.index, h.disp); 1934 h.base = TCG_TMP0; 1935 h.index = TCG_REG_NONE; 1936 h.disp = 0; 1937 } 1938 tcg_out_insn_RXY(s, insn, d1, h.base, h.index, h.disp); 1939 tcg_out_insn_RXY(s, insn, d2, h.base, h.index, h.disp + 8); 1940 } 1941 if (l2) { 1942 tcg_out_label(s, l2); 1943 } 1944 1945 if (ldst) { 1946 ldst->type = TCG_TYPE_I128; 1947 ldst->datalo_reg = datalo; 1948 ldst->datahi_reg = datahi; 1949 ldst->raddr = tcg_splitwx_to_rx(s->code_ptr); 1950 } 1951} 1952 1953static void tcg_out_exit_tb(TCGContext *s, uintptr_t a0) 1954{ 1955 /* Reuse the zeroing that exists for goto_ptr. */ 1956 if (a0 == 0) { 1957 tgen_gotoi(s, S390_CC_ALWAYS, tcg_code_gen_epilogue); 1958 } else { 1959 tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, a0); 1960 tgen_gotoi(s, S390_CC_ALWAYS, tb_ret_addr); 1961 } 1962} 1963 1964static void tcg_out_goto_tb(TCGContext *s, int which) 1965{ 1966 /* 1967 * Branch displacement must be aligned for atomic patching; 1968 * see if we need to add extra nop before branch 1969 */ 1970 if (!QEMU_PTR_IS_ALIGNED(s->code_ptr + 1, 4)) { 1971 tcg_out16(s, NOP); 1972 } 1973 tcg_out16(s, RIL_BRCL | (S390_CC_ALWAYS << 4)); 1974 set_jmp_insn_offset(s, which); 1975 s->code_ptr += 2; 1976 set_jmp_reset_offset(s, which); 1977} 1978 1979void tb_target_set_jmp_target(const TranslationBlock *tb, int n, 1980 uintptr_t jmp_rx, uintptr_t jmp_rw) 1981{ 1982 if (!HAVE_FACILITY(GEN_INST_EXT)) { 1983 return; 1984 } 1985 /* patch the branch destination */ 1986 uintptr_t addr = tb->jmp_target_addr[n]; 1987 intptr_t disp = addr - (jmp_rx - 2); 1988 qatomic_set((int32_t *)jmp_rw, disp / 2); 1989 /* no need to flush icache explicitly */ 1990} 1991 1992# define OP_32_64(x) \ 1993 case glue(glue(INDEX_op_,x),_i32): \ 1994 case glue(glue(INDEX_op_,x),_i64) 1995 1996static inline void tcg_out_op(TCGContext *s, TCGOpcode opc, 1997 const TCGArg args[TCG_MAX_OP_ARGS], 1998 const int const_args[TCG_MAX_OP_ARGS]) 1999{ 2000 S390Opcode op, op2; 2001 TCGArg a0, a1, a2; 2002 2003 switch (opc) { 2004 case INDEX_op_goto_ptr: 2005 a0 = args[0]; 2006 tcg_out_insn(s, RR, BCR, S390_CC_ALWAYS, a0); 2007 break; 2008 2009 OP_32_64(ld8u): 2010 /* ??? LLC (RXY format) is only present with the extended-immediate 2011 facility, whereas LLGC is always present. */ 2012 tcg_out_mem(s, 0, RXY_LLGC, args[0], args[1], TCG_REG_NONE, args[2]); 2013 break; 2014 2015 OP_32_64(ld8s): 2016 /* ??? LB is no smaller than LGB, so no point to using it. */ 2017 tcg_out_mem(s, 0, RXY_LGB, args[0], args[1], TCG_REG_NONE, args[2]); 2018 break; 2019 2020 OP_32_64(ld16u): 2021 /* ??? LLH (RXY format) is only present with the extended-immediate 2022 facility, whereas LLGH is always present. */ 2023 tcg_out_mem(s, 0, RXY_LLGH, args[0], args[1], TCG_REG_NONE, args[2]); 2024 break; 2025 2026 case INDEX_op_ld16s_i32: 2027 tcg_out_mem(s, RX_LH, RXY_LHY, args[0], args[1], TCG_REG_NONE, args[2]); 2028 break; 2029 2030 case INDEX_op_ld_i32: 2031 tcg_out_ld(s, TCG_TYPE_I32, args[0], args[1], args[2]); 2032 break; 2033 2034 OP_32_64(st8): 2035 tcg_out_mem(s, RX_STC, RXY_STCY, args[0], args[1], 2036 TCG_REG_NONE, args[2]); 2037 break; 2038 2039 OP_32_64(st16): 2040 tcg_out_mem(s, RX_STH, RXY_STHY, args[0], args[1], 2041 TCG_REG_NONE, args[2]); 2042 break; 2043 2044 case INDEX_op_st_i32: 2045 tcg_out_st(s, TCG_TYPE_I32, args[0], args[1], args[2]); 2046 break; 2047 2048 case INDEX_op_add_i32: 2049 a0 = args[0], a1 = args[1], a2 = (int32_t)args[2]; 2050 if (const_args[2]) { 2051 do_addi_32: 2052 if (a0 == a1) { 2053 if (a2 == (int16_t)a2) { 2054 tcg_out_insn(s, RI, AHI, a0, a2); 2055 break; 2056 } 2057 tcg_out_insn(s, RIL, AFI, a0, a2); 2058 break; 2059 } 2060 tcg_out_mem(s, RX_LA, RXY_LAY, a0, a1, TCG_REG_NONE, a2); 2061 } else if (a0 == a1) { 2062 tcg_out_insn(s, RR, AR, a0, a2); 2063 } else { 2064 tcg_out_insn(s, RX, LA, a0, a1, a2, 0); 2065 } 2066 break; 2067 case INDEX_op_sub_i32: 2068 a0 = args[0], a1 = args[1], a2 = (int32_t)args[2]; 2069 if (const_args[2]) { 2070 a2 = -a2; 2071 goto do_addi_32; 2072 } else if (a0 == a1) { 2073 tcg_out_insn(s, RR, SR, a0, a2); 2074 } else { 2075 tcg_out_insn(s, RRFa, SRK, a0, a1, a2); 2076 } 2077 break; 2078 2079 case INDEX_op_and_i32: 2080 a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; 2081 if (const_args[2]) { 2082 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2083 tgen_andi(s, TCG_TYPE_I32, a0, a2); 2084 } else if (a0 == a1) { 2085 tcg_out_insn(s, RR, NR, a0, a2); 2086 } else { 2087 tcg_out_insn(s, RRFa, NRK, a0, a1, a2); 2088 } 2089 break; 2090 case INDEX_op_or_i32: 2091 a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; 2092 if (const_args[2]) { 2093 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2094 tgen_ori(s, a0, a2); 2095 } else if (a0 == a1) { 2096 tcg_out_insn(s, RR, OR, a0, a2); 2097 } else { 2098 tcg_out_insn(s, RRFa, ORK, a0, a1, a2); 2099 } 2100 break; 2101 case INDEX_op_xor_i32: 2102 a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; 2103 if (const_args[2]) { 2104 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2105 tcg_out_insn(s, RIL, XILF, a0, a2); 2106 } else if (a0 == a1) { 2107 tcg_out_insn(s, RR, XR, args[0], args[2]); 2108 } else { 2109 tcg_out_insn(s, RRFa, XRK, a0, a1, a2); 2110 } 2111 break; 2112 2113 case INDEX_op_andc_i32: 2114 a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; 2115 if (const_args[2]) { 2116 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2117 tgen_andi(s, TCG_TYPE_I32, a0, (uint32_t)~a2); 2118 } else { 2119 tcg_out_insn(s, RRFa, NCRK, a0, a1, a2); 2120 } 2121 break; 2122 case INDEX_op_orc_i32: 2123 a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; 2124 if (const_args[2]) { 2125 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2126 tgen_ori(s, a0, (uint32_t)~a2); 2127 } else { 2128 tcg_out_insn(s, RRFa, OCRK, a0, a1, a2); 2129 } 2130 break; 2131 case INDEX_op_eqv_i32: 2132 a0 = args[0], a1 = args[1], a2 = (uint32_t)args[2]; 2133 if (const_args[2]) { 2134 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2135 tcg_out_insn(s, RIL, XILF, a0, ~a2); 2136 } else { 2137 tcg_out_insn(s, RRFa, NXRK, a0, a1, a2); 2138 } 2139 break; 2140 case INDEX_op_nand_i32: 2141 tcg_out_insn(s, RRFa, NNRK, args[0], args[1], args[2]); 2142 break; 2143 case INDEX_op_nor_i32: 2144 tcg_out_insn(s, RRFa, NORK, args[0], args[1], args[2]); 2145 break; 2146 2147 case INDEX_op_neg_i32: 2148 tcg_out_insn(s, RR, LCR, args[0], args[1]); 2149 break; 2150 case INDEX_op_not_i32: 2151 tcg_out_insn(s, RRFa, NORK, args[0], args[1], args[1]); 2152 break; 2153 2154 case INDEX_op_mul_i32: 2155 a0 = args[0], a1 = args[1], a2 = (int32_t)args[2]; 2156 if (const_args[2]) { 2157 tcg_out_mov(s, TCG_TYPE_I32, a0, a1); 2158 if (a2 == (int16_t)a2) { 2159 tcg_out_insn(s, RI, MHI, a0, a2); 2160 } else { 2161 tcg_out_insn(s, RIL, MSFI, a0, a2); 2162 } 2163 } else if (a0 == a1) { 2164 tcg_out_insn(s, RRE, MSR, a0, a2); 2165 } else { 2166 tcg_out_insn(s, RRFa, MSRKC, a0, a1, a2); 2167 } 2168 break; 2169 2170 case INDEX_op_div2_i32: 2171 tcg_debug_assert(args[0] == args[2]); 2172 tcg_debug_assert(args[1] == args[3]); 2173 tcg_debug_assert((args[1] & 1) == 0); 2174 tcg_debug_assert(args[0] == args[1] + 1); 2175 tcg_out_insn(s, RR, DR, args[1], args[4]); 2176 break; 2177 case INDEX_op_divu2_i32: 2178 tcg_debug_assert(args[0] == args[2]); 2179 tcg_debug_assert(args[1] == args[3]); 2180 tcg_debug_assert((args[1] & 1) == 0); 2181 tcg_debug_assert(args[0] == args[1] + 1); 2182 tcg_out_insn(s, RRE, DLR, args[1], args[4]); 2183 break; 2184 2185 case INDEX_op_shl_i32: 2186 op = RS_SLL; 2187 op2 = RSY_SLLK; 2188 do_shift32: 2189 a0 = args[0], a1 = args[1], a2 = (int32_t)args[2]; 2190 if (a0 == a1) { 2191 if (const_args[2]) { 2192 tcg_out_sh32(s, op, a0, TCG_REG_NONE, a2); 2193 } else { 2194 tcg_out_sh32(s, op, a0, a2, 0); 2195 } 2196 } else { 2197 /* Using tcg_out_sh64 here for the format; it is a 32-bit shift. */ 2198 if (const_args[2]) { 2199 tcg_out_sh64(s, op2, a0, a1, TCG_REG_NONE, a2); 2200 } else { 2201 tcg_out_sh64(s, op2, a0, a1, a2, 0); 2202 } 2203 } 2204 break; 2205 case INDEX_op_shr_i32: 2206 op = RS_SRL; 2207 op2 = RSY_SRLK; 2208 goto do_shift32; 2209 case INDEX_op_sar_i32: 2210 op = RS_SRA; 2211 op2 = RSY_SRAK; 2212 goto do_shift32; 2213 2214 case INDEX_op_rotl_i32: 2215 /* ??? Using tcg_out_sh64 here for the format; it is a 32-bit rol. */ 2216 if (const_args[2]) { 2217 tcg_out_sh64(s, RSY_RLL, args[0], args[1], TCG_REG_NONE, args[2]); 2218 } else { 2219 tcg_out_sh64(s, RSY_RLL, args[0], args[1], args[2], 0); 2220 } 2221 break; 2222 case INDEX_op_rotr_i32: 2223 if (const_args[2]) { 2224 tcg_out_sh64(s, RSY_RLL, args[0], args[1], 2225 TCG_REG_NONE, (32 - args[2]) & 31); 2226 } else { 2227 tcg_out_insn(s, RR, LCR, TCG_TMP0, args[2]); 2228 tcg_out_sh64(s, RSY_RLL, args[0], args[1], TCG_TMP0, 0); 2229 } 2230 break; 2231 2232 case INDEX_op_bswap16_i32: 2233 a0 = args[0], a1 = args[1], a2 = args[2]; 2234 tcg_out_insn(s, RRE, LRVR, a0, a1); 2235 if (a2 & TCG_BSWAP_OS) { 2236 tcg_out_sh32(s, RS_SRA, a0, TCG_REG_NONE, 16); 2237 } else { 2238 tcg_out_sh32(s, RS_SRL, a0, TCG_REG_NONE, 16); 2239 } 2240 break; 2241 case INDEX_op_bswap16_i64: 2242 a0 = args[0], a1 = args[1], a2 = args[2]; 2243 tcg_out_insn(s, RRE, LRVGR, a0, a1); 2244 if (a2 & TCG_BSWAP_OS) { 2245 tcg_out_sh64(s, RSY_SRAG, a0, a0, TCG_REG_NONE, 48); 2246 } else { 2247 tcg_out_sh64(s, RSY_SRLG, a0, a0, TCG_REG_NONE, 48); 2248 } 2249 break; 2250 2251 case INDEX_op_bswap32_i32: 2252 tcg_out_insn(s, RRE, LRVR, args[0], args[1]); 2253 break; 2254 case INDEX_op_bswap32_i64: 2255 a0 = args[0], a1 = args[1], a2 = args[2]; 2256 tcg_out_insn(s, RRE, LRVR, a0, a1); 2257 if (a2 & TCG_BSWAP_OS) { 2258 tcg_out_ext32s(s, a0, a0); 2259 } else if ((a2 & (TCG_BSWAP_IZ | TCG_BSWAP_OZ)) == TCG_BSWAP_OZ) { 2260 tcg_out_ext32u(s, a0, a0); 2261 } 2262 break; 2263 2264 case INDEX_op_add2_i32: 2265 if (const_args[4]) { 2266 tcg_out_insn(s, RIL, ALFI, args[0], args[4]); 2267 } else { 2268 tcg_out_insn(s, RR, ALR, args[0], args[4]); 2269 } 2270 tcg_out_insn(s, RRE, ALCR, args[1], args[5]); 2271 break; 2272 case INDEX_op_sub2_i32: 2273 if (const_args[4]) { 2274 tcg_out_insn(s, RIL, SLFI, args[0], args[4]); 2275 } else { 2276 tcg_out_insn(s, RR, SLR, args[0], args[4]); 2277 } 2278 tcg_out_insn(s, RRE, SLBR, args[1], args[5]); 2279 break; 2280 2281 case INDEX_op_br: 2282 tgen_branch(s, S390_CC_ALWAYS, arg_label(args[0])); 2283 break; 2284 2285 case INDEX_op_brcond_i32: 2286 tgen_brcond(s, TCG_TYPE_I32, args[2], args[0], 2287 args[1], const_args[1], arg_label(args[3])); 2288 break; 2289 case INDEX_op_setcond_i32: 2290 tgen_setcond(s, TCG_TYPE_I32, args[3], args[0], args[1], 2291 args[2], const_args[2]); 2292 break; 2293 case INDEX_op_movcond_i32: 2294 tgen_movcond(s, TCG_TYPE_I32, args[5], args[0], args[1], 2295 args[2], const_args[2], args[3], const_args[3], args[4]); 2296 break; 2297 2298 case INDEX_op_qemu_ld_a32_i32: 2299 case INDEX_op_qemu_ld_a64_i32: 2300 tcg_out_qemu_ld(s, args[0], args[1], args[2], TCG_TYPE_I32); 2301 break; 2302 case INDEX_op_qemu_ld_a32_i64: 2303 case INDEX_op_qemu_ld_a64_i64: 2304 tcg_out_qemu_ld(s, args[0], args[1], args[2], TCG_TYPE_I64); 2305 break; 2306 case INDEX_op_qemu_st_a32_i32: 2307 case INDEX_op_qemu_st_a64_i32: 2308 tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I32); 2309 break; 2310 case INDEX_op_qemu_st_a32_i64: 2311 case INDEX_op_qemu_st_a64_i64: 2312 tcg_out_qemu_st(s, args[0], args[1], args[2], TCG_TYPE_I64); 2313 break; 2314 case INDEX_op_qemu_ld_a32_i128: 2315 case INDEX_op_qemu_ld_a64_i128: 2316 tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], true); 2317 break; 2318 case INDEX_op_qemu_st_a32_i128: 2319 case INDEX_op_qemu_st_a64_i128: 2320 tcg_out_qemu_ldst_i128(s, args[0], args[1], args[2], args[3], false); 2321 break; 2322 2323 case INDEX_op_ld16s_i64: 2324 tcg_out_mem(s, 0, RXY_LGH, args[0], args[1], TCG_REG_NONE, args[2]); 2325 break; 2326 case INDEX_op_ld32u_i64: 2327 tcg_out_mem(s, 0, RXY_LLGF, args[0], args[1], TCG_REG_NONE, args[2]); 2328 break; 2329 case INDEX_op_ld32s_i64: 2330 tcg_out_mem(s, 0, RXY_LGF, args[0], args[1], TCG_REG_NONE, args[2]); 2331 break; 2332 case INDEX_op_ld_i64: 2333 tcg_out_ld(s, TCG_TYPE_I64, args[0], args[1], args[2]); 2334 break; 2335 2336 case INDEX_op_st32_i64: 2337 tcg_out_st(s, TCG_TYPE_I32, args[0], args[1], args[2]); 2338 break; 2339 case INDEX_op_st_i64: 2340 tcg_out_st(s, TCG_TYPE_I64, args[0], args[1], args[2]); 2341 break; 2342 2343 case INDEX_op_add_i64: 2344 a0 = args[0], a1 = args[1], a2 = args[2]; 2345 if (const_args[2]) { 2346 do_addi_64: 2347 if (a0 == a1) { 2348 if (a2 == (int16_t)a2) { 2349 tcg_out_insn(s, RI, AGHI, a0, a2); 2350 break; 2351 } 2352 if (a2 == (int32_t)a2) { 2353 tcg_out_insn(s, RIL, AGFI, a0, a2); 2354 break; 2355 } 2356 if (a2 == (uint32_t)a2) { 2357 tcg_out_insn(s, RIL, ALGFI, a0, a2); 2358 break; 2359 } 2360 if (-a2 == (uint32_t)-a2) { 2361 tcg_out_insn(s, RIL, SLGFI, a0, -a2); 2362 break; 2363 } 2364 } 2365 tcg_out_mem(s, RX_LA, RXY_LAY, a0, a1, TCG_REG_NONE, a2); 2366 } else if (a0 == a1) { 2367 tcg_out_insn(s, RRE, AGR, a0, a2); 2368 } else { 2369 tcg_out_insn(s, RX, LA, a0, a1, a2, 0); 2370 } 2371 break; 2372 case INDEX_op_sub_i64: 2373 a0 = args[0], a1 = args[1], a2 = args[2]; 2374 if (const_args[2]) { 2375 a2 = -a2; 2376 goto do_addi_64; 2377 } else { 2378 tcg_out_insn(s, RRFa, SGRK, a0, a1, a2); 2379 } 2380 break; 2381 2382 case INDEX_op_and_i64: 2383 a0 = args[0], a1 = args[1], a2 = args[2]; 2384 if (const_args[2]) { 2385 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2386 tgen_andi(s, TCG_TYPE_I64, args[0], args[2]); 2387 } else { 2388 tcg_out_insn(s, RRFa, NGRK, a0, a1, a2); 2389 } 2390 break; 2391 case INDEX_op_or_i64: 2392 a0 = args[0], a1 = args[1], a2 = args[2]; 2393 if (const_args[2]) { 2394 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2395 tgen_ori(s, a0, a2); 2396 } else { 2397 tcg_out_insn(s, RRFa, OGRK, a0, a1, a2); 2398 } 2399 break; 2400 case INDEX_op_xor_i64: 2401 a0 = args[0], a1 = args[1], a2 = args[2]; 2402 if (const_args[2]) { 2403 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2404 tgen_xori(s, a0, a2); 2405 } else { 2406 tcg_out_insn(s, RRFa, XGRK, a0, a1, a2); 2407 } 2408 break; 2409 2410 case INDEX_op_andc_i64: 2411 a0 = args[0], a1 = args[1], a2 = args[2]; 2412 if (const_args[2]) { 2413 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2414 tgen_andi(s, TCG_TYPE_I64, a0, ~a2); 2415 } else { 2416 tcg_out_insn(s, RRFa, NCGRK, a0, a1, a2); 2417 } 2418 break; 2419 case INDEX_op_orc_i64: 2420 a0 = args[0], a1 = args[1], a2 = args[2]; 2421 if (const_args[2]) { 2422 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2423 tgen_ori(s, a0, ~a2); 2424 } else { 2425 tcg_out_insn(s, RRFa, OCGRK, a0, a1, a2); 2426 } 2427 break; 2428 case INDEX_op_eqv_i64: 2429 a0 = args[0], a1 = args[1], a2 = args[2]; 2430 if (const_args[2]) { 2431 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2432 tgen_xori(s, a0, ~a2); 2433 } else { 2434 tcg_out_insn(s, RRFa, NXGRK, a0, a1, a2); 2435 } 2436 break; 2437 case INDEX_op_nand_i64: 2438 tcg_out_insn(s, RRFa, NNGRK, args[0], args[1], args[2]); 2439 break; 2440 case INDEX_op_nor_i64: 2441 tcg_out_insn(s, RRFa, NOGRK, args[0], args[1], args[2]); 2442 break; 2443 2444 case INDEX_op_neg_i64: 2445 tcg_out_insn(s, RRE, LCGR, args[0], args[1]); 2446 break; 2447 case INDEX_op_not_i64: 2448 tcg_out_insn(s, RRFa, NOGRK, args[0], args[1], args[1]); 2449 break; 2450 case INDEX_op_bswap64_i64: 2451 tcg_out_insn(s, RRE, LRVGR, args[0], args[1]); 2452 break; 2453 2454 case INDEX_op_mul_i64: 2455 a0 = args[0], a1 = args[1], a2 = args[2]; 2456 if (const_args[2]) { 2457 tcg_out_mov(s, TCG_TYPE_I64, a0, a1); 2458 if (a2 == (int16_t)a2) { 2459 tcg_out_insn(s, RI, MGHI, a0, a2); 2460 } else { 2461 tcg_out_insn(s, RIL, MSGFI, a0, a2); 2462 } 2463 } else if (a0 == a1) { 2464 tcg_out_insn(s, RRE, MSGR, a0, a2); 2465 } else { 2466 tcg_out_insn(s, RRFa, MSGRKC, a0, a1, a2); 2467 } 2468 break; 2469 2470 case INDEX_op_div2_i64: 2471 /* 2472 * ??? We get an unnecessary sign-extension of the dividend 2473 * into op0 with this definition, but as we do in fact always 2474 * produce both quotient and remainder using INDEX_op_div_i64 2475 * instead requires jumping through even more hoops. 2476 */ 2477 tcg_debug_assert(args[0] == args[2]); 2478 tcg_debug_assert(args[1] == args[3]); 2479 tcg_debug_assert((args[1] & 1) == 0); 2480 tcg_debug_assert(args[0] == args[1] + 1); 2481 tcg_out_insn(s, RRE, DSGR, args[1], args[4]); 2482 break; 2483 case INDEX_op_divu2_i64: 2484 tcg_debug_assert(args[0] == args[2]); 2485 tcg_debug_assert(args[1] == args[3]); 2486 tcg_debug_assert((args[1] & 1) == 0); 2487 tcg_debug_assert(args[0] == args[1] + 1); 2488 tcg_out_insn(s, RRE, DLGR, args[1], args[4]); 2489 break; 2490 case INDEX_op_mulu2_i64: 2491 tcg_debug_assert(args[0] == args[2]); 2492 tcg_debug_assert((args[1] & 1) == 0); 2493 tcg_debug_assert(args[0] == args[1] + 1); 2494 tcg_out_insn(s, RRE, MLGR, args[1], args[3]); 2495 break; 2496 case INDEX_op_muls2_i64: 2497 tcg_debug_assert((args[1] & 1) == 0); 2498 tcg_debug_assert(args[0] == args[1] + 1); 2499 tcg_out_insn(s, RRFa, MGRK, args[1], args[2], args[3]); 2500 break; 2501 2502 case INDEX_op_shl_i64: 2503 op = RSY_SLLG; 2504 do_shift64: 2505 if (const_args[2]) { 2506 tcg_out_sh64(s, op, args[0], args[1], TCG_REG_NONE, args[2]); 2507 } else { 2508 tcg_out_sh64(s, op, args[0], args[1], args[2], 0); 2509 } 2510 break; 2511 case INDEX_op_shr_i64: 2512 op = RSY_SRLG; 2513 goto do_shift64; 2514 case INDEX_op_sar_i64: 2515 op = RSY_SRAG; 2516 goto do_shift64; 2517 2518 case INDEX_op_rotl_i64: 2519 if (const_args[2]) { 2520 tcg_out_sh64(s, RSY_RLLG, args[0], args[1], 2521 TCG_REG_NONE, args[2]); 2522 } else { 2523 tcg_out_sh64(s, RSY_RLLG, args[0], args[1], args[2], 0); 2524 } 2525 break; 2526 case INDEX_op_rotr_i64: 2527 if (const_args[2]) { 2528 tcg_out_sh64(s, RSY_RLLG, args[0], args[1], 2529 TCG_REG_NONE, (64 - args[2]) & 63); 2530 } else { 2531 /* We can use the smaller 32-bit negate because only the 2532 low 6 bits are examined for the rotate. */ 2533 tcg_out_insn(s, RR, LCR, TCG_TMP0, args[2]); 2534 tcg_out_sh64(s, RSY_RLLG, args[0], args[1], TCG_TMP0, 0); 2535 } 2536 break; 2537 2538 case INDEX_op_add2_i64: 2539 if (const_args[4]) { 2540 if ((int64_t)args[4] >= 0) { 2541 tcg_out_insn(s, RIL, ALGFI, args[0], args[4]); 2542 } else { 2543 tcg_out_insn(s, RIL, SLGFI, args[0], -args[4]); 2544 } 2545 } else { 2546 tcg_out_insn(s, RRE, ALGR, args[0], args[4]); 2547 } 2548 tcg_out_insn(s, RRE, ALCGR, args[1], args[5]); 2549 break; 2550 case INDEX_op_sub2_i64: 2551 if (const_args[4]) { 2552 if ((int64_t)args[4] >= 0) { 2553 tcg_out_insn(s, RIL, SLGFI, args[0], args[4]); 2554 } else { 2555 tcg_out_insn(s, RIL, ALGFI, args[0], -args[4]); 2556 } 2557 } else { 2558 tcg_out_insn(s, RRE, SLGR, args[0], args[4]); 2559 } 2560 tcg_out_insn(s, RRE, SLBGR, args[1], args[5]); 2561 break; 2562 2563 case INDEX_op_brcond_i64: 2564 tgen_brcond(s, TCG_TYPE_I64, args[2], args[0], 2565 args[1], const_args[1], arg_label(args[3])); 2566 break; 2567 case INDEX_op_setcond_i64: 2568 tgen_setcond(s, TCG_TYPE_I64, args[3], args[0], args[1], 2569 args[2], const_args[2]); 2570 break; 2571 case INDEX_op_movcond_i64: 2572 tgen_movcond(s, TCG_TYPE_I64, args[5], args[0], args[1], 2573 args[2], const_args[2], args[3], const_args[3], args[4]); 2574 break; 2575 2576 OP_32_64(deposit): 2577 a0 = args[0], a1 = args[1], a2 = args[2]; 2578 if (const_args[1]) { 2579 tgen_deposit(s, a0, a2, args[3], args[4], 1); 2580 } else { 2581 /* Since we can't support "0Z" as a constraint, we allow a1 in 2582 any register. Fix things up as if a matching constraint. */ 2583 if (a0 != a1) { 2584 TCGType type = (opc == INDEX_op_deposit_i64); 2585 if (a0 == a2) { 2586 tcg_out_mov(s, type, TCG_TMP0, a2); 2587 a2 = TCG_TMP0; 2588 } 2589 tcg_out_mov(s, type, a0, a1); 2590 } 2591 tgen_deposit(s, a0, a2, args[3], args[4], 0); 2592 } 2593 break; 2594 2595 OP_32_64(extract): 2596 tgen_extract(s, args[0], args[1], args[2], args[3]); 2597 break; 2598 2599 case INDEX_op_clz_i64: 2600 tgen_clz(s, args[0], args[1], args[2], const_args[2]); 2601 break; 2602 2603 case INDEX_op_ctpop_i32: 2604 tgen_ctpop(s, TCG_TYPE_I32, args[0], args[1]); 2605 break; 2606 case INDEX_op_ctpop_i64: 2607 tgen_ctpop(s, TCG_TYPE_I64, args[0], args[1]); 2608 break; 2609 2610 case INDEX_op_mb: 2611 /* The host memory model is quite strong, we simply need to 2612 serialize the instruction stream. */ 2613 if (args[0] & TCG_MO_ST_LD) { 2614 /* fast-bcr-serialization facility (45) is present */ 2615 tcg_out_insn(s, RR, BCR, 14, 0); 2616 } 2617 break; 2618 2619 case INDEX_op_mov_i32: /* Always emitted via tcg_out_mov. */ 2620 case INDEX_op_mov_i64: 2621 case INDEX_op_call: /* Always emitted via tcg_out_call. */ 2622 case INDEX_op_exit_tb: /* Always emitted via tcg_out_exit_tb. */ 2623 case INDEX_op_goto_tb: /* Always emitted via tcg_out_goto_tb. */ 2624 case INDEX_op_ext8s_i32: /* Always emitted via tcg_reg_alloc_op. */ 2625 case INDEX_op_ext8s_i64: 2626 case INDEX_op_ext8u_i32: 2627 case INDEX_op_ext8u_i64: 2628 case INDEX_op_ext16s_i32: 2629 case INDEX_op_ext16s_i64: 2630 case INDEX_op_ext16u_i32: 2631 case INDEX_op_ext16u_i64: 2632 case INDEX_op_ext32s_i64: 2633 case INDEX_op_ext32u_i64: 2634 case INDEX_op_ext_i32_i64: 2635 case INDEX_op_extu_i32_i64: 2636 case INDEX_op_extrl_i64_i32: 2637 default: 2638 g_assert_not_reached(); 2639 } 2640} 2641 2642static bool tcg_out_dup_vec(TCGContext *s, TCGType type, unsigned vece, 2643 TCGReg dst, TCGReg src) 2644{ 2645 if (is_general_reg(src)) { 2646 /* Replicate general register into two MO_64. */ 2647 tcg_out_insn(s, VRRf, VLVGP, dst, src, src); 2648 if (vece == MO_64) { 2649 return true; 2650 } 2651 src = dst; 2652 } 2653 2654 /* 2655 * Recall that the "standard" integer, within a vector, is the 2656 * rightmost element of the leftmost doubleword, a-la VLLEZ. 2657 */ 2658 tcg_out_insn(s, VRIc, VREP, dst, (8 >> vece) - 1, src, vece); 2659 return true; 2660} 2661 2662static bool tcg_out_dupm_vec(TCGContext *s, TCGType type, unsigned vece, 2663 TCGReg dst, TCGReg base, intptr_t offset) 2664{ 2665 tcg_out_vrx_mem(s, VRX_VLREP, dst, base, TCG_REG_NONE, offset, vece); 2666 return true; 2667} 2668 2669static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece, 2670 TCGReg dst, int64_t val) 2671{ 2672 int i, mask, msb, lsb; 2673 2674 /* Look for int16_t elements. */ 2675 if (vece <= MO_16 || 2676 (vece == MO_32 ? (int32_t)val : val) == (int16_t)val) { 2677 tcg_out_insn(s, VRIa, VREPI, dst, val, vece); 2678 return; 2679 } 2680 2681 /* Look for bit masks. */ 2682 if (vece == MO_32) { 2683 if (risbg_mask((int32_t)val)) { 2684 /* Handle wraparound by swapping msb and lsb. */ 2685 if ((val & 0x80000001u) == 0x80000001u) { 2686 msb = 32 - ctz32(~val); 2687 lsb = clz32(~val) - 1; 2688 } else { 2689 msb = clz32(val); 2690 lsb = 31 - ctz32(val); 2691 } 2692 tcg_out_insn(s, VRIb, VGM, dst, msb, lsb, MO_32); 2693 return; 2694 } 2695 } else { 2696 if (risbg_mask(val)) { 2697 /* Handle wraparound by swapping msb and lsb. */ 2698 if ((val & 0x8000000000000001ull) == 0x8000000000000001ull) { 2699 /* Handle wraparound by swapping msb and lsb. */ 2700 msb = 64 - ctz64(~val); 2701 lsb = clz64(~val) - 1; 2702 } else { 2703 msb = clz64(val); 2704 lsb = 63 - ctz64(val); 2705 } 2706 tcg_out_insn(s, VRIb, VGM, dst, msb, lsb, MO_64); 2707 return; 2708 } 2709 } 2710 2711 /* Look for all bytes 0x00 or 0xff. */ 2712 for (i = mask = 0; i < 8; i++) { 2713 uint8_t byte = val >> (i * 8); 2714 if (byte == 0xff) { 2715 mask |= 1 << i; 2716 } else if (byte != 0) { 2717 break; 2718 } 2719 } 2720 if (i == 8) { 2721 tcg_out_insn(s, VRIa, VGBM, dst, mask * 0x0101, 0); 2722 return; 2723 } 2724 2725 /* Otherwise, stuff it in the constant pool. */ 2726 tcg_out_insn(s, RIL, LARL, TCG_TMP0, 0); 2727 new_pool_label(s, val, R_390_PC32DBL, s->code_ptr - 2, 2); 2728 tcg_out_insn(s, VRX, VLREP, dst, TCG_TMP0, TCG_REG_NONE, 0, MO_64); 2729} 2730 2731static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc, 2732 unsigned vecl, unsigned vece, 2733 const TCGArg args[TCG_MAX_OP_ARGS], 2734 const int const_args[TCG_MAX_OP_ARGS]) 2735{ 2736 TCGType type = vecl + TCG_TYPE_V64; 2737 TCGArg a0 = args[0], a1 = args[1], a2 = args[2]; 2738 2739 switch (opc) { 2740 case INDEX_op_ld_vec: 2741 tcg_out_ld(s, type, a0, a1, a2); 2742 break; 2743 case INDEX_op_st_vec: 2744 tcg_out_st(s, type, a0, a1, a2); 2745 break; 2746 case INDEX_op_dupm_vec: 2747 tcg_out_dupm_vec(s, type, vece, a0, a1, a2); 2748 break; 2749 2750 case INDEX_op_abs_vec: 2751 tcg_out_insn(s, VRRa, VLP, a0, a1, vece); 2752 break; 2753 case INDEX_op_neg_vec: 2754 tcg_out_insn(s, VRRa, VLC, a0, a1, vece); 2755 break; 2756 case INDEX_op_not_vec: 2757 tcg_out_insn(s, VRRc, VNO, a0, a1, a1, 0); 2758 break; 2759 2760 case INDEX_op_add_vec: 2761 tcg_out_insn(s, VRRc, VA, a0, a1, a2, vece); 2762 break; 2763 case INDEX_op_sub_vec: 2764 tcg_out_insn(s, VRRc, VS, a0, a1, a2, vece); 2765 break; 2766 case INDEX_op_and_vec: 2767 tcg_out_insn(s, VRRc, VN, a0, a1, a2, 0); 2768 break; 2769 case INDEX_op_andc_vec: 2770 tcg_out_insn(s, VRRc, VNC, a0, a1, a2, 0); 2771 break; 2772 case INDEX_op_mul_vec: 2773 tcg_out_insn(s, VRRc, VML, a0, a1, a2, vece); 2774 break; 2775 case INDEX_op_or_vec: 2776 tcg_out_insn(s, VRRc, VO, a0, a1, a2, 0); 2777 break; 2778 case INDEX_op_orc_vec: 2779 tcg_out_insn(s, VRRc, VOC, a0, a1, a2, 0); 2780 break; 2781 case INDEX_op_xor_vec: 2782 tcg_out_insn(s, VRRc, VX, a0, a1, a2, 0); 2783 break; 2784 case INDEX_op_nand_vec: 2785 tcg_out_insn(s, VRRc, VNN, a0, a1, a2, 0); 2786 break; 2787 case INDEX_op_nor_vec: 2788 tcg_out_insn(s, VRRc, VNO, a0, a1, a2, 0); 2789 break; 2790 case INDEX_op_eqv_vec: 2791 tcg_out_insn(s, VRRc, VNX, a0, a1, a2, 0); 2792 break; 2793 2794 case INDEX_op_shli_vec: 2795 tcg_out_insn(s, VRSa, VESL, a0, a2, TCG_REG_NONE, a1, vece); 2796 break; 2797 case INDEX_op_shri_vec: 2798 tcg_out_insn(s, VRSa, VESRL, a0, a2, TCG_REG_NONE, a1, vece); 2799 break; 2800 case INDEX_op_sari_vec: 2801 tcg_out_insn(s, VRSa, VESRA, a0, a2, TCG_REG_NONE, a1, vece); 2802 break; 2803 case INDEX_op_rotli_vec: 2804 tcg_out_insn(s, VRSa, VERLL, a0, a2, TCG_REG_NONE, a1, vece); 2805 break; 2806 case INDEX_op_shls_vec: 2807 tcg_out_insn(s, VRSa, VESL, a0, 0, a2, a1, vece); 2808 break; 2809 case INDEX_op_shrs_vec: 2810 tcg_out_insn(s, VRSa, VESRL, a0, 0, a2, a1, vece); 2811 break; 2812 case INDEX_op_sars_vec: 2813 tcg_out_insn(s, VRSa, VESRA, a0, 0, a2, a1, vece); 2814 break; 2815 case INDEX_op_rotls_vec: 2816 tcg_out_insn(s, VRSa, VERLL, a0, 0, a2, a1, vece); 2817 break; 2818 case INDEX_op_shlv_vec: 2819 tcg_out_insn(s, VRRc, VESLV, a0, a1, a2, vece); 2820 break; 2821 case INDEX_op_shrv_vec: 2822 tcg_out_insn(s, VRRc, VESRLV, a0, a1, a2, vece); 2823 break; 2824 case INDEX_op_sarv_vec: 2825 tcg_out_insn(s, VRRc, VESRAV, a0, a1, a2, vece); 2826 break; 2827 case INDEX_op_rotlv_vec: 2828 tcg_out_insn(s, VRRc, VERLLV, a0, a1, a2, vece); 2829 break; 2830 2831 case INDEX_op_smin_vec: 2832 tcg_out_insn(s, VRRc, VMN, a0, a1, a2, vece); 2833 break; 2834 case INDEX_op_smax_vec: 2835 tcg_out_insn(s, VRRc, VMX, a0, a1, a2, vece); 2836 break; 2837 case INDEX_op_umin_vec: 2838 tcg_out_insn(s, VRRc, VMNL, a0, a1, a2, vece); 2839 break; 2840 case INDEX_op_umax_vec: 2841 tcg_out_insn(s, VRRc, VMXL, a0, a1, a2, vece); 2842 break; 2843 2844 case INDEX_op_bitsel_vec: 2845 tcg_out_insn(s, VRRe, VSEL, a0, a2, args[3], a1); 2846 break; 2847 2848 case INDEX_op_cmp_vec: 2849 switch ((TCGCond)args[3]) { 2850 case TCG_COND_EQ: 2851 tcg_out_insn(s, VRRc, VCEQ, a0, a1, a2, vece); 2852 break; 2853 case TCG_COND_GT: 2854 tcg_out_insn(s, VRRc, VCH, a0, a1, a2, vece); 2855 break; 2856 case TCG_COND_GTU: 2857 tcg_out_insn(s, VRRc, VCHL, a0, a1, a2, vece); 2858 break; 2859 default: 2860 g_assert_not_reached(); 2861 } 2862 break; 2863 2864 case INDEX_op_s390_vuph_vec: 2865 tcg_out_insn(s, VRRa, VUPH, a0, a1, vece); 2866 break; 2867 case INDEX_op_s390_vupl_vec: 2868 tcg_out_insn(s, VRRa, VUPL, a0, a1, vece); 2869 break; 2870 case INDEX_op_s390_vpks_vec: 2871 tcg_out_insn(s, VRRc, VPKS, a0, a1, a2, vece); 2872 break; 2873 2874 case INDEX_op_mov_vec: /* Always emitted via tcg_out_mov. */ 2875 case INDEX_op_dup_vec: /* Always emitted via tcg_out_dup_vec. */ 2876 default: 2877 g_assert_not_reached(); 2878 } 2879} 2880 2881int tcg_can_emit_vec_op(TCGOpcode opc, TCGType type, unsigned vece) 2882{ 2883 switch (opc) { 2884 case INDEX_op_abs_vec: 2885 case INDEX_op_add_vec: 2886 case INDEX_op_and_vec: 2887 case INDEX_op_andc_vec: 2888 case INDEX_op_bitsel_vec: 2889 case INDEX_op_eqv_vec: 2890 case INDEX_op_nand_vec: 2891 case INDEX_op_neg_vec: 2892 case INDEX_op_nor_vec: 2893 case INDEX_op_not_vec: 2894 case INDEX_op_or_vec: 2895 case INDEX_op_orc_vec: 2896 case INDEX_op_rotli_vec: 2897 case INDEX_op_rotls_vec: 2898 case INDEX_op_rotlv_vec: 2899 case INDEX_op_sari_vec: 2900 case INDEX_op_sars_vec: 2901 case INDEX_op_sarv_vec: 2902 case INDEX_op_shli_vec: 2903 case INDEX_op_shls_vec: 2904 case INDEX_op_shlv_vec: 2905 case INDEX_op_shri_vec: 2906 case INDEX_op_shrs_vec: 2907 case INDEX_op_shrv_vec: 2908 case INDEX_op_smax_vec: 2909 case INDEX_op_smin_vec: 2910 case INDEX_op_sub_vec: 2911 case INDEX_op_umax_vec: 2912 case INDEX_op_umin_vec: 2913 case INDEX_op_xor_vec: 2914 return 1; 2915 case INDEX_op_cmp_vec: 2916 case INDEX_op_cmpsel_vec: 2917 case INDEX_op_rotrv_vec: 2918 return -1; 2919 case INDEX_op_mul_vec: 2920 return vece < MO_64; 2921 case INDEX_op_ssadd_vec: 2922 case INDEX_op_sssub_vec: 2923 return vece < MO_64 ? -1 : 0; 2924 default: 2925 return 0; 2926 } 2927} 2928 2929static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0, 2930 TCGv_vec v1, TCGv_vec v2, TCGCond cond) 2931{ 2932 bool need_swap = false, need_inv = false; 2933 2934 switch (cond) { 2935 case TCG_COND_EQ: 2936 case TCG_COND_GT: 2937 case TCG_COND_GTU: 2938 break; 2939 case TCG_COND_NE: 2940 case TCG_COND_LE: 2941 case TCG_COND_LEU: 2942 need_inv = true; 2943 break; 2944 case TCG_COND_LT: 2945 case TCG_COND_LTU: 2946 need_swap = true; 2947 break; 2948 case TCG_COND_GE: 2949 case TCG_COND_GEU: 2950 need_swap = need_inv = true; 2951 break; 2952 default: 2953 g_assert_not_reached(); 2954 } 2955 2956 if (need_inv) { 2957 cond = tcg_invert_cond(cond); 2958 } 2959 if (need_swap) { 2960 TCGv_vec t1; 2961 t1 = v1, v1 = v2, v2 = t1; 2962 cond = tcg_swap_cond(cond); 2963 } 2964 2965 vec_gen_4(INDEX_op_cmp_vec, type, vece, tcgv_vec_arg(v0), 2966 tcgv_vec_arg(v1), tcgv_vec_arg(v2), cond); 2967 2968 return need_inv; 2969} 2970 2971static void expand_vec_cmp(TCGType type, unsigned vece, TCGv_vec v0, 2972 TCGv_vec v1, TCGv_vec v2, TCGCond cond) 2973{ 2974 if (expand_vec_cmp_noinv(type, vece, v0, v1, v2, cond)) { 2975 tcg_gen_not_vec(vece, v0, v0); 2976 } 2977} 2978 2979static void expand_vec_cmpsel(TCGType type, unsigned vece, TCGv_vec v0, 2980 TCGv_vec c1, TCGv_vec c2, 2981 TCGv_vec v3, TCGv_vec v4, TCGCond cond) 2982{ 2983 TCGv_vec t = tcg_temp_new_vec(type); 2984 2985 if (expand_vec_cmp_noinv(type, vece, t, c1, c2, cond)) { 2986 /* Invert the sense of the compare by swapping arguments. */ 2987 tcg_gen_bitsel_vec(vece, v0, t, v4, v3); 2988 } else { 2989 tcg_gen_bitsel_vec(vece, v0, t, v3, v4); 2990 } 2991 tcg_temp_free_vec(t); 2992} 2993 2994static void expand_vec_sat(TCGType type, unsigned vece, TCGv_vec v0, 2995 TCGv_vec v1, TCGv_vec v2, TCGOpcode add_sub_opc) 2996{ 2997 TCGv_vec h1 = tcg_temp_new_vec(type); 2998 TCGv_vec h2 = tcg_temp_new_vec(type); 2999 TCGv_vec l1 = tcg_temp_new_vec(type); 3000 TCGv_vec l2 = tcg_temp_new_vec(type); 3001 3002 tcg_debug_assert (vece < MO_64); 3003 3004 /* Unpack with sign-extension. */ 3005 vec_gen_2(INDEX_op_s390_vuph_vec, type, vece, 3006 tcgv_vec_arg(h1), tcgv_vec_arg(v1)); 3007 vec_gen_2(INDEX_op_s390_vuph_vec, type, vece, 3008 tcgv_vec_arg(h2), tcgv_vec_arg(v2)); 3009 3010 vec_gen_2(INDEX_op_s390_vupl_vec, type, vece, 3011 tcgv_vec_arg(l1), tcgv_vec_arg(v1)); 3012 vec_gen_2(INDEX_op_s390_vupl_vec, type, vece, 3013 tcgv_vec_arg(l2), tcgv_vec_arg(v2)); 3014 3015 /* Arithmetic on a wider element size. */ 3016 vec_gen_3(add_sub_opc, type, vece + 1, tcgv_vec_arg(h1), 3017 tcgv_vec_arg(h1), tcgv_vec_arg(h2)); 3018 vec_gen_3(add_sub_opc, type, vece + 1, tcgv_vec_arg(l1), 3019 tcgv_vec_arg(l1), tcgv_vec_arg(l2)); 3020 3021 /* Pack with saturation. */ 3022 vec_gen_3(INDEX_op_s390_vpks_vec, type, vece + 1, 3023 tcgv_vec_arg(v0), tcgv_vec_arg(h1), tcgv_vec_arg(l1)); 3024 3025 tcg_temp_free_vec(h1); 3026 tcg_temp_free_vec(h2); 3027 tcg_temp_free_vec(l1); 3028 tcg_temp_free_vec(l2); 3029} 3030 3031void tcg_expand_vec_op(TCGOpcode opc, TCGType type, unsigned vece, 3032 TCGArg a0, ...) 3033{ 3034 va_list va; 3035 TCGv_vec v0, v1, v2, v3, v4, t0; 3036 3037 va_start(va, a0); 3038 v0 = temp_tcgv_vec(arg_temp(a0)); 3039 v1 = temp_tcgv_vec(arg_temp(va_arg(va, TCGArg))); 3040 v2 = temp_tcgv_vec(arg_temp(va_arg(va, TCGArg))); 3041 3042 switch (opc) { 3043 case INDEX_op_cmp_vec: 3044 expand_vec_cmp(type, vece, v0, v1, v2, va_arg(va, TCGArg)); 3045 break; 3046 3047 case INDEX_op_cmpsel_vec: 3048 v3 = temp_tcgv_vec(arg_temp(va_arg(va, TCGArg))); 3049 v4 = temp_tcgv_vec(arg_temp(va_arg(va, TCGArg))); 3050 expand_vec_cmpsel(type, vece, v0, v1, v2, v3, v4, va_arg(va, TCGArg)); 3051 break; 3052 3053 case INDEX_op_rotrv_vec: 3054 t0 = tcg_temp_new_vec(type); 3055 tcg_gen_neg_vec(vece, t0, v2); 3056 tcg_gen_rotlv_vec(vece, v0, v1, t0); 3057 tcg_temp_free_vec(t0); 3058 break; 3059 3060 case INDEX_op_ssadd_vec: 3061 expand_vec_sat(type, vece, v0, v1, v2, INDEX_op_add_vec); 3062 break; 3063 case INDEX_op_sssub_vec: 3064 expand_vec_sat(type, vece, v0, v1, v2, INDEX_op_sub_vec); 3065 break; 3066 3067 default: 3068 g_assert_not_reached(); 3069 } 3070 va_end(va); 3071} 3072 3073static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op) 3074{ 3075 switch (op) { 3076 case INDEX_op_goto_ptr: 3077 return C_O0_I1(r); 3078 3079 case INDEX_op_ld8u_i32: 3080 case INDEX_op_ld8u_i64: 3081 case INDEX_op_ld8s_i32: 3082 case INDEX_op_ld8s_i64: 3083 case INDEX_op_ld16u_i32: 3084 case INDEX_op_ld16u_i64: 3085 case INDEX_op_ld16s_i32: 3086 case INDEX_op_ld16s_i64: 3087 case INDEX_op_ld_i32: 3088 case INDEX_op_ld32u_i64: 3089 case INDEX_op_ld32s_i64: 3090 case INDEX_op_ld_i64: 3091 return C_O1_I1(r, r); 3092 3093 case INDEX_op_st8_i32: 3094 case INDEX_op_st8_i64: 3095 case INDEX_op_st16_i32: 3096 case INDEX_op_st16_i64: 3097 case INDEX_op_st_i32: 3098 case INDEX_op_st32_i64: 3099 case INDEX_op_st_i64: 3100 return C_O0_I2(r, r); 3101 3102 case INDEX_op_add_i32: 3103 case INDEX_op_add_i64: 3104 case INDEX_op_shl_i64: 3105 case INDEX_op_shr_i64: 3106 case INDEX_op_sar_i64: 3107 case INDEX_op_rotl_i32: 3108 case INDEX_op_rotl_i64: 3109 case INDEX_op_rotr_i32: 3110 case INDEX_op_rotr_i64: 3111 case INDEX_op_setcond_i32: 3112 return C_O1_I2(r, r, ri); 3113 case INDEX_op_setcond_i64: 3114 return C_O1_I2(r, r, rA); 3115 3116 case INDEX_op_clz_i64: 3117 return C_O1_I2(r, r, rI); 3118 3119 case INDEX_op_sub_i32: 3120 case INDEX_op_sub_i64: 3121 case INDEX_op_and_i32: 3122 case INDEX_op_or_i32: 3123 case INDEX_op_xor_i32: 3124 return C_O1_I2(r, r, ri); 3125 case INDEX_op_and_i64: 3126 return C_O1_I2(r, r, rNKR); 3127 case INDEX_op_or_i64: 3128 case INDEX_op_xor_i64: 3129 return C_O1_I2(r, r, rK); 3130 3131 case INDEX_op_andc_i32: 3132 case INDEX_op_orc_i32: 3133 case INDEX_op_eqv_i32: 3134 return C_O1_I2(r, r, ri); 3135 case INDEX_op_andc_i64: 3136 return C_O1_I2(r, r, rKR); 3137 case INDEX_op_orc_i64: 3138 case INDEX_op_eqv_i64: 3139 return C_O1_I2(r, r, rNK); 3140 3141 case INDEX_op_nand_i32: 3142 case INDEX_op_nand_i64: 3143 case INDEX_op_nor_i32: 3144 case INDEX_op_nor_i64: 3145 return C_O1_I2(r, r, r); 3146 3147 case INDEX_op_mul_i32: 3148 return (HAVE_FACILITY(MISC_INSN_EXT2) 3149 ? C_O1_I2(r, r, ri) 3150 : C_O1_I2(r, 0, ri)); 3151 case INDEX_op_mul_i64: 3152 return (HAVE_FACILITY(MISC_INSN_EXT2) 3153 ? C_O1_I2(r, r, rJ) 3154 : C_O1_I2(r, 0, rJ)); 3155 3156 case INDEX_op_shl_i32: 3157 case INDEX_op_shr_i32: 3158 case INDEX_op_sar_i32: 3159 return C_O1_I2(r, r, ri); 3160 3161 case INDEX_op_brcond_i32: 3162 return C_O0_I2(r, ri); 3163 case INDEX_op_brcond_i64: 3164 return C_O0_I2(r, rA); 3165 3166 case INDEX_op_bswap16_i32: 3167 case INDEX_op_bswap16_i64: 3168 case INDEX_op_bswap32_i32: 3169 case INDEX_op_bswap32_i64: 3170 case INDEX_op_bswap64_i64: 3171 case INDEX_op_neg_i32: 3172 case INDEX_op_neg_i64: 3173 case INDEX_op_not_i32: 3174 case INDEX_op_not_i64: 3175 case INDEX_op_ext8s_i32: 3176 case INDEX_op_ext8s_i64: 3177 case INDEX_op_ext8u_i32: 3178 case INDEX_op_ext8u_i64: 3179 case INDEX_op_ext16s_i32: 3180 case INDEX_op_ext16s_i64: 3181 case INDEX_op_ext16u_i32: 3182 case INDEX_op_ext16u_i64: 3183 case INDEX_op_ext32s_i64: 3184 case INDEX_op_ext32u_i64: 3185 case INDEX_op_ext_i32_i64: 3186 case INDEX_op_extu_i32_i64: 3187 case INDEX_op_extract_i32: 3188 case INDEX_op_extract_i64: 3189 case INDEX_op_ctpop_i32: 3190 case INDEX_op_ctpop_i64: 3191 return C_O1_I1(r, r); 3192 3193 case INDEX_op_qemu_ld_a32_i32: 3194 case INDEX_op_qemu_ld_a64_i32: 3195 case INDEX_op_qemu_ld_a32_i64: 3196 case INDEX_op_qemu_ld_a64_i64: 3197 return C_O1_I1(r, r); 3198 case INDEX_op_qemu_st_a32_i64: 3199 case INDEX_op_qemu_st_a64_i64: 3200 case INDEX_op_qemu_st_a32_i32: 3201 case INDEX_op_qemu_st_a64_i32: 3202 return C_O0_I2(r, r); 3203 case INDEX_op_qemu_ld_a32_i128: 3204 case INDEX_op_qemu_ld_a64_i128: 3205 return C_O2_I1(o, m, r); 3206 case INDEX_op_qemu_st_a32_i128: 3207 case INDEX_op_qemu_st_a64_i128: 3208 return C_O0_I3(o, m, r); 3209 3210 case INDEX_op_deposit_i32: 3211 case INDEX_op_deposit_i64: 3212 return C_O1_I2(r, rZ, r); 3213 3214 case INDEX_op_movcond_i32: 3215 return C_O1_I4(r, r, ri, rI, r); 3216 case INDEX_op_movcond_i64: 3217 return C_O1_I4(r, r, rA, rI, r); 3218 3219 case INDEX_op_div2_i32: 3220 case INDEX_op_div2_i64: 3221 case INDEX_op_divu2_i32: 3222 case INDEX_op_divu2_i64: 3223 return C_O2_I3(o, m, 0, 1, r); 3224 3225 case INDEX_op_mulu2_i64: 3226 return C_O2_I2(o, m, 0, r); 3227 case INDEX_op_muls2_i64: 3228 return C_O2_I2(o, m, r, r); 3229 3230 case INDEX_op_add2_i32: 3231 case INDEX_op_sub2_i32: 3232 return C_N1_O1_I4(r, r, 0, 1, ri, r); 3233 3234 case INDEX_op_add2_i64: 3235 case INDEX_op_sub2_i64: 3236 return C_N1_O1_I4(r, r, 0, 1, rA, r); 3237 3238 case INDEX_op_st_vec: 3239 return C_O0_I2(v, r); 3240 case INDEX_op_ld_vec: 3241 case INDEX_op_dupm_vec: 3242 return C_O1_I1(v, r); 3243 case INDEX_op_dup_vec: 3244 return C_O1_I1(v, vr); 3245 case INDEX_op_abs_vec: 3246 case INDEX_op_neg_vec: 3247 case INDEX_op_not_vec: 3248 case INDEX_op_rotli_vec: 3249 case INDEX_op_sari_vec: 3250 case INDEX_op_shli_vec: 3251 case INDEX_op_shri_vec: 3252 case INDEX_op_s390_vuph_vec: 3253 case INDEX_op_s390_vupl_vec: 3254 return C_O1_I1(v, v); 3255 case INDEX_op_add_vec: 3256 case INDEX_op_sub_vec: 3257 case INDEX_op_and_vec: 3258 case INDEX_op_andc_vec: 3259 case INDEX_op_or_vec: 3260 case INDEX_op_orc_vec: 3261 case INDEX_op_xor_vec: 3262 case INDEX_op_nand_vec: 3263 case INDEX_op_nor_vec: 3264 case INDEX_op_eqv_vec: 3265 case INDEX_op_cmp_vec: 3266 case INDEX_op_mul_vec: 3267 case INDEX_op_rotlv_vec: 3268 case INDEX_op_rotrv_vec: 3269 case INDEX_op_shlv_vec: 3270 case INDEX_op_shrv_vec: 3271 case INDEX_op_sarv_vec: 3272 case INDEX_op_smax_vec: 3273 case INDEX_op_smin_vec: 3274 case INDEX_op_umax_vec: 3275 case INDEX_op_umin_vec: 3276 case INDEX_op_s390_vpks_vec: 3277 return C_O1_I2(v, v, v); 3278 case INDEX_op_rotls_vec: 3279 case INDEX_op_shls_vec: 3280 case INDEX_op_shrs_vec: 3281 case INDEX_op_sars_vec: 3282 return C_O1_I2(v, v, r); 3283 case INDEX_op_bitsel_vec: 3284 return C_O1_I3(v, v, v, v); 3285 3286 default: 3287 g_assert_not_reached(); 3288 } 3289} 3290 3291/* 3292 * Mainline glibc added HWCAP_S390_VX before it was kernel abi. 3293 * Some distros have fixed this up locally, others have not. 3294 */ 3295#ifndef HWCAP_S390_VXRS 3296#define HWCAP_S390_VXRS 2048 3297#endif 3298 3299static void query_s390_facilities(void) 3300{ 3301 unsigned long hwcap = qemu_getauxval(AT_HWCAP); 3302 const char *which; 3303 3304 /* Is STORE FACILITY LIST EXTENDED available? Honestly, I believe this 3305 is present on all 64-bit systems, but let's check for it anyway. */ 3306 if (hwcap & HWCAP_S390_STFLE) { 3307 register int r0 __asm__("0") = ARRAY_SIZE(s390_facilities) - 1; 3308 register void *r1 __asm__("1") = s390_facilities; 3309 3310 /* stfle 0(%r1) */ 3311 asm volatile(".word 0xb2b0,0x1000" 3312 : "=r"(r0) : "r"(r0), "r"(r1) : "memory", "cc"); 3313 } 3314 3315 /* 3316 * Use of vector registers requires os support beyond the facility bit. 3317 * If the kernel does not advertise support, disable the facility bits. 3318 * There is nothing else we currently care about in the 3rd word, so 3319 * disable VECTOR with one store. 3320 */ 3321 if (!(hwcap & HWCAP_S390_VXRS)) { 3322 s390_facilities[2] = 0; 3323 } 3324 3325 /* 3326 * Minimum supported cpu revision is z196. 3327 * Check for all required facilities. 3328 * ZARCH_ACTIVE is done via preprocessor check for 64-bit. 3329 */ 3330 if (!HAVE_FACILITY(LONG_DISP)) { 3331 which = "long-displacement"; 3332 goto fail; 3333 } 3334 if (!HAVE_FACILITY(EXT_IMM)) { 3335 which = "extended-immediate"; 3336 goto fail; 3337 } 3338 if (!HAVE_FACILITY(GEN_INST_EXT)) { 3339 which = "general-instructions-extension"; 3340 goto fail; 3341 } 3342 /* 3343 * Facility 45 is a big bin that contains: distinct-operands, 3344 * fast-BCR-serialization, high-word, population-count, 3345 * interlocked-access-1, and load/store-on-condition-1 3346 */ 3347 if (!HAVE_FACILITY(45)) { 3348 which = "45"; 3349 goto fail; 3350 } 3351 return; 3352 3353 fail: 3354 error_report("%s: missing required facility %s", __func__, which); 3355 exit(EXIT_FAILURE); 3356} 3357 3358static void tcg_target_init(TCGContext *s) 3359{ 3360 query_s390_facilities(); 3361 3362 tcg_target_available_regs[TCG_TYPE_I32] = 0xffff; 3363 tcg_target_available_regs[TCG_TYPE_I64] = 0xffff; 3364 if (HAVE_FACILITY(VECTOR)) { 3365 tcg_target_available_regs[TCG_TYPE_V64] = 0xffffffff00000000ull; 3366 tcg_target_available_regs[TCG_TYPE_V128] = 0xffffffff00000000ull; 3367 } 3368 3369 tcg_target_call_clobber_regs = 0; 3370 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R0); 3371 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R1); 3372 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R2); 3373 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R3); 3374 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R4); 3375 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R5); 3376 /* The r6 register is technically call-saved, but it's also a parameter 3377 register, so it can get killed by setup for the qemu_st helper. */ 3378 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R6); 3379 /* The return register can be considered call-clobbered. */ 3380 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_R14); 3381 3382 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V0); 3383 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V1); 3384 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V2); 3385 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V3); 3386 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V4); 3387 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V5); 3388 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V6); 3389 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V7); 3390 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V16); 3391 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V17); 3392 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V18); 3393 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V19); 3394 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V20); 3395 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V21); 3396 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V22); 3397 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V23); 3398 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V24); 3399 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V25); 3400 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V26); 3401 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V27); 3402 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V28); 3403 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V29); 3404 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V30); 3405 tcg_regset_set_reg(tcg_target_call_clobber_regs, TCG_REG_V31); 3406 3407 s->reserved_regs = 0; 3408 tcg_regset_set_reg(s->reserved_regs, TCG_TMP0); 3409 /* XXX many insns can't be used with R0, so we better avoid it for now */ 3410 tcg_regset_set_reg(s->reserved_regs, TCG_REG_R0); 3411 tcg_regset_set_reg(s->reserved_regs, TCG_REG_CALL_STACK); 3412} 3413 3414#define FRAME_SIZE ((int)(TCG_TARGET_CALL_STACK_OFFSET \ 3415 + TCG_STATIC_CALL_ARGS_SIZE \ 3416 + CPU_TEMP_BUF_NLONGS * sizeof(long))) 3417 3418static void tcg_target_qemu_prologue(TCGContext *s) 3419{ 3420 /* stmg %r6,%r15,48(%r15) (save registers) */ 3421 tcg_out_insn(s, RXY, STMG, TCG_REG_R6, TCG_REG_R15, TCG_REG_R15, 48); 3422 3423 /* aghi %r15,-frame_size */ 3424 tcg_out_insn(s, RI, AGHI, TCG_REG_R15, -FRAME_SIZE); 3425 3426 tcg_set_frame(s, TCG_REG_CALL_STACK, 3427 TCG_STATIC_CALL_ARGS_SIZE + TCG_TARGET_CALL_STACK_OFFSET, 3428 CPU_TEMP_BUF_NLONGS * sizeof(long)); 3429 3430#ifndef CONFIG_SOFTMMU 3431 if (guest_base >= 0x80000) { 3432 tcg_out_movi(s, TCG_TYPE_PTR, TCG_GUEST_BASE_REG, guest_base); 3433 tcg_regset_set_reg(s->reserved_regs, TCG_GUEST_BASE_REG); 3434 } 3435#endif 3436 3437 tcg_out_mov(s, TCG_TYPE_PTR, TCG_AREG0, tcg_target_call_iarg_regs[0]); 3438 3439 /* br %r3 (go to TB) */ 3440 tcg_out_insn(s, RR, BCR, S390_CC_ALWAYS, tcg_target_call_iarg_regs[1]); 3441 3442 /* 3443 * Return path for goto_ptr. Set return value to 0, a-la exit_tb, 3444 * and fall through to the rest of the epilogue. 3445 */ 3446 tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr); 3447 tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, 0); 3448 3449 /* TB epilogue */ 3450 tb_ret_addr = tcg_splitwx_to_rx(s->code_ptr); 3451 3452 /* lmg %r6,%r15,fs+48(%r15) (restore registers) */ 3453 tcg_out_insn(s, RXY, LMG, TCG_REG_R6, TCG_REG_R15, TCG_REG_R15, 3454 FRAME_SIZE + 48); 3455 3456 /* br %r14 (return) */ 3457 tcg_out_insn(s, RR, BCR, S390_CC_ALWAYS, TCG_REG_R14); 3458} 3459 3460static void tcg_out_nop_fill(tcg_insn_unit *p, int count) 3461{ 3462 memset(p, 0x07, count * sizeof(tcg_insn_unit)); 3463} 3464 3465typedef struct { 3466 DebugFrameHeader h; 3467 uint8_t fde_def_cfa[4]; 3468 uint8_t fde_reg_ofs[18]; 3469} DebugFrame; 3470 3471/* We're expecting a 2 byte uleb128 encoded value. */ 3472QEMU_BUILD_BUG_ON(FRAME_SIZE >= (1 << 14)); 3473 3474#define ELF_HOST_MACHINE EM_S390 3475 3476static const DebugFrame debug_frame = { 3477 .h.cie.len = sizeof(DebugFrameCIE)-4, /* length after .len member */ 3478 .h.cie.id = -1, 3479 .h.cie.version = 1, 3480 .h.cie.code_align = 1, 3481 .h.cie.data_align = 8, /* sleb128 8 */ 3482 .h.cie.return_column = TCG_REG_R14, 3483 3484 /* Total FDE size does not include the "len" member. */ 3485 .h.fde.len = sizeof(DebugFrame) - offsetof(DebugFrame, h.fde.cie_offset), 3486 3487 .fde_def_cfa = { 3488 12, TCG_REG_CALL_STACK, /* DW_CFA_def_cfa %r15, ... */ 3489 (FRAME_SIZE & 0x7f) | 0x80, /* ... uleb128 FRAME_SIZE */ 3490 (FRAME_SIZE >> 7) 3491 }, 3492 .fde_reg_ofs = { 3493 0x86, 6, /* DW_CFA_offset, %r6, 48 */ 3494 0x87, 7, /* DW_CFA_offset, %r7, 56 */ 3495 0x88, 8, /* DW_CFA_offset, %r8, 64 */ 3496 0x89, 9, /* DW_CFA_offset, %r92, 72 */ 3497 0x8a, 10, /* DW_CFA_offset, %r10, 80 */ 3498 0x8b, 11, /* DW_CFA_offset, %r11, 88 */ 3499 0x8c, 12, /* DW_CFA_offset, %r12, 96 */ 3500 0x8d, 13, /* DW_CFA_offset, %r13, 104 */ 3501 0x8e, 14, /* DW_CFA_offset, %r14, 112 */ 3502 } 3503}; 3504 3505void tcg_register_jit(const void *buf, size_t buf_size) 3506{ 3507 tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame)); 3508} 3509