/* * ARM SVE Operations * * Copyright (c) 2018 Linaro, Ltd. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, see . */ #include "qemu/osdep.h" #include "cpu.h" #include "internals.h" #include "exec/exec-all.h" #include "exec/cpu_ldst.h" #include "exec/helper-proto.h" #include "tcg/tcg-gvec-desc.h" #include "fpu/softfloat.h" #include "tcg/tcg.h" #include "vec_internal.h" /* Return a value for NZCV as per the ARM PredTest pseudofunction. * * The return value has bit 31 set if N is set, bit 1 set if Z is clear, * and bit 0 set if C is set. Compare the definitions of these variables * within CPUARMState. */ /* For no G bits set, NZCV = C. */ #define PREDTEST_INIT 1 /* This is an iterative function, called for each Pd and Pg word * moving forward. */ static uint32_t iter_predtest_fwd(uint64_t d, uint64_t g, uint32_t flags) { if (likely(g)) { /* Compute N from first D & G. Use bit 2 to signal first G bit seen. */ if (!(flags & 4)) { flags |= ((d & (g & -g)) != 0) << 31; flags |= 4; } /* Accumulate Z from each D & G. */ flags |= ((d & g) != 0) << 1; /* Compute C from last !(D & G). Replace previous. */ flags = deposit32(flags, 0, 1, (d & pow2floor(g)) == 0); } return flags; } /* This is an iterative function, called for each Pd and Pg word * moving backward. */ static uint32_t iter_predtest_bwd(uint64_t d, uint64_t g, uint32_t flags) { if (likely(g)) { /* Compute C from first (i.e last) !(D & G). Use bit 2 to signal first G bit seen. */ if (!(flags & 4)) { flags += 4 - 1; /* add bit 2, subtract C from PREDTEST_INIT */ flags |= (d & pow2floor(g)) == 0; } /* Accumulate Z from each D & G. */ flags |= ((d & g) != 0) << 1; /* Compute N from last (i.e first) D & G. Replace previous. */ flags = deposit32(flags, 31, 1, (d & (g & -g)) != 0); } return flags; } /* The same for a single word predicate. */ uint32_t HELPER(sve_predtest1)(uint64_t d, uint64_t g) { return iter_predtest_fwd(d, g, PREDTEST_INIT); } /* The same for a multi-word predicate. */ uint32_t HELPER(sve_predtest)(void *vd, void *vg, uint32_t words) { uint32_t flags = PREDTEST_INIT; uint64_t *d = vd, *g = vg; uintptr_t i = 0; do { flags = iter_predtest_fwd(d[i], g[i], flags); } while (++i < words); return flags; } /* * Expand active predicate bits to bytes, for byte elements. * (The data table itself is in vec_helper.c as MVE also needs it.) */ static inline uint64_t expand_pred_b(uint8_t byte) { return expand_pred_b_data[byte]; } /* Similarly for half-word elements. * for (i = 0; i < 256; ++i) { * unsigned long m = 0; * if (i & 0xaa) { * continue; * } * for (j = 0; j < 8; j += 2) { * if ((i >> j) & 1) { * m |= 0xfffful << (j << 3); * } * } * printf("[0x%x] = 0x%016lx,\n", i, m); * } */ static inline uint64_t expand_pred_h(uint8_t byte) { static const uint64_t word[] = { [0x01] = 0x000000000000ffff, [0x04] = 0x00000000ffff0000, [0x05] = 0x00000000ffffffff, [0x10] = 0x0000ffff00000000, [0x11] = 0x0000ffff0000ffff, [0x14] = 0x0000ffffffff0000, [0x15] = 0x0000ffffffffffff, [0x40] = 0xffff000000000000, [0x41] = 0xffff00000000ffff, [0x44] = 0xffff0000ffff0000, [0x45] = 0xffff0000ffffffff, [0x50] = 0xffffffff00000000, [0x51] = 0xffffffff0000ffff, [0x54] = 0xffffffffffff0000, [0x55] = 0xffffffffffffffff, }; return word[byte & 0x55]; } /* Similarly for single word elements. */ static inline uint64_t expand_pred_s(uint8_t byte) { static const uint64_t word[] = { [0x01] = 0x00000000ffffffffull, [0x10] = 0xffffffff00000000ull, [0x11] = 0xffffffffffffffffull, }; return word[byte & 0x11]; } #define LOGICAL_PPPP(NAME, FUNC) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ uintptr_t opr_sz = simd_oprsz(desc); \ uint64_t *d = vd, *n = vn, *m = vm, *g = vg; \ uintptr_t i; \ for (i = 0; i < opr_sz / 8; ++i) { \ d[i] = FUNC(n[i], m[i], g[i]); \ } \ } #define DO_AND(N, M, G) (((N) & (M)) & (G)) #define DO_BIC(N, M, G) (((N) & ~(M)) & (G)) #define DO_EOR(N, M, G) (((N) ^ (M)) & (G)) #define DO_ORR(N, M, G) (((N) | (M)) & (G)) #define DO_ORN(N, M, G) (((N) | ~(M)) & (G)) #define DO_NOR(N, M, G) (~((N) | (M)) & (G)) #define DO_NAND(N, M, G) (~((N) & (M)) & (G)) #define DO_SEL(N, M, G) (((N) & (G)) | ((M) & ~(G))) LOGICAL_PPPP(sve_and_pppp, DO_AND) LOGICAL_PPPP(sve_bic_pppp, DO_BIC) LOGICAL_PPPP(sve_eor_pppp, DO_EOR) LOGICAL_PPPP(sve_sel_pppp, DO_SEL) LOGICAL_PPPP(sve_orr_pppp, DO_ORR) LOGICAL_PPPP(sve_orn_pppp, DO_ORN) LOGICAL_PPPP(sve_nor_pppp, DO_NOR) LOGICAL_PPPP(sve_nand_pppp, DO_NAND) #undef DO_AND #undef DO_BIC #undef DO_EOR #undef DO_ORR #undef DO_ORN #undef DO_NOR #undef DO_NAND #undef DO_SEL #undef LOGICAL_PPPP /* Fully general three-operand expander, controlled by a predicate. * This is complicated by the host-endian storage of the register file. */ /* ??? I don't expect the compiler could ever vectorize this itself. * With some tables we can convert bit masks to byte masks, and with * extra care wrt byte/word ordering we could use gcc generic vectors * and do 16 bytes at a time. */ #define DO_ZPZZ(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ if (pg & 1) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ TYPE mm = *(TYPE *)(vm + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, mm); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ } /* Similarly, specialized for 64-bit operands. */ #define DO_ZPZZ_D(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / 8; \ TYPE *d = vd, *n = vn, *m = vm; \ uint8_t *pg = vg; \ for (i = 0; i < opr_sz; i += 1) { \ if (pg[H1(i)] & 1) { \ TYPE nn = n[i], mm = m[i]; \ d[i] = OP(nn, mm); \ } \ } \ } #define DO_AND(N, M) (N & M) #define DO_EOR(N, M) (N ^ M) #define DO_ORR(N, M) (N | M) #define DO_BIC(N, M) (N & ~M) #define DO_ADD(N, M) (N + M) #define DO_SUB(N, M) (N - M) #define DO_MAX(N, M) ((N) >= (M) ? (N) : (M)) #define DO_MIN(N, M) ((N) >= (M) ? (M) : (N)) #define DO_ABD(N, M) ((N) >= (M) ? (N) - (M) : (M) - (N)) #define DO_MUL(N, M) (N * M) /* * We must avoid the C undefined behaviour cases: division by * zero and signed division of INT_MIN by -1. Both of these * have architecturally defined required results for Arm. * We special case all signed divisions by -1 to avoid having * to deduce the minimum integer for the type involved. */ #define DO_SDIV(N, M) (unlikely(M == 0) ? 0 : unlikely(M == -1) ? -N : N / M) #define DO_UDIV(N, M) (unlikely(M == 0) ? 0 : N / M) DO_ZPZZ(sve_and_zpzz_b, uint8_t, H1, DO_AND) DO_ZPZZ(sve_and_zpzz_h, uint16_t, H1_2, DO_AND) DO_ZPZZ(sve_and_zpzz_s, uint32_t, H1_4, DO_AND) DO_ZPZZ_D(sve_and_zpzz_d, uint64_t, DO_AND) DO_ZPZZ(sve_orr_zpzz_b, uint8_t, H1, DO_ORR) DO_ZPZZ(sve_orr_zpzz_h, uint16_t, H1_2, DO_ORR) DO_ZPZZ(sve_orr_zpzz_s, uint32_t, H1_4, DO_ORR) DO_ZPZZ_D(sve_orr_zpzz_d, uint64_t, DO_ORR) DO_ZPZZ(sve_eor_zpzz_b, uint8_t, H1, DO_EOR) DO_ZPZZ(sve_eor_zpzz_h, uint16_t, H1_2, DO_EOR) DO_ZPZZ(sve_eor_zpzz_s, uint32_t, H1_4, DO_EOR) DO_ZPZZ_D(sve_eor_zpzz_d, uint64_t, DO_EOR) DO_ZPZZ(sve_bic_zpzz_b, uint8_t, H1, DO_BIC) DO_ZPZZ(sve_bic_zpzz_h, uint16_t, H1_2, DO_BIC) DO_ZPZZ(sve_bic_zpzz_s, uint32_t, H1_4, DO_BIC) DO_ZPZZ_D(sve_bic_zpzz_d, uint64_t, DO_BIC) DO_ZPZZ(sve_add_zpzz_b, uint8_t, H1, DO_ADD) DO_ZPZZ(sve_add_zpzz_h, uint16_t, H1_2, DO_ADD) DO_ZPZZ(sve_add_zpzz_s, uint32_t, H1_4, DO_ADD) DO_ZPZZ_D(sve_add_zpzz_d, uint64_t, DO_ADD) DO_ZPZZ(sve_sub_zpzz_b, uint8_t, H1, DO_SUB) DO_ZPZZ(sve_sub_zpzz_h, uint16_t, H1_2, DO_SUB) DO_ZPZZ(sve_sub_zpzz_s, uint32_t, H1_4, DO_SUB) DO_ZPZZ_D(sve_sub_zpzz_d, uint64_t, DO_SUB) DO_ZPZZ(sve_smax_zpzz_b, int8_t, H1, DO_MAX) DO_ZPZZ(sve_smax_zpzz_h, int16_t, H1_2, DO_MAX) DO_ZPZZ(sve_smax_zpzz_s, int32_t, H1_4, DO_MAX) DO_ZPZZ_D(sve_smax_zpzz_d, int64_t, DO_MAX) DO_ZPZZ(sve_umax_zpzz_b, uint8_t, H1, DO_MAX) DO_ZPZZ(sve_umax_zpzz_h, uint16_t, H1_2, DO_MAX) DO_ZPZZ(sve_umax_zpzz_s, uint32_t, H1_4, DO_MAX) DO_ZPZZ_D(sve_umax_zpzz_d, uint64_t, DO_MAX) DO_ZPZZ(sve_smin_zpzz_b, int8_t, H1, DO_MIN) DO_ZPZZ(sve_smin_zpzz_h, int16_t, H1_2, DO_MIN) DO_ZPZZ(sve_smin_zpzz_s, int32_t, H1_4, DO_MIN) DO_ZPZZ_D(sve_smin_zpzz_d, int64_t, DO_MIN) DO_ZPZZ(sve_umin_zpzz_b, uint8_t, H1, DO_MIN) DO_ZPZZ(sve_umin_zpzz_h, uint16_t, H1_2, DO_MIN) DO_ZPZZ(sve_umin_zpzz_s, uint32_t, H1_4, DO_MIN) DO_ZPZZ_D(sve_umin_zpzz_d, uint64_t, DO_MIN) DO_ZPZZ(sve_sabd_zpzz_b, int8_t, H1, DO_ABD) DO_ZPZZ(sve_sabd_zpzz_h, int16_t, H1_2, DO_ABD) DO_ZPZZ(sve_sabd_zpzz_s, int32_t, H1_4, DO_ABD) DO_ZPZZ_D(sve_sabd_zpzz_d, int64_t, DO_ABD) DO_ZPZZ(sve_uabd_zpzz_b, uint8_t, H1, DO_ABD) DO_ZPZZ(sve_uabd_zpzz_h, uint16_t, H1_2, DO_ABD) DO_ZPZZ(sve_uabd_zpzz_s, uint32_t, H1_4, DO_ABD) DO_ZPZZ_D(sve_uabd_zpzz_d, uint64_t, DO_ABD) /* Because the computation type is at least twice as large as required, these work for both signed and unsigned source types. */ static inline uint8_t do_mulh_b(int32_t n, int32_t m) { return (n * m) >> 8; } static inline uint16_t do_mulh_h(int32_t n, int32_t m) { return (n * m) >> 16; } static inline uint32_t do_mulh_s(int64_t n, int64_t m) { return (n * m) >> 32; } static inline uint64_t do_smulh_d(uint64_t n, uint64_t m) { uint64_t lo, hi; muls64(&lo, &hi, n, m); return hi; } static inline uint64_t do_umulh_d(uint64_t n, uint64_t m) { uint64_t lo, hi; mulu64(&lo, &hi, n, m); return hi; } DO_ZPZZ(sve_mul_zpzz_b, uint8_t, H1, DO_MUL) DO_ZPZZ(sve_mul_zpzz_h, uint16_t, H1_2, DO_MUL) DO_ZPZZ(sve_mul_zpzz_s, uint32_t, H1_4, DO_MUL) DO_ZPZZ_D(sve_mul_zpzz_d, uint64_t, DO_MUL) DO_ZPZZ(sve_smulh_zpzz_b, int8_t, H1, do_mulh_b) DO_ZPZZ(sve_smulh_zpzz_h, int16_t, H1_2, do_mulh_h) DO_ZPZZ(sve_smulh_zpzz_s, int32_t, H1_4, do_mulh_s) DO_ZPZZ_D(sve_smulh_zpzz_d, uint64_t, do_smulh_d) DO_ZPZZ(sve_umulh_zpzz_b, uint8_t, H1, do_mulh_b) DO_ZPZZ(sve_umulh_zpzz_h, uint16_t, H1_2, do_mulh_h) DO_ZPZZ(sve_umulh_zpzz_s, uint32_t, H1_4, do_mulh_s) DO_ZPZZ_D(sve_umulh_zpzz_d, uint64_t, do_umulh_d) DO_ZPZZ(sve_sdiv_zpzz_s, int32_t, H1_4, DO_SDIV) DO_ZPZZ_D(sve_sdiv_zpzz_d, int64_t, DO_SDIV) DO_ZPZZ(sve_udiv_zpzz_s, uint32_t, H1_4, DO_UDIV) DO_ZPZZ_D(sve_udiv_zpzz_d, uint64_t, DO_UDIV) /* Note that all bits of the shift are significant and not modulo the element size. */ #define DO_ASR(N, M) (N >> MIN(M, sizeof(N) * 8 - 1)) #define DO_LSR(N, M) (M < sizeof(N) * 8 ? N >> M : 0) #define DO_LSL(N, M) (M < sizeof(N) * 8 ? N << M : 0) DO_ZPZZ(sve_asr_zpzz_b, int8_t, H1, DO_ASR) DO_ZPZZ(sve_lsr_zpzz_b, uint8_t, H1_2, DO_LSR) DO_ZPZZ(sve_lsl_zpzz_b, uint8_t, H1_4, DO_LSL) DO_ZPZZ(sve_asr_zpzz_h, int16_t, H1, DO_ASR) DO_ZPZZ(sve_lsr_zpzz_h, uint16_t, H1_2, DO_LSR) DO_ZPZZ(sve_lsl_zpzz_h, uint16_t, H1_4, DO_LSL) DO_ZPZZ(sve_asr_zpzz_s, int32_t, H1, DO_ASR) DO_ZPZZ(sve_lsr_zpzz_s, uint32_t, H1_2, DO_LSR) DO_ZPZZ(sve_lsl_zpzz_s, uint32_t, H1_4, DO_LSL) DO_ZPZZ_D(sve_asr_zpzz_d, int64_t, DO_ASR) DO_ZPZZ_D(sve_lsr_zpzz_d, uint64_t, DO_LSR) DO_ZPZZ_D(sve_lsl_zpzz_d, uint64_t, DO_LSL) static inline uint16_t do_sadalp_h(int16_t n, int16_t m) { int8_t n1 = n, n2 = n >> 8; return m + n1 + n2; } static inline uint32_t do_sadalp_s(int32_t n, int32_t m) { int16_t n1 = n, n2 = n >> 16; return m + n1 + n2; } static inline uint64_t do_sadalp_d(int64_t n, int64_t m) { int32_t n1 = n, n2 = n >> 32; return m + n1 + n2; } DO_ZPZZ(sve2_sadalp_zpzz_h, int16_t, H1_2, do_sadalp_h) DO_ZPZZ(sve2_sadalp_zpzz_s, int32_t, H1_4, do_sadalp_s) DO_ZPZZ_D(sve2_sadalp_zpzz_d, int64_t, do_sadalp_d) static inline uint16_t do_uadalp_h(uint16_t n, uint16_t m) { uint8_t n1 = n, n2 = n >> 8; return m + n1 + n2; } static inline uint32_t do_uadalp_s(uint32_t n, uint32_t m) { uint16_t n1 = n, n2 = n >> 16; return m + n1 + n2; } static inline uint64_t do_uadalp_d(uint64_t n, uint64_t m) { uint32_t n1 = n, n2 = n >> 32; return m + n1 + n2; } DO_ZPZZ(sve2_uadalp_zpzz_h, uint16_t, H1_2, do_uadalp_h) DO_ZPZZ(sve2_uadalp_zpzz_s, uint32_t, H1_4, do_uadalp_s) DO_ZPZZ_D(sve2_uadalp_zpzz_d, uint64_t, do_uadalp_d) #define do_srshl_b(n, m) do_sqrshl_bhs(n, m, 8, true, NULL) #define do_srshl_h(n, m) do_sqrshl_bhs(n, m, 16, true, NULL) #define do_srshl_s(n, m) do_sqrshl_bhs(n, m, 32, true, NULL) #define do_srshl_d(n, m) do_sqrshl_d(n, m, true, NULL) DO_ZPZZ(sve2_srshl_zpzz_b, int8_t, H1, do_srshl_b) DO_ZPZZ(sve2_srshl_zpzz_h, int16_t, H1_2, do_srshl_h) DO_ZPZZ(sve2_srshl_zpzz_s, int32_t, H1_4, do_srshl_s) DO_ZPZZ_D(sve2_srshl_zpzz_d, int64_t, do_srshl_d) #define do_urshl_b(n, m) do_uqrshl_bhs(n, (int8_t)m, 8, true, NULL) #define do_urshl_h(n, m) do_uqrshl_bhs(n, (int16_t)m, 16, true, NULL) #define do_urshl_s(n, m) do_uqrshl_bhs(n, m, 32, true, NULL) #define do_urshl_d(n, m) do_uqrshl_d(n, m, true, NULL) DO_ZPZZ(sve2_urshl_zpzz_b, uint8_t, H1, do_urshl_b) DO_ZPZZ(sve2_urshl_zpzz_h, uint16_t, H1_2, do_urshl_h) DO_ZPZZ(sve2_urshl_zpzz_s, uint32_t, H1_4, do_urshl_s) DO_ZPZZ_D(sve2_urshl_zpzz_d, uint64_t, do_urshl_d) /* * Unlike the NEON and AdvSIMD versions, there is no QC bit to set. * We pass in a pointer to a dummy saturation field to trigger * the saturating arithmetic but discard the information about * whether it has occurred. */ #define do_sqshl_b(n, m) \ ({ uint32_t discard; do_sqrshl_bhs(n, m, 8, false, &discard); }) #define do_sqshl_h(n, m) \ ({ uint32_t discard; do_sqrshl_bhs(n, m, 16, false, &discard); }) #define do_sqshl_s(n, m) \ ({ uint32_t discard; do_sqrshl_bhs(n, m, 32, false, &discard); }) #define do_sqshl_d(n, m) \ ({ uint32_t discard; do_sqrshl_d(n, m, false, &discard); }) DO_ZPZZ(sve2_sqshl_zpzz_b, int8_t, H1_2, do_sqshl_b) DO_ZPZZ(sve2_sqshl_zpzz_h, int16_t, H1_2, do_sqshl_h) DO_ZPZZ(sve2_sqshl_zpzz_s, int32_t, H1_4, do_sqshl_s) DO_ZPZZ_D(sve2_sqshl_zpzz_d, int64_t, do_sqshl_d) #define do_uqshl_b(n, m) \ ({ uint32_t discard; do_uqrshl_bhs(n, (int8_t)m, 8, false, &discard); }) #define do_uqshl_h(n, m) \ ({ uint32_t discard; do_uqrshl_bhs(n, (int16_t)m, 16, false, &discard); }) #define do_uqshl_s(n, m) \ ({ uint32_t discard; do_uqrshl_bhs(n, m, 32, false, &discard); }) #define do_uqshl_d(n, m) \ ({ uint32_t discard; do_uqrshl_d(n, m, false, &discard); }) DO_ZPZZ(sve2_uqshl_zpzz_b, uint8_t, H1_2, do_uqshl_b) DO_ZPZZ(sve2_uqshl_zpzz_h, uint16_t, H1_2, do_uqshl_h) DO_ZPZZ(sve2_uqshl_zpzz_s, uint32_t, H1_4, do_uqshl_s) DO_ZPZZ_D(sve2_uqshl_zpzz_d, uint64_t, do_uqshl_d) #define do_sqrshl_b(n, m) \ ({ uint32_t discard; do_sqrshl_bhs(n, m, 8, true, &discard); }) #define do_sqrshl_h(n, m) \ ({ uint32_t discard; do_sqrshl_bhs(n, m, 16, true, &discard); }) #define do_sqrshl_s(n, m) \ ({ uint32_t discard; do_sqrshl_bhs(n, m, 32, true, &discard); }) #define do_sqrshl_d(n, m) \ ({ uint32_t discard; do_sqrshl_d(n, m, true, &discard); }) DO_ZPZZ(sve2_sqrshl_zpzz_b, int8_t, H1_2, do_sqrshl_b) DO_ZPZZ(sve2_sqrshl_zpzz_h, int16_t, H1_2, do_sqrshl_h) DO_ZPZZ(sve2_sqrshl_zpzz_s, int32_t, H1_4, do_sqrshl_s) DO_ZPZZ_D(sve2_sqrshl_zpzz_d, int64_t, do_sqrshl_d) #undef do_sqrshl_d #define do_uqrshl_b(n, m) \ ({ uint32_t discard; do_uqrshl_bhs(n, (int8_t)m, 8, true, &discard); }) #define do_uqrshl_h(n, m) \ ({ uint32_t discard; do_uqrshl_bhs(n, (int16_t)m, 16, true, &discard); }) #define do_uqrshl_s(n, m) \ ({ uint32_t discard; do_uqrshl_bhs(n, m, 32, true, &discard); }) #define do_uqrshl_d(n, m) \ ({ uint32_t discard; do_uqrshl_d(n, m, true, &discard); }) DO_ZPZZ(sve2_uqrshl_zpzz_b, uint8_t, H1_2, do_uqrshl_b) DO_ZPZZ(sve2_uqrshl_zpzz_h, uint16_t, H1_2, do_uqrshl_h) DO_ZPZZ(sve2_uqrshl_zpzz_s, uint32_t, H1_4, do_uqrshl_s) DO_ZPZZ_D(sve2_uqrshl_zpzz_d, uint64_t, do_uqrshl_d) #undef do_uqrshl_d #define DO_HADD_BHS(n, m) (((int64_t)n + m) >> 1) #define DO_HADD_D(n, m) ((n >> 1) + (m >> 1) + (n & m & 1)) DO_ZPZZ(sve2_shadd_zpzz_b, int8_t, H1, DO_HADD_BHS) DO_ZPZZ(sve2_shadd_zpzz_h, int16_t, H1_2, DO_HADD_BHS) DO_ZPZZ(sve2_shadd_zpzz_s, int32_t, H1_4, DO_HADD_BHS) DO_ZPZZ_D(sve2_shadd_zpzz_d, int64_t, DO_HADD_D) DO_ZPZZ(sve2_uhadd_zpzz_b, uint8_t, H1, DO_HADD_BHS) DO_ZPZZ(sve2_uhadd_zpzz_h, uint16_t, H1_2, DO_HADD_BHS) DO_ZPZZ(sve2_uhadd_zpzz_s, uint32_t, H1_4, DO_HADD_BHS) DO_ZPZZ_D(sve2_uhadd_zpzz_d, uint64_t, DO_HADD_D) #define DO_RHADD_BHS(n, m) (((int64_t)n + m + 1) >> 1) #define DO_RHADD_D(n, m) ((n >> 1) + (m >> 1) + ((n | m) & 1)) DO_ZPZZ(sve2_srhadd_zpzz_b, int8_t, H1, DO_RHADD_BHS) DO_ZPZZ(sve2_srhadd_zpzz_h, int16_t, H1_2, DO_RHADD_BHS) DO_ZPZZ(sve2_srhadd_zpzz_s, int32_t, H1_4, DO_RHADD_BHS) DO_ZPZZ_D(sve2_srhadd_zpzz_d, int64_t, DO_RHADD_D) DO_ZPZZ(sve2_urhadd_zpzz_b, uint8_t, H1, DO_RHADD_BHS) DO_ZPZZ(sve2_urhadd_zpzz_h, uint16_t, H1_2, DO_RHADD_BHS) DO_ZPZZ(sve2_urhadd_zpzz_s, uint32_t, H1_4, DO_RHADD_BHS) DO_ZPZZ_D(sve2_urhadd_zpzz_d, uint64_t, DO_RHADD_D) #define DO_HSUB_BHS(n, m) (((int64_t)n - m) >> 1) #define DO_HSUB_D(n, m) ((n >> 1) - (m >> 1) - (~n & m & 1)) DO_ZPZZ(sve2_shsub_zpzz_b, int8_t, H1, DO_HSUB_BHS) DO_ZPZZ(sve2_shsub_zpzz_h, int16_t, H1_2, DO_HSUB_BHS) DO_ZPZZ(sve2_shsub_zpzz_s, int32_t, H1_4, DO_HSUB_BHS) DO_ZPZZ_D(sve2_shsub_zpzz_d, int64_t, DO_HSUB_D) DO_ZPZZ(sve2_uhsub_zpzz_b, uint8_t, H1, DO_HSUB_BHS) DO_ZPZZ(sve2_uhsub_zpzz_h, uint16_t, H1_2, DO_HSUB_BHS) DO_ZPZZ(sve2_uhsub_zpzz_s, uint32_t, H1_4, DO_HSUB_BHS) DO_ZPZZ_D(sve2_uhsub_zpzz_d, uint64_t, DO_HSUB_D) static inline int32_t do_sat_bhs(int64_t val, int64_t min, int64_t max) { return val >= max ? max : val <= min ? min : val; } #define DO_SQADD_B(n, m) do_sat_bhs((int64_t)n + m, INT8_MIN, INT8_MAX) #define DO_SQADD_H(n, m) do_sat_bhs((int64_t)n + m, INT16_MIN, INT16_MAX) #define DO_SQADD_S(n, m) do_sat_bhs((int64_t)n + m, INT32_MIN, INT32_MAX) static inline int64_t do_sqadd_d(int64_t n, int64_t m) { int64_t r = n + m; if (((r ^ n) & ~(n ^ m)) < 0) { /* Signed overflow. */ return r < 0 ? INT64_MAX : INT64_MIN; } return r; } DO_ZPZZ(sve2_sqadd_zpzz_b, int8_t, H1, DO_SQADD_B) DO_ZPZZ(sve2_sqadd_zpzz_h, int16_t, H1_2, DO_SQADD_H) DO_ZPZZ(sve2_sqadd_zpzz_s, int32_t, H1_4, DO_SQADD_S) DO_ZPZZ_D(sve2_sqadd_zpzz_d, int64_t, do_sqadd_d) #define DO_UQADD_B(n, m) do_sat_bhs((int64_t)n + m, 0, UINT8_MAX) #define DO_UQADD_H(n, m) do_sat_bhs((int64_t)n + m, 0, UINT16_MAX) #define DO_UQADD_S(n, m) do_sat_bhs((int64_t)n + m, 0, UINT32_MAX) static inline uint64_t do_uqadd_d(uint64_t n, uint64_t m) { uint64_t r = n + m; return r < n ? UINT64_MAX : r; } DO_ZPZZ(sve2_uqadd_zpzz_b, uint8_t, H1, DO_UQADD_B) DO_ZPZZ(sve2_uqadd_zpzz_h, uint16_t, H1_2, DO_UQADD_H) DO_ZPZZ(sve2_uqadd_zpzz_s, uint32_t, H1_4, DO_UQADD_S) DO_ZPZZ_D(sve2_uqadd_zpzz_d, uint64_t, do_uqadd_d) #define DO_SQSUB_B(n, m) do_sat_bhs((int64_t)n - m, INT8_MIN, INT8_MAX) #define DO_SQSUB_H(n, m) do_sat_bhs((int64_t)n - m, INT16_MIN, INT16_MAX) #define DO_SQSUB_S(n, m) do_sat_bhs((int64_t)n - m, INT32_MIN, INT32_MAX) static inline int64_t do_sqsub_d(int64_t n, int64_t m) { int64_t r = n - m; if (((r ^ n) & (n ^ m)) < 0) { /* Signed overflow. */ return r < 0 ? INT64_MAX : INT64_MIN; } return r; } DO_ZPZZ(sve2_sqsub_zpzz_b, int8_t, H1, DO_SQSUB_B) DO_ZPZZ(sve2_sqsub_zpzz_h, int16_t, H1_2, DO_SQSUB_H) DO_ZPZZ(sve2_sqsub_zpzz_s, int32_t, H1_4, DO_SQSUB_S) DO_ZPZZ_D(sve2_sqsub_zpzz_d, int64_t, do_sqsub_d) #define DO_UQSUB_B(n, m) do_sat_bhs((int64_t)n - m, 0, UINT8_MAX) #define DO_UQSUB_H(n, m) do_sat_bhs((int64_t)n - m, 0, UINT16_MAX) #define DO_UQSUB_S(n, m) do_sat_bhs((int64_t)n - m, 0, UINT32_MAX) static inline uint64_t do_uqsub_d(uint64_t n, uint64_t m) { return n > m ? n - m : 0; } DO_ZPZZ(sve2_uqsub_zpzz_b, uint8_t, H1, DO_UQSUB_B) DO_ZPZZ(sve2_uqsub_zpzz_h, uint16_t, H1_2, DO_UQSUB_H) DO_ZPZZ(sve2_uqsub_zpzz_s, uint32_t, H1_4, DO_UQSUB_S) DO_ZPZZ_D(sve2_uqsub_zpzz_d, uint64_t, do_uqsub_d) #define DO_SUQADD_B(n, m) \ do_sat_bhs((int64_t)(int8_t)n + m, INT8_MIN, INT8_MAX) #define DO_SUQADD_H(n, m) \ do_sat_bhs((int64_t)(int16_t)n + m, INT16_MIN, INT16_MAX) #define DO_SUQADD_S(n, m) \ do_sat_bhs((int64_t)(int32_t)n + m, INT32_MIN, INT32_MAX) static inline int64_t do_suqadd_d(int64_t n, uint64_t m) { uint64_t r = n + m; if (n < 0) { /* Note that m - abs(n) cannot underflow. */ if (r > INT64_MAX) { /* Result is either very large positive or negative. */ if (m > -n) { /* m > abs(n), so r is a very large positive. */ return INT64_MAX; } /* Result is negative. */ } } else { /* Both inputs are positive: check for overflow. */ if (r < m || r > INT64_MAX) { return INT64_MAX; } } return r; } DO_ZPZZ(sve2_suqadd_zpzz_b, uint8_t, H1, DO_SUQADD_B) DO_ZPZZ(sve2_suqadd_zpzz_h, uint16_t, H1_2, DO_SUQADD_H) DO_ZPZZ(sve2_suqadd_zpzz_s, uint32_t, H1_4, DO_SUQADD_S) DO_ZPZZ_D(sve2_suqadd_zpzz_d, uint64_t, do_suqadd_d) #define DO_USQADD_B(n, m) \ do_sat_bhs((int64_t)n + (int8_t)m, 0, UINT8_MAX) #define DO_USQADD_H(n, m) \ do_sat_bhs((int64_t)n + (int16_t)m, 0, UINT16_MAX) #define DO_USQADD_S(n, m) \ do_sat_bhs((int64_t)n + (int32_t)m, 0, UINT32_MAX) static inline uint64_t do_usqadd_d(uint64_t n, int64_t m) { uint64_t r = n + m; if (m < 0) { return n < -m ? 0 : r; } return r < n ? UINT64_MAX : r; } DO_ZPZZ(sve2_usqadd_zpzz_b, uint8_t, H1, DO_USQADD_B) DO_ZPZZ(sve2_usqadd_zpzz_h, uint16_t, H1_2, DO_USQADD_H) DO_ZPZZ(sve2_usqadd_zpzz_s, uint32_t, H1_4, DO_USQADD_S) DO_ZPZZ_D(sve2_usqadd_zpzz_d, uint64_t, do_usqadd_d) #undef DO_ZPZZ #undef DO_ZPZZ_D /* * Three operand expander, operating on element pairs. * If the slot I is even, the elements from from VN {I, I+1}. * If the slot I is odd, the elements from from VM {I-1, I}. * Load all of the input elements in each pair before overwriting output. */ #define DO_ZPZZ_PAIR(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ TYPE n0 = *(TYPE *)(vn + H(i)); \ TYPE m0 = *(TYPE *)(vm + H(i)); \ TYPE n1 = *(TYPE *)(vn + H(i + sizeof(TYPE))); \ TYPE m1 = *(TYPE *)(vm + H(i + sizeof(TYPE))); \ if (pg & 1) { \ *(TYPE *)(vd + H(i)) = OP(n0, n1); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ if (pg & 1) { \ *(TYPE *)(vd + H(i)) = OP(m0, m1); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ } /* Similarly, specialized for 64-bit operands. */ #define DO_ZPZZ_PAIR_D(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / 8; \ TYPE *d = vd, *n = vn, *m = vm; \ uint8_t *pg = vg; \ for (i = 0; i < opr_sz; i += 2) { \ TYPE n0 = n[i], n1 = n[i + 1]; \ TYPE m0 = m[i], m1 = m[i + 1]; \ if (pg[H1(i)] & 1) { \ d[i] = OP(n0, n1); \ } \ if (pg[H1(i + 1)] & 1) { \ d[i + 1] = OP(m0, m1); \ } \ } \ } DO_ZPZZ_PAIR(sve2_addp_zpzz_b, uint8_t, H1, DO_ADD) DO_ZPZZ_PAIR(sve2_addp_zpzz_h, uint16_t, H1_2, DO_ADD) DO_ZPZZ_PAIR(sve2_addp_zpzz_s, uint32_t, H1_4, DO_ADD) DO_ZPZZ_PAIR_D(sve2_addp_zpzz_d, uint64_t, DO_ADD) DO_ZPZZ_PAIR(sve2_umaxp_zpzz_b, uint8_t, H1, DO_MAX) DO_ZPZZ_PAIR(sve2_umaxp_zpzz_h, uint16_t, H1_2, DO_MAX) DO_ZPZZ_PAIR(sve2_umaxp_zpzz_s, uint32_t, H1_4, DO_MAX) DO_ZPZZ_PAIR_D(sve2_umaxp_zpzz_d, uint64_t, DO_MAX) DO_ZPZZ_PAIR(sve2_uminp_zpzz_b, uint8_t, H1, DO_MIN) DO_ZPZZ_PAIR(sve2_uminp_zpzz_h, uint16_t, H1_2, DO_MIN) DO_ZPZZ_PAIR(sve2_uminp_zpzz_s, uint32_t, H1_4, DO_MIN) DO_ZPZZ_PAIR_D(sve2_uminp_zpzz_d, uint64_t, DO_MIN) DO_ZPZZ_PAIR(sve2_smaxp_zpzz_b, int8_t, H1, DO_MAX) DO_ZPZZ_PAIR(sve2_smaxp_zpzz_h, int16_t, H1_2, DO_MAX) DO_ZPZZ_PAIR(sve2_smaxp_zpzz_s, int32_t, H1_4, DO_MAX) DO_ZPZZ_PAIR_D(sve2_smaxp_zpzz_d, int64_t, DO_MAX) DO_ZPZZ_PAIR(sve2_sminp_zpzz_b, int8_t, H1, DO_MIN) DO_ZPZZ_PAIR(sve2_sminp_zpzz_h, int16_t, H1_2, DO_MIN) DO_ZPZZ_PAIR(sve2_sminp_zpzz_s, int32_t, H1_4, DO_MIN) DO_ZPZZ_PAIR_D(sve2_sminp_zpzz_d, int64_t, DO_MIN) #undef DO_ZPZZ_PAIR #undef DO_ZPZZ_PAIR_D #define DO_ZPZZ_PAIR_FP(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, \ void *status, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ TYPE n0 = *(TYPE *)(vn + H(i)); \ TYPE m0 = *(TYPE *)(vm + H(i)); \ TYPE n1 = *(TYPE *)(vn + H(i + sizeof(TYPE))); \ TYPE m1 = *(TYPE *)(vm + H(i + sizeof(TYPE))); \ if (pg & 1) { \ *(TYPE *)(vd + H(i)) = OP(n0, n1, status); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ if (pg & 1) { \ *(TYPE *)(vd + H(i)) = OP(m0, m1, status); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ } DO_ZPZZ_PAIR_FP(sve2_faddp_zpzz_h, float16, H1_2, float16_add) DO_ZPZZ_PAIR_FP(sve2_faddp_zpzz_s, float32, H1_4, float32_add) DO_ZPZZ_PAIR_FP(sve2_faddp_zpzz_d, float64, H1_8, float64_add) DO_ZPZZ_PAIR_FP(sve2_fmaxnmp_zpzz_h, float16, H1_2, float16_maxnum) DO_ZPZZ_PAIR_FP(sve2_fmaxnmp_zpzz_s, float32, H1_4, float32_maxnum) DO_ZPZZ_PAIR_FP(sve2_fmaxnmp_zpzz_d, float64, H1_8, float64_maxnum) DO_ZPZZ_PAIR_FP(sve2_fminnmp_zpzz_h, float16, H1_2, float16_minnum) DO_ZPZZ_PAIR_FP(sve2_fminnmp_zpzz_s, float32, H1_4, float32_minnum) DO_ZPZZ_PAIR_FP(sve2_fminnmp_zpzz_d, float64, H1_8, float64_minnum) DO_ZPZZ_PAIR_FP(sve2_fmaxp_zpzz_h, float16, H1_2, float16_max) DO_ZPZZ_PAIR_FP(sve2_fmaxp_zpzz_s, float32, H1_4, float32_max) DO_ZPZZ_PAIR_FP(sve2_fmaxp_zpzz_d, float64, H1_8, float64_max) DO_ZPZZ_PAIR_FP(sve2_fminp_zpzz_h, float16, H1_2, float16_min) DO_ZPZZ_PAIR_FP(sve2_fminp_zpzz_s, float32, H1_4, float32_min) DO_ZPZZ_PAIR_FP(sve2_fminp_zpzz_d, float64, H1_8, float64_min) #undef DO_ZPZZ_PAIR_FP /* Three-operand expander, controlled by a predicate, in which the * third operand is "wide". That is, for D = N op M, the same 64-bit * value of M is used with all of the narrower values of N. */ #define DO_ZPZW(NAME, TYPE, TYPEW, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ uint8_t pg = *(uint8_t *)(vg + H1(i >> 3)); \ TYPEW mm = *(TYPEW *)(vm + i); \ do { \ if (pg & 1) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, mm); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 7); \ } \ } DO_ZPZW(sve_asr_zpzw_b, int8_t, uint64_t, H1, DO_ASR) DO_ZPZW(sve_lsr_zpzw_b, uint8_t, uint64_t, H1, DO_LSR) DO_ZPZW(sve_lsl_zpzw_b, uint8_t, uint64_t, H1, DO_LSL) DO_ZPZW(sve_asr_zpzw_h, int16_t, uint64_t, H1_2, DO_ASR) DO_ZPZW(sve_lsr_zpzw_h, uint16_t, uint64_t, H1_2, DO_LSR) DO_ZPZW(sve_lsl_zpzw_h, uint16_t, uint64_t, H1_2, DO_LSL) DO_ZPZW(sve_asr_zpzw_s, int32_t, uint64_t, H1_4, DO_ASR) DO_ZPZW(sve_lsr_zpzw_s, uint32_t, uint64_t, H1_4, DO_LSR) DO_ZPZW(sve_lsl_zpzw_s, uint32_t, uint64_t, H1_4, DO_LSL) #undef DO_ZPZW /* Fully general two-operand expander, controlled by a predicate. */ #define DO_ZPZ(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ if (pg & 1) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ } /* Similarly, specialized for 64-bit operands. */ #define DO_ZPZ_D(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / 8; \ TYPE *d = vd, *n = vn; \ uint8_t *pg = vg; \ for (i = 0; i < opr_sz; i += 1) { \ if (pg[H1(i)] & 1) { \ TYPE nn = n[i]; \ d[i] = OP(nn); \ } \ } \ } #define DO_CLS_B(N) (clrsb32(N) - 24) #define DO_CLS_H(N) (clrsb32(N) - 16) DO_ZPZ(sve_cls_b, int8_t, H1, DO_CLS_B) DO_ZPZ(sve_cls_h, int16_t, H1_2, DO_CLS_H) DO_ZPZ(sve_cls_s, int32_t, H1_4, clrsb32) DO_ZPZ_D(sve_cls_d, int64_t, clrsb64) #define DO_CLZ_B(N) (clz32(N) - 24) #define DO_CLZ_H(N) (clz32(N) - 16) DO_ZPZ(sve_clz_b, uint8_t, H1, DO_CLZ_B) DO_ZPZ(sve_clz_h, uint16_t, H1_2, DO_CLZ_H) DO_ZPZ(sve_clz_s, uint32_t, H1_4, clz32) DO_ZPZ_D(sve_clz_d, uint64_t, clz64) DO_ZPZ(sve_cnt_zpz_b, uint8_t, H1, ctpop8) DO_ZPZ(sve_cnt_zpz_h, uint16_t, H1_2, ctpop16) DO_ZPZ(sve_cnt_zpz_s, uint32_t, H1_4, ctpop32) DO_ZPZ_D(sve_cnt_zpz_d, uint64_t, ctpop64) #define DO_CNOT(N) (N == 0) DO_ZPZ(sve_cnot_b, uint8_t, H1, DO_CNOT) DO_ZPZ(sve_cnot_h, uint16_t, H1_2, DO_CNOT) DO_ZPZ(sve_cnot_s, uint32_t, H1_4, DO_CNOT) DO_ZPZ_D(sve_cnot_d, uint64_t, DO_CNOT) #define DO_FABS(N) (N & ((__typeof(N))-1 >> 1)) DO_ZPZ(sve_fabs_h, uint16_t, H1_2, DO_FABS) DO_ZPZ(sve_fabs_s, uint32_t, H1_4, DO_FABS) DO_ZPZ_D(sve_fabs_d, uint64_t, DO_FABS) #define DO_FNEG(N) (N ^ ~((__typeof(N))-1 >> 1)) DO_ZPZ(sve_fneg_h, uint16_t, H1_2, DO_FNEG) DO_ZPZ(sve_fneg_s, uint32_t, H1_4, DO_FNEG) DO_ZPZ_D(sve_fneg_d, uint64_t, DO_FNEG) #define DO_NOT(N) (~N) DO_ZPZ(sve_not_zpz_b, uint8_t, H1, DO_NOT) DO_ZPZ(sve_not_zpz_h, uint16_t, H1_2, DO_NOT) DO_ZPZ(sve_not_zpz_s, uint32_t, H1_4, DO_NOT) DO_ZPZ_D(sve_not_zpz_d, uint64_t, DO_NOT) #define DO_SXTB(N) ((int8_t)N) #define DO_SXTH(N) ((int16_t)N) #define DO_SXTS(N) ((int32_t)N) #define DO_UXTB(N) ((uint8_t)N) #define DO_UXTH(N) ((uint16_t)N) #define DO_UXTS(N) ((uint32_t)N) DO_ZPZ(sve_sxtb_h, uint16_t, H1_2, DO_SXTB) DO_ZPZ(sve_sxtb_s, uint32_t, H1_4, DO_SXTB) DO_ZPZ(sve_sxth_s, uint32_t, H1_4, DO_SXTH) DO_ZPZ_D(sve_sxtb_d, uint64_t, DO_SXTB) DO_ZPZ_D(sve_sxth_d, uint64_t, DO_SXTH) DO_ZPZ_D(sve_sxtw_d, uint64_t, DO_SXTS) DO_ZPZ(sve_uxtb_h, uint16_t, H1_2, DO_UXTB) DO_ZPZ(sve_uxtb_s, uint32_t, H1_4, DO_UXTB) DO_ZPZ(sve_uxth_s, uint32_t, H1_4, DO_UXTH) DO_ZPZ_D(sve_uxtb_d, uint64_t, DO_UXTB) DO_ZPZ_D(sve_uxth_d, uint64_t, DO_UXTH) DO_ZPZ_D(sve_uxtw_d, uint64_t, DO_UXTS) #define DO_ABS(N) (N < 0 ? -N : N) DO_ZPZ(sve_abs_b, int8_t, H1, DO_ABS) DO_ZPZ(sve_abs_h, int16_t, H1_2, DO_ABS) DO_ZPZ(sve_abs_s, int32_t, H1_4, DO_ABS) DO_ZPZ_D(sve_abs_d, int64_t, DO_ABS) #define DO_NEG(N) (-N) DO_ZPZ(sve_neg_b, uint8_t, H1, DO_NEG) DO_ZPZ(sve_neg_h, uint16_t, H1_2, DO_NEG) DO_ZPZ(sve_neg_s, uint32_t, H1_4, DO_NEG) DO_ZPZ_D(sve_neg_d, uint64_t, DO_NEG) DO_ZPZ(sve_revb_h, uint16_t, H1_2, bswap16) DO_ZPZ(sve_revb_s, uint32_t, H1_4, bswap32) DO_ZPZ_D(sve_revb_d, uint64_t, bswap64) DO_ZPZ(sve_revh_s, uint32_t, H1_4, hswap32) DO_ZPZ_D(sve_revh_d, uint64_t, hswap64) DO_ZPZ_D(sve_revw_d, uint64_t, wswap64) DO_ZPZ(sve_rbit_b, uint8_t, H1, revbit8) DO_ZPZ(sve_rbit_h, uint16_t, H1_2, revbit16) DO_ZPZ(sve_rbit_s, uint32_t, H1_4, revbit32) DO_ZPZ_D(sve_rbit_d, uint64_t, revbit64) #define DO_SQABS(X) \ ({ __typeof(X) x_ = (X), min_ = 1ull << (sizeof(X) * 8 - 1); \ x_ >= 0 ? x_ : x_ == min_ ? -min_ - 1 : -x_; }) DO_ZPZ(sve2_sqabs_b, int8_t, H1, DO_SQABS) DO_ZPZ(sve2_sqabs_h, int16_t, H1_2, DO_SQABS) DO_ZPZ(sve2_sqabs_s, int32_t, H1_4, DO_SQABS) DO_ZPZ_D(sve2_sqabs_d, int64_t, DO_SQABS) #define DO_SQNEG(X) \ ({ __typeof(X) x_ = (X), min_ = 1ull << (sizeof(X) * 8 - 1); \ x_ == min_ ? -min_ - 1 : -x_; }) DO_ZPZ(sve2_sqneg_b, uint8_t, H1, DO_SQNEG) DO_ZPZ(sve2_sqneg_h, uint16_t, H1_2, DO_SQNEG) DO_ZPZ(sve2_sqneg_s, uint32_t, H1_4, DO_SQNEG) DO_ZPZ_D(sve2_sqneg_d, uint64_t, DO_SQNEG) DO_ZPZ(sve2_urecpe_s, uint32_t, H1_4, helper_recpe_u32) DO_ZPZ(sve2_ursqrte_s, uint32_t, H1_4, helper_rsqrte_u32) /* Three-operand expander, unpredicated, in which the third operand is "wide". */ #define DO_ZZW(NAME, TYPE, TYPEW, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ TYPEW mm = *(TYPEW *)(vm + i); \ do { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, mm); \ i += sizeof(TYPE); \ } while (i & 7); \ } \ } DO_ZZW(sve_asr_zzw_b, int8_t, uint64_t, H1, DO_ASR) DO_ZZW(sve_lsr_zzw_b, uint8_t, uint64_t, H1, DO_LSR) DO_ZZW(sve_lsl_zzw_b, uint8_t, uint64_t, H1, DO_LSL) DO_ZZW(sve_asr_zzw_h, int16_t, uint64_t, H1_2, DO_ASR) DO_ZZW(sve_lsr_zzw_h, uint16_t, uint64_t, H1_2, DO_LSR) DO_ZZW(sve_lsl_zzw_h, uint16_t, uint64_t, H1_2, DO_LSL) DO_ZZW(sve_asr_zzw_s, int32_t, uint64_t, H1_4, DO_ASR) DO_ZZW(sve_lsr_zzw_s, uint32_t, uint64_t, H1_4, DO_LSR) DO_ZZW(sve_lsl_zzw_s, uint32_t, uint64_t, H1_4, DO_LSL) #undef DO_ZZW #undef DO_CLS_B #undef DO_CLS_H #undef DO_CLZ_B #undef DO_CLZ_H #undef DO_CNOT #undef DO_FABS #undef DO_FNEG #undef DO_ABS #undef DO_NEG #undef DO_ZPZ #undef DO_ZPZ_D /* * Three-operand expander, unpredicated, in which the two inputs are * selected from the top or bottom half of the wide column. */ #define DO_ZZZ_TB(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ int sel1 = extract32(desc, SIMD_DATA_SHIFT, 1) * sizeof(TYPEN); \ int sel2 = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(TYPEN); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEN *)(vn + HN(i + sel1)); \ TYPEW mm = *(TYPEN *)(vm + HN(i + sel2)); \ *(TYPEW *)(vd + HW(i)) = OP(nn, mm); \ } \ } DO_ZZZ_TB(sve2_saddl_h, int16_t, int8_t, H1_2, H1, DO_ADD) DO_ZZZ_TB(sve2_saddl_s, int32_t, int16_t, H1_4, H1_2, DO_ADD) DO_ZZZ_TB(sve2_saddl_d, int64_t, int32_t, H1_8, H1_4, DO_ADD) DO_ZZZ_TB(sve2_ssubl_h, int16_t, int8_t, H1_2, H1, DO_SUB) DO_ZZZ_TB(sve2_ssubl_s, int32_t, int16_t, H1_4, H1_2, DO_SUB) DO_ZZZ_TB(sve2_ssubl_d, int64_t, int32_t, H1_8, H1_4, DO_SUB) DO_ZZZ_TB(sve2_sabdl_h, int16_t, int8_t, H1_2, H1, DO_ABD) DO_ZZZ_TB(sve2_sabdl_s, int32_t, int16_t, H1_4, H1_2, DO_ABD) DO_ZZZ_TB(sve2_sabdl_d, int64_t, int32_t, H1_8, H1_4, DO_ABD) DO_ZZZ_TB(sve2_uaddl_h, uint16_t, uint8_t, H1_2, H1, DO_ADD) DO_ZZZ_TB(sve2_uaddl_s, uint32_t, uint16_t, H1_4, H1_2, DO_ADD) DO_ZZZ_TB(sve2_uaddl_d, uint64_t, uint32_t, H1_8, H1_4, DO_ADD) DO_ZZZ_TB(sve2_usubl_h, uint16_t, uint8_t, H1_2, H1, DO_SUB) DO_ZZZ_TB(sve2_usubl_s, uint32_t, uint16_t, H1_4, H1_2, DO_SUB) DO_ZZZ_TB(sve2_usubl_d, uint64_t, uint32_t, H1_8, H1_4, DO_SUB) DO_ZZZ_TB(sve2_uabdl_h, uint16_t, uint8_t, H1_2, H1, DO_ABD) DO_ZZZ_TB(sve2_uabdl_s, uint32_t, uint16_t, H1_4, H1_2, DO_ABD) DO_ZZZ_TB(sve2_uabdl_d, uint64_t, uint32_t, H1_8, H1_4, DO_ABD) DO_ZZZ_TB(sve2_smull_zzz_h, int16_t, int8_t, H1_2, H1, DO_MUL) DO_ZZZ_TB(sve2_smull_zzz_s, int32_t, int16_t, H1_4, H1_2, DO_MUL) DO_ZZZ_TB(sve2_smull_zzz_d, int64_t, int32_t, H1_8, H1_4, DO_MUL) DO_ZZZ_TB(sve2_umull_zzz_h, uint16_t, uint8_t, H1_2, H1, DO_MUL) DO_ZZZ_TB(sve2_umull_zzz_s, uint32_t, uint16_t, H1_4, H1_2, DO_MUL) DO_ZZZ_TB(sve2_umull_zzz_d, uint64_t, uint32_t, H1_8, H1_4, DO_MUL) /* Note that the multiply cannot overflow, but the doubling can. */ static inline int16_t do_sqdmull_h(int16_t n, int16_t m) { int16_t val = n * m; return DO_SQADD_H(val, val); } static inline int32_t do_sqdmull_s(int32_t n, int32_t m) { int32_t val = n * m; return DO_SQADD_S(val, val); } static inline int64_t do_sqdmull_d(int64_t n, int64_t m) { int64_t val = n * m; return do_sqadd_d(val, val); } DO_ZZZ_TB(sve2_sqdmull_zzz_h, int16_t, int8_t, H1_2, H1, do_sqdmull_h) DO_ZZZ_TB(sve2_sqdmull_zzz_s, int32_t, int16_t, H1_4, H1_2, do_sqdmull_s) DO_ZZZ_TB(sve2_sqdmull_zzz_d, int64_t, int32_t, H1_8, H1_4, do_sqdmull_d) #undef DO_ZZZ_TB #define DO_ZZZ_WTB(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ int sel2 = extract32(desc, SIMD_DATA_SHIFT, 1) * sizeof(TYPEN); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEW *)(vn + HW(i)); \ TYPEW mm = *(TYPEN *)(vm + HN(i + sel2)); \ *(TYPEW *)(vd + HW(i)) = OP(nn, mm); \ } \ } DO_ZZZ_WTB(sve2_saddw_h, int16_t, int8_t, H1_2, H1, DO_ADD) DO_ZZZ_WTB(sve2_saddw_s, int32_t, int16_t, H1_4, H1_2, DO_ADD) DO_ZZZ_WTB(sve2_saddw_d, int64_t, int32_t, H1_8, H1_4, DO_ADD) DO_ZZZ_WTB(sve2_ssubw_h, int16_t, int8_t, H1_2, H1, DO_SUB) DO_ZZZ_WTB(sve2_ssubw_s, int32_t, int16_t, H1_4, H1_2, DO_SUB) DO_ZZZ_WTB(sve2_ssubw_d, int64_t, int32_t, H1_8, H1_4, DO_SUB) DO_ZZZ_WTB(sve2_uaddw_h, uint16_t, uint8_t, H1_2, H1, DO_ADD) DO_ZZZ_WTB(sve2_uaddw_s, uint32_t, uint16_t, H1_4, H1_2, DO_ADD) DO_ZZZ_WTB(sve2_uaddw_d, uint64_t, uint32_t, H1_8, H1_4, DO_ADD) DO_ZZZ_WTB(sve2_usubw_h, uint16_t, uint8_t, H1_2, H1, DO_SUB) DO_ZZZ_WTB(sve2_usubw_s, uint32_t, uint16_t, H1_4, H1_2, DO_SUB) DO_ZZZ_WTB(sve2_usubw_d, uint64_t, uint32_t, H1_8, H1_4, DO_SUB) #undef DO_ZZZ_WTB #define DO_ZZZ_NTB(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ intptr_t sel1 = extract32(desc, SIMD_DATA_SHIFT, 1) * sizeof(TYPE); \ intptr_t sel2 = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(TYPE); \ for (i = 0; i < opr_sz; i += 2 * sizeof(TYPE)) { \ TYPE nn = *(TYPE *)(vn + H(i + sel1)); \ TYPE mm = *(TYPE *)(vm + H(i + sel2)); \ *(TYPE *)(vd + H(i + sel1)) = OP(nn, mm); \ } \ } DO_ZZZ_NTB(sve2_eoril_b, uint8_t, H1, DO_EOR) DO_ZZZ_NTB(sve2_eoril_h, uint16_t, H1_2, DO_EOR) DO_ZZZ_NTB(sve2_eoril_s, uint32_t, H1_4, DO_EOR) DO_ZZZ_NTB(sve2_eoril_d, uint64_t, H1_8, DO_EOR) #undef DO_ZZZ_NTB #define DO_ZZZW_ACC(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ intptr_t sel1 = simd_data(desc) * sizeof(TYPEN); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEN *)(vn + HN(i + sel1)); \ TYPEW mm = *(TYPEN *)(vm + HN(i + sel1)); \ TYPEW aa = *(TYPEW *)(va + HW(i)); \ *(TYPEW *)(vd + HW(i)) = OP(nn, mm) + aa; \ } \ } DO_ZZZW_ACC(sve2_sabal_h, int16_t, int8_t, H1_2, H1, DO_ABD) DO_ZZZW_ACC(sve2_sabal_s, int32_t, int16_t, H1_4, H1_2, DO_ABD) DO_ZZZW_ACC(sve2_sabal_d, int64_t, int32_t, H1_8, H1_4, DO_ABD) DO_ZZZW_ACC(sve2_uabal_h, uint16_t, uint8_t, H1_2, H1, DO_ABD) DO_ZZZW_ACC(sve2_uabal_s, uint32_t, uint16_t, H1_4, H1_2, DO_ABD) DO_ZZZW_ACC(sve2_uabal_d, uint64_t, uint32_t, H1_8, H1_4, DO_ABD) DO_ZZZW_ACC(sve2_smlal_zzzw_h, int16_t, int8_t, H1_2, H1, DO_MUL) DO_ZZZW_ACC(sve2_smlal_zzzw_s, int32_t, int16_t, H1_4, H1_2, DO_MUL) DO_ZZZW_ACC(sve2_smlal_zzzw_d, int64_t, int32_t, H1_8, H1_4, DO_MUL) DO_ZZZW_ACC(sve2_umlal_zzzw_h, uint16_t, uint8_t, H1_2, H1, DO_MUL) DO_ZZZW_ACC(sve2_umlal_zzzw_s, uint32_t, uint16_t, H1_4, H1_2, DO_MUL) DO_ZZZW_ACC(sve2_umlal_zzzw_d, uint64_t, uint32_t, H1_8, H1_4, DO_MUL) #define DO_NMUL(N, M) -(N * M) DO_ZZZW_ACC(sve2_smlsl_zzzw_h, int16_t, int8_t, H1_2, H1, DO_NMUL) DO_ZZZW_ACC(sve2_smlsl_zzzw_s, int32_t, int16_t, H1_4, H1_2, DO_NMUL) DO_ZZZW_ACC(sve2_smlsl_zzzw_d, int64_t, int32_t, H1_8, H1_4, DO_NMUL) DO_ZZZW_ACC(sve2_umlsl_zzzw_h, uint16_t, uint8_t, H1_2, H1, DO_NMUL) DO_ZZZW_ACC(sve2_umlsl_zzzw_s, uint32_t, uint16_t, H1_4, H1_2, DO_NMUL) DO_ZZZW_ACC(sve2_umlsl_zzzw_d, uint64_t, uint32_t, H1_8, H1_4, DO_NMUL) #undef DO_ZZZW_ACC #define DO_XTNB(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; i += sizeof(TYPE)) { \ TYPE nn = *(TYPE *)(vn + i); \ nn = OP(nn) & MAKE_64BIT_MASK(0, sizeof(TYPE) * 4); \ *(TYPE *)(vd + i) = nn; \ } \ } #define DO_XTNT(NAME, TYPE, TYPEN, H, OP) \ void HELPER(NAME)(void *vd, void *vn, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc), odd = H(sizeof(TYPEN)); \ for (i = 0; i < opr_sz; i += sizeof(TYPE)) { \ TYPE nn = *(TYPE *)(vn + i); \ *(TYPEN *)(vd + i + odd) = OP(nn); \ } \ } #define DO_SQXTN_H(n) do_sat_bhs(n, INT8_MIN, INT8_MAX) #define DO_SQXTN_S(n) do_sat_bhs(n, INT16_MIN, INT16_MAX) #define DO_SQXTN_D(n) do_sat_bhs(n, INT32_MIN, INT32_MAX) DO_XTNB(sve2_sqxtnb_h, int16_t, DO_SQXTN_H) DO_XTNB(sve2_sqxtnb_s, int32_t, DO_SQXTN_S) DO_XTNB(sve2_sqxtnb_d, int64_t, DO_SQXTN_D) DO_XTNT(sve2_sqxtnt_h, int16_t, int8_t, H1, DO_SQXTN_H) DO_XTNT(sve2_sqxtnt_s, int32_t, int16_t, H1_2, DO_SQXTN_S) DO_XTNT(sve2_sqxtnt_d, int64_t, int32_t, H1_4, DO_SQXTN_D) #define DO_UQXTN_H(n) do_sat_bhs(n, 0, UINT8_MAX) #define DO_UQXTN_S(n) do_sat_bhs(n, 0, UINT16_MAX) #define DO_UQXTN_D(n) do_sat_bhs(n, 0, UINT32_MAX) DO_XTNB(sve2_uqxtnb_h, uint16_t, DO_UQXTN_H) DO_XTNB(sve2_uqxtnb_s, uint32_t, DO_UQXTN_S) DO_XTNB(sve2_uqxtnb_d, uint64_t, DO_UQXTN_D) DO_XTNT(sve2_uqxtnt_h, uint16_t, uint8_t, H1, DO_UQXTN_H) DO_XTNT(sve2_uqxtnt_s, uint32_t, uint16_t, H1_2, DO_UQXTN_S) DO_XTNT(sve2_uqxtnt_d, uint64_t, uint32_t, H1_4, DO_UQXTN_D) DO_XTNB(sve2_sqxtunb_h, int16_t, DO_UQXTN_H) DO_XTNB(sve2_sqxtunb_s, int32_t, DO_UQXTN_S) DO_XTNB(sve2_sqxtunb_d, int64_t, DO_UQXTN_D) DO_XTNT(sve2_sqxtunt_h, int16_t, int8_t, H1, DO_UQXTN_H) DO_XTNT(sve2_sqxtunt_s, int32_t, int16_t, H1_2, DO_UQXTN_S) DO_XTNT(sve2_sqxtunt_d, int64_t, int32_t, H1_4, DO_UQXTN_D) #undef DO_XTNB #undef DO_XTNT void HELPER(sve2_adcl_s)(void *vd, void *vn, void *vm, void *va, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc); int sel = H4(extract32(desc, SIMD_DATA_SHIFT, 1)); uint32_t inv = -extract32(desc, SIMD_DATA_SHIFT + 1, 1); uint32_t *a = va, *n = vn; uint64_t *d = vd, *m = vm; for (i = 0; i < opr_sz / 8; ++i) { uint32_t e1 = a[2 * i + H4(0)]; uint32_t e2 = n[2 * i + sel] ^ inv; uint64_t c = extract64(m[i], 32, 1); /* Compute and store the entire 33-bit result at once. */ d[i] = c + e1 + e2; } } void HELPER(sve2_adcl_d)(void *vd, void *vn, void *vm, void *va, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc); int sel = extract32(desc, SIMD_DATA_SHIFT, 1); uint64_t inv = -(uint64_t)extract32(desc, SIMD_DATA_SHIFT + 1, 1); uint64_t *d = vd, *a = va, *n = vn, *m = vm; for (i = 0; i < opr_sz / 8; i += 2) { Int128 e1 = int128_make64(a[i]); Int128 e2 = int128_make64(n[i + sel] ^ inv); Int128 c = int128_make64(m[i + 1] & 1); Int128 r = int128_add(int128_add(e1, e2), c); d[i + 0] = int128_getlo(r); d[i + 1] = int128_gethi(r); } } #define DO_SQDMLAL(NAME, TYPEW, TYPEN, HW, HN, DMUL_OP, SUM_OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ int sel1 = extract32(desc, SIMD_DATA_SHIFT, 1) * sizeof(TYPEN); \ int sel2 = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(TYPEN); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEN *)(vn + HN(i + sel1)); \ TYPEW mm = *(TYPEN *)(vm + HN(i + sel2)); \ TYPEW aa = *(TYPEW *)(va + HW(i)); \ *(TYPEW *)(vd + HW(i)) = SUM_OP(aa, DMUL_OP(nn, mm)); \ } \ } DO_SQDMLAL(sve2_sqdmlal_zzzw_h, int16_t, int8_t, H1_2, H1, do_sqdmull_h, DO_SQADD_H) DO_SQDMLAL(sve2_sqdmlal_zzzw_s, int32_t, int16_t, H1_4, H1_2, do_sqdmull_s, DO_SQADD_S) DO_SQDMLAL(sve2_sqdmlal_zzzw_d, int64_t, int32_t, H1_8, H1_4, do_sqdmull_d, do_sqadd_d) DO_SQDMLAL(sve2_sqdmlsl_zzzw_h, int16_t, int8_t, H1_2, H1, do_sqdmull_h, DO_SQSUB_H) DO_SQDMLAL(sve2_sqdmlsl_zzzw_s, int32_t, int16_t, H1_4, H1_2, do_sqdmull_s, DO_SQSUB_S) DO_SQDMLAL(sve2_sqdmlsl_zzzw_d, int64_t, int32_t, H1_8, H1_4, do_sqdmull_d, do_sqsub_d) #undef DO_SQDMLAL #define DO_CMLA_FUNC(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / sizeof(TYPE); \ int rot = simd_data(desc); \ int sel_a = rot & 1, sel_b = sel_a ^ 1; \ bool sub_r = rot == 1 || rot == 2; \ bool sub_i = rot >= 2; \ TYPE *d = vd, *n = vn, *m = vm, *a = va; \ for (i = 0; i < opr_sz; i += 2) { \ TYPE elt1_a = n[H(i + sel_a)]; \ TYPE elt2_a = m[H(i + sel_a)]; \ TYPE elt2_b = m[H(i + sel_b)]; \ d[H(i)] = OP(elt1_a, elt2_a, a[H(i)], sub_r); \ d[H(i + 1)] = OP(elt1_a, elt2_b, a[H(i + 1)], sub_i); \ } \ } #define DO_CMLA(N, M, A, S) (A + (N * M) * (S ? -1 : 1)) DO_CMLA_FUNC(sve2_cmla_zzzz_b, uint8_t, H1, DO_CMLA) DO_CMLA_FUNC(sve2_cmla_zzzz_h, uint16_t, H2, DO_CMLA) DO_CMLA_FUNC(sve2_cmla_zzzz_s, uint32_t, H4, DO_CMLA) DO_CMLA_FUNC(sve2_cmla_zzzz_d, uint64_t, H8, DO_CMLA) #define DO_SQRDMLAH_B(N, M, A, S) \ do_sqrdmlah_b(N, M, A, S, true) #define DO_SQRDMLAH_H(N, M, A, S) \ ({ uint32_t discard; do_sqrdmlah_h(N, M, A, S, true, &discard); }) #define DO_SQRDMLAH_S(N, M, A, S) \ ({ uint32_t discard; do_sqrdmlah_s(N, M, A, S, true, &discard); }) #define DO_SQRDMLAH_D(N, M, A, S) \ do_sqrdmlah_d(N, M, A, S, true) DO_CMLA_FUNC(sve2_sqrdcmlah_zzzz_b, int8_t, H1, DO_SQRDMLAH_B) DO_CMLA_FUNC(sve2_sqrdcmlah_zzzz_h, int16_t, H2, DO_SQRDMLAH_H) DO_CMLA_FUNC(sve2_sqrdcmlah_zzzz_s, int32_t, H4, DO_SQRDMLAH_S) DO_CMLA_FUNC(sve2_sqrdcmlah_zzzz_d, int64_t, H8, DO_SQRDMLAH_D) #define DO_CMLA_IDX_FUNC(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ intptr_t i, j, oprsz = simd_oprsz(desc); \ int rot = extract32(desc, SIMD_DATA_SHIFT, 2); \ int idx = extract32(desc, SIMD_DATA_SHIFT + 2, 2) * 2; \ int sel_a = rot & 1, sel_b = sel_a ^ 1; \ bool sub_r = rot == 1 || rot == 2; \ bool sub_i = rot >= 2; \ TYPE *d = vd, *n = vn, *m = vm, *a = va; \ for (i = 0; i < oprsz / sizeof(TYPE); i += 16 / sizeof(TYPE)) { \ TYPE elt2_a = m[H(i + idx + sel_a)]; \ TYPE elt2_b = m[H(i + idx + sel_b)]; \ for (j = 0; j < 16 / sizeof(TYPE); j += 2) { \ TYPE elt1_a = n[H(i + j + sel_a)]; \ d[H2(i + j)] = OP(elt1_a, elt2_a, a[H(i + j)], sub_r); \ d[H2(i + j + 1)] = OP(elt1_a, elt2_b, a[H(i + j + 1)], sub_i); \ } \ } \ } DO_CMLA_IDX_FUNC(sve2_cmla_idx_h, int16_t, H2, DO_CMLA) DO_CMLA_IDX_FUNC(sve2_cmla_idx_s, int32_t, H4, DO_CMLA) DO_CMLA_IDX_FUNC(sve2_sqrdcmlah_idx_h, int16_t, H2, DO_SQRDMLAH_H) DO_CMLA_IDX_FUNC(sve2_sqrdcmlah_idx_s, int32_t, H4, DO_SQRDMLAH_S) #undef DO_CMLA #undef DO_CMLA_FUNC #undef DO_CMLA_IDX_FUNC #undef DO_SQRDMLAH_B #undef DO_SQRDMLAH_H #undef DO_SQRDMLAH_S #undef DO_SQRDMLAH_D /* Note N and M are 4 elements bundled into one unit. */ static int32_t do_cdot_s(uint32_t n, uint32_t m, int32_t a, int sel_a, int sel_b, int sub_i) { for (int i = 0; i <= 1; i++) { int32_t elt1_r = (int8_t)(n >> (16 * i)); int32_t elt1_i = (int8_t)(n >> (16 * i + 8)); int32_t elt2_a = (int8_t)(m >> (16 * i + 8 * sel_a)); int32_t elt2_b = (int8_t)(m >> (16 * i + 8 * sel_b)); a += elt1_r * elt2_a + elt1_i * elt2_b * sub_i; } return a; } static int64_t do_cdot_d(uint64_t n, uint64_t m, int64_t a, int sel_a, int sel_b, int sub_i) { for (int i = 0; i <= 1; i++) { int64_t elt1_r = (int16_t)(n >> (32 * i + 0)); int64_t elt1_i = (int16_t)(n >> (32 * i + 16)); int64_t elt2_a = (int16_t)(m >> (32 * i + 16 * sel_a)); int64_t elt2_b = (int16_t)(m >> (32 * i + 16 * sel_b)); a += elt1_r * elt2_a + elt1_i * elt2_b * sub_i; } return a; } void HELPER(sve2_cdot_zzzz_s)(void *vd, void *vn, void *vm, void *va, uint32_t desc) { int opr_sz = simd_oprsz(desc); int rot = simd_data(desc); int sel_a = rot & 1; int sel_b = sel_a ^ 1; int sub_i = (rot == 0 || rot == 3 ? -1 : 1); uint32_t *d = vd, *n = vn, *m = vm, *a = va; for (int e = 0; e < opr_sz / 4; e++) { d[e] = do_cdot_s(n[e], m[e], a[e], sel_a, sel_b, sub_i); } } void HELPER(sve2_cdot_zzzz_d)(void *vd, void *vn, void *vm, void *va, uint32_t desc) { int opr_sz = simd_oprsz(desc); int rot = simd_data(desc); int sel_a = rot & 1; int sel_b = sel_a ^ 1; int sub_i = (rot == 0 || rot == 3 ? -1 : 1); uint64_t *d = vd, *n = vn, *m = vm, *a = va; for (int e = 0; e < opr_sz / 8; e++) { d[e] = do_cdot_d(n[e], m[e], a[e], sel_a, sel_b, sub_i); } } void HELPER(sve2_cdot_idx_s)(void *vd, void *vn, void *vm, void *va, uint32_t desc) { int opr_sz = simd_oprsz(desc); int rot = extract32(desc, SIMD_DATA_SHIFT, 2); int idx = H4(extract32(desc, SIMD_DATA_SHIFT + 2, 2)); int sel_a = rot & 1; int sel_b = sel_a ^ 1; int sub_i = (rot == 0 || rot == 3 ? -1 : 1); uint32_t *d = vd, *n = vn, *m = vm, *a = va; for (int seg = 0; seg < opr_sz / 4; seg += 4) { uint32_t seg_m = m[seg + idx]; for (int e = 0; e < 4; e++) { d[seg + e] = do_cdot_s(n[seg + e], seg_m, a[seg + e], sel_a, sel_b, sub_i); } } } void HELPER(sve2_cdot_idx_d)(void *vd, void *vn, void *vm, void *va, uint32_t desc) { int seg, opr_sz = simd_oprsz(desc); int rot = extract32(desc, SIMD_DATA_SHIFT, 2); int idx = extract32(desc, SIMD_DATA_SHIFT + 2, 2); int sel_a = rot & 1; int sel_b = sel_a ^ 1; int sub_i = (rot == 0 || rot == 3 ? -1 : 1); uint64_t *d = vd, *n = vn, *m = vm, *a = va; for (seg = 0; seg < opr_sz / 8; seg += 2) { uint64_t seg_m = m[seg + idx]; for (int e = 0; e < 2; e++) { d[seg + e] = do_cdot_d(n[seg + e], seg_m, a[seg + e], sel_a, sel_b, sub_i); } } } #define DO_ZZXZ(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ intptr_t oprsz = simd_oprsz(desc), segment = 16 / sizeof(TYPE); \ intptr_t i, j, idx = simd_data(desc); \ TYPE *d = vd, *a = va, *n = vn, *m = (TYPE *)vm + H(idx); \ for (i = 0; i < oprsz / sizeof(TYPE); i += segment) { \ TYPE mm = m[i]; \ for (j = 0; j < segment; j++) { \ d[i + j] = OP(n[i + j], mm, a[i + j]); \ } \ } \ } #define DO_SQRDMLAH_H(N, M, A) \ ({ uint32_t discard; do_sqrdmlah_h(N, M, A, false, true, &discard); }) #define DO_SQRDMLAH_S(N, M, A) \ ({ uint32_t discard; do_sqrdmlah_s(N, M, A, false, true, &discard); }) #define DO_SQRDMLAH_D(N, M, A) do_sqrdmlah_d(N, M, A, false, true) DO_ZZXZ(sve2_sqrdmlah_idx_h, int16_t, H2, DO_SQRDMLAH_H) DO_ZZXZ(sve2_sqrdmlah_idx_s, int32_t, H4, DO_SQRDMLAH_S) DO_ZZXZ(sve2_sqrdmlah_idx_d, int64_t, H8, DO_SQRDMLAH_D) #define DO_SQRDMLSH_H(N, M, A) \ ({ uint32_t discard; do_sqrdmlah_h(N, M, A, true, true, &discard); }) #define DO_SQRDMLSH_S(N, M, A) \ ({ uint32_t discard; do_sqrdmlah_s(N, M, A, true, true, &discard); }) #define DO_SQRDMLSH_D(N, M, A) do_sqrdmlah_d(N, M, A, true, true) DO_ZZXZ(sve2_sqrdmlsh_idx_h, int16_t, H2, DO_SQRDMLSH_H) DO_ZZXZ(sve2_sqrdmlsh_idx_s, int32_t, H4, DO_SQRDMLSH_S) DO_ZZXZ(sve2_sqrdmlsh_idx_d, int64_t, H8, DO_SQRDMLSH_D) #undef DO_ZZXZ #define DO_ZZXW(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *va, uint32_t desc) \ { \ intptr_t i, j, oprsz = simd_oprsz(desc); \ intptr_t sel = extract32(desc, SIMD_DATA_SHIFT, 1) * sizeof(TYPEN); \ intptr_t idx = extract32(desc, SIMD_DATA_SHIFT + 1, 3) * sizeof(TYPEN); \ for (i = 0; i < oprsz; i += 16) { \ TYPEW mm = *(TYPEN *)(vm + HN(i + idx)); \ for (j = 0; j < 16; j += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEN *)(vn + HN(i + j + sel)); \ TYPEW aa = *(TYPEW *)(va + HW(i + j)); \ *(TYPEW *)(vd + HW(i + j)) = OP(nn, mm, aa); \ } \ } \ } #define DO_MLA(N, M, A) (A + N * M) DO_ZZXW(sve2_smlal_idx_s, int32_t, int16_t, H1_4, H1_2, DO_MLA) DO_ZZXW(sve2_smlal_idx_d, int64_t, int32_t, H1_8, H1_4, DO_MLA) DO_ZZXW(sve2_umlal_idx_s, uint32_t, uint16_t, H1_4, H1_2, DO_MLA) DO_ZZXW(sve2_umlal_idx_d, uint64_t, uint32_t, H1_8, H1_4, DO_MLA) #define DO_MLS(N, M, A) (A - N * M) DO_ZZXW(sve2_smlsl_idx_s, int32_t, int16_t, H1_4, H1_2, DO_MLS) DO_ZZXW(sve2_smlsl_idx_d, int64_t, int32_t, H1_8, H1_4, DO_MLS) DO_ZZXW(sve2_umlsl_idx_s, uint32_t, uint16_t, H1_4, H1_2, DO_MLS) DO_ZZXW(sve2_umlsl_idx_d, uint64_t, uint32_t, H1_8, H1_4, DO_MLS) #define DO_SQDMLAL_S(N, M, A) DO_SQADD_S(A, do_sqdmull_s(N, M)) #define DO_SQDMLAL_D(N, M, A) do_sqadd_d(A, do_sqdmull_d(N, M)) DO_ZZXW(sve2_sqdmlal_idx_s, int32_t, int16_t, H1_4, H1_2, DO_SQDMLAL_S) DO_ZZXW(sve2_sqdmlal_idx_d, int64_t, int32_t, H1_8, H1_4, DO_SQDMLAL_D) #define DO_SQDMLSL_S(N, M, A) DO_SQSUB_S(A, do_sqdmull_s(N, M)) #define DO_SQDMLSL_D(N, M, A) do_sqsub_d(A, do_sqdmull_d(N, M)) DO_ZZXW(sve2_sqdmlsl_idx_s, int32_t, int16_t, H1_4, H1_2, DO_SQDMLSL_S) DO_ZZXW(sve2_sqdmlsl_idx_d, int64_t, int32_t, H1_8, H1_4, DO_SQDMLSL_D) #undef DO_MLA #undef DO_MLS #undef DO_ZZXW #define DO_ZZX(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, j, oprsz = simd_oprsz(desc); \ intptr_t sel = extract32(desc, SIMD_DATA_SHIFT, 1) * sizeof(TYPEN); \ intptr_t idx = extract32(desc, SIMD_DATA_SHIFT + 1, 3) * sizeof(TYPEN); \ for (i = 0; i < oprsz; i += 16) { \ TYPEW mm = *(TYPEN *)(vm + HN(i + idx)); \ for (j = 0; j < 16; j += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEN *)(vn + HN(i + j + sel)); \ *(TYPEW *)(vd + HW(i + j)) = OP(nn, mm); \ } \ } \ } DO_ZZX(sve2_sqdmull_idx_s, int32_t, int16_t, H1_4, H1_2, do_sqdmull_s) DO_ZZX(sve2_sqdmull_idx_d, int64_t, int32_t, H1_8, H1_4, do_sqdmull_d) DO_ZZX(sve2_smull_idx_s, int32_t, int16_t, H1_4, H1_2, DO_MUL) DO_ZZX(sve2_smull_idx_d, int64_t, int32_t, H1_8, H1_4, DO_MUL) DO_ZZX(sve2_umull_idx_s, uint32_t, uint16_t, H1_4, H1_2, DO_MUL) DO_ZZX(sve2_umull_idx_d, uint64_t, uint32_t, H1_8, H1_4, DO_MUL) #undef DO_ZZX #define DO_BITPERM(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; i += sizeof(TYPE)) { \ TYPE nn = *(TYPE *)(vn + i); \ TYPE mm = *(TYPE *)(vm + i); \ *(TYPE *)(vd + i) = OP(nn, mm, sizeof(TYPE) * 8); \ } \ } static uint64_t bitextract(uint64_t data, uint64_t mask, int n) { uint64_t res = 0; int db, rb = 0; for (db = 0; db < n; ++db) { if ((mask >> db) & 1) { res |= ((data >> db) & 1) << rb; ++rb; } } return res; } DO_BITPERM(sve2_bext_b, uint8_t, bitextract) DO_BITPERM(sve2_bext_h, uint16_t, bitextract) DO_BITPERM(sve2_bext_s, uint32_t, bitextract) DO_BITPERM(sve2_bext_d, uint64_t, bitextract) static uint64_t bitdeposit(uint64_t data, uint64_t mask, int n) { uint64_t res = 0; int rb, db = 0; for (rb = 0; rb < n; ++rb) { if ((mask >> rb) & 1) { res |= ((data >> db) & 1) << rb; ++db; } } return res; } DO_BITPERM(sve2_bdep_b, uint8_t, bitdeposit) DO_BITPERM(sve2_bdep_h, uint16_t, bitdeposit) DO_BITPERM(sve2_bdep_s, uint32_t, bitdeposit) DO_BITPERM(sve2_bdep_d, uint64_t, bitdeposit) static uint64_t bitgroup(uint64_t data, uint64_t mask, int n) { uint64_t resm = 0, resu = 0; int db, rbm = 0, rbu = 0; for (db = 0; db < n; ++db) { uint64_t val = (data >> db) & 1; if ((mask >> db) & 1) { resm |= val << rbm++; } else { resu |= val << rbu++; } } return resm | (resu << rbm); } DO_BITPERM(sve2_bgrp_b, uint8_t, bitgroup) DO_BITPERM(sve2_bgrp_h, uint16_t, bitgroup) DO_BITPERM(sve2_bgrp_s, uint32_t, bitgroup) DO_BITPERM(sve2_bgrp_d, uint64_t, bitgroup) #undef DO_BITPERM #define DO_CADD(NAME, TYPE, H, ADD_OP, SUB_OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ int sub_r = simd_data(desc); \ if (sub_r) { \ for (i = 0; i < opr_sz; i += 2 * sizeof(TYPE)) { \ TYPE acc_r = *(TYPE *)(vn + H(i)); \ TYPE acc_i = *(TYPE *)(vn + H(i + sizeof(TYPE))); \ TYPE el2_r = *(TYPE *)(vm + H(i)); \ TYPE el2_i = *(TYPE *)(vm + H(i + sizeof(TYPE))); \ acc_r = ADD_OP(acc_r, el2_i); \ acc_i = SUB_OP(acc_i, el2_r); \ *(TYPE *)(vd + H(i)) = acc_r; \ *(TYPE *)(vd + H(i + sizeof(TYPE))) = acc_i; \ } \ } else { \ for (i = 0; i < opr_sz; i += 2 * sizeof(TYPE)) { \ TYPE acc_r = *(TYPE *)(vn + H(i)); \ TYPE acc_i = *(TYPE *)(vn + H(i + sizeof(TYPE))); \ TYPE el2_r = *(TYPE *)(vm + H(i)); \ TYPE el2_i = *(TYPE *)(vm + H(i + sizeof(TYPE))); \ acc_r = SUB_OP(acc_r, el2_i); \ acc_i = ADD_OP(acc_i, el2_r); \ *(TYPE *)(vd + H(i)) = acc_r; \ *(TYPE *)(vd + H(i + sizeof(TYPE))) = acc_i; \ } \ } \ } DO_CADD(sve2_cadd_b, int8_t, H1, DO_ADD, DO_SUB) DO_CADD(sve2_cadd_h, int16_t, H1_2, DO_ADD, DO_SUB) DO_CADD(sve2_cadd_s, int32_t, H1_4, DO_ADD, DO_SUB) DO_CADD(sve2_cadd_d, int64_t, H1_8, DO_ADD, DO_SUB) DO_CADD(sve2_sqcadd_b, int8_t, H1, DO_SQADD_B, DO_SQSUB_B) DO_CADD(sve2_sqcadd_h, int16_t, H1_2, DO_SQADD_H, DO_SQSUB_H) DO_CADD(sve2_sqcadd_s, int32_t, H1_4, DO_SQADD_S, DO_SQSUB_S) DO_CADD(sve2_sqcadd_d, int64_t, H1_8, do_sqadd_d, do_sqsub_d) #undef DO_CADD #define DO_ZZI_SHLL(NAME, TYPEW, TYPEN, HW, HN) \ void HELPER(NAME)(void *vd, void *vn, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ intptr_t sel = (simd_data(desc) & 1) * sizeof(TYPEN); \ int shift = simd_data(desc) >> 1; \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEN *)(vn + HN(i + sel)); \ *(TYPEW *)(vd + HW(i)) = nn << shift; \ } \ } DO_ZZI_SHLL(sve2_sshll_h, int16_t, int8_t, H1_2, H1) DO_ZZI_SHLL(sve2_sshll_s, int32_t, int16_t, H1_4, H1_2) DO_ZZI_SHLL(sve2_sshll_d, int64_t, int32_t, H1_8, H1_4) DO_ZZI_SHLL(sve2_ushll_h, uint16_t, uint8_t, H1_2, H1) DO_ZZI_SHLL(sve2_ushll_s, uint32_t, uint16_t, H1_4, H1_2) DO_ZZI_SHLL(sve2_ushll_d, uint64_t, uint32_t, H1_8, H1_4) #undef DO_ZZI_SHLL /* Two-operand reduction expander, controlled by a predicate. * The difference between TYPERED and TYPERET has to do with * sign-extension. E.g. for SMAX, TYPERED must be signed, * but TYPERET must be unsigned so that e.g. a 32-bit value * is not sign-extended to the ABI uint64_t return type. */ /* ??? If we were to vectorize this by hand the reduction ordering * would change. For integer operands, this is perfectly fine. */ #define DO_VPZ(NAME, TYPEELT, TYPERED, TYPERET, H, INIT, OP) \ uint64_t HELPER(NAME)(void *vn, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ TYPERED ret = INIT; \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ if (pg & 1) { \ TYPEELT nn = *(TYPEELT *)(vn + H(i)); \ ret = OP(ret, nn); \ } \ i += sizeof(TYPEELT), pg >>= sizeof(TYPEELT); \ } while (i & 15); \ } \ return (TYPERET)ret; \ } #define DO_VPZ_D(NAME, TYPEE, TYPER, INIT, OP) \ uint64_t HELPER(NAME)(void *vn, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / 8; \ TYPEE *n = vn; \ uint8_t *pg = vg; \ TYPER ret = INIT; \ for (i = 0; i < opr_sz; i += 1) { \ if (pg[H1(i)] & 1) { \ TYPEE nn = n[i]; \ ret = OP(ret, nn); \ } \ } \ return ret; \ } DO_VPZ(sve_orv_b, uint8_t, uint8_t, uint8_t, H1, 0, DO_ORR) DO_VPZ(sve_orv_h, uint16_t, uint16_t, uint16_t, H1_2, 0, DO_ORR) DO_VPZ(sve_orv_s, uint32_t, uint32_t, uint32_t, H1_4, 0, DO_ORR) DO_VPZ_D(sve_orv_d, uint64_t, uint64_t, 0, DO_ORR) DO_VPZ(sve_eorv_b, uint8_t, uint8_t, uint8_t, H1, 0, DO_EOR) DO_VPZ(sve_eorv_h, uint16_t, uint16_t, uint16_t, H1_2, 0, DO_EOR) DO_VPZ(sve_eorv_s, uint32_t, uint32_t, uint32_t, H1_4, 0, DO_EOR) DO_VPZ_D(sve_eorv_d, uint64_t, uint64_t, 0, DO_EOR) DO_VPZ(sve_andv_b, uint8_t, uint8_t, uint8_t, H1, -1, DO_AND) DO_VPZ(sve_andv_h, uint16_t, uint16_t, uint16_t, H1_2, -1, DO_AND) DO_VPZ(sve_andv_s, uint32_t, uint32_t, uint32_t, H1_4, -1, DO_AND) DO_VPZ_D(sve_andv_d, uint64_t, uint64_t, -1, DO_AND) DO_VPZ(sve_saddv_b, int8_t, uint64_t, uint64_t, H1, 0, DO_ADD) DO_VPZ(sve_saddv_h, int16_t, uint64_t, uint64_t, H1_2, 0, DO_ADD) DO_VPZ(sve_saddv_s, int32_t, uint64_t, uint64_t, H1_4, 0, DO_ADD) DO_VPZ(sve_uaddv_b, uint8_t, uint64_t, uint64_t, H1, 0, DO_ADD) DO_VPZ(sve_uaddv_h, uint16_t, uint64_t, uint64_t, H1_2, 0, DO_ADD) DO_VPZ(sve_uaddv_s, uint32_t, uint64_t, uint64_t, H1_4, 0, DO_ADD) DO_VPZ_D(sve_uaddv_d, uint64_t, uint64_t, 0, DO_ADD) DO_VPZ(sve_smaxv_b, int8_t, int8_t, uint8_t, H1, INT8_MIN, DO_MAX) DO_VPZ(sve_smaxv_h, int16_t, int16_t, uint16_t, H1_2, INT16_MIN, DO_MAX) DO_VPZ(sve_smaxv_s, int32_t, int32_t, uint32_t, H1_4, INT32_MIN, DO_MAX) DO_VPZ_D(sve_smaxv_d, int64_t, int64_t, INT64_MIN, DO_MAX) DO_VPZ(sve_umaxv_b, uint8_t, uint8_t, uint8_t, H1, 0, DO_MAX) DO_VPZ(sve_umaxv_h, uint16_t, uint16_t, uint16_t, H1_2, 0, DO_MAX) DO_VPZ(sve_umaxv_s, uint32_t, uint32_t, uint32_t, H1_4, 0, DO_MAX) DO_VPZ_D(sve_umaxv_d, uint64_t, uint64_t, 0, DO_MAX) DO_VPZ(sve_sminv_b, int8_t, int8_t, uint8_t, H1, INT8_MAX, DO_MIN) DO_VPZ(sve_sminv_h, int16_t, int16_t, uint16_t, H1_2, INT16_MAX, DO_MIN) DO_VPZ(sve_sminv_s, int32_t, int32_t, uint32_t, H1_4, INT32_MAX, DO_MIN) DO_VPZ_D(sve_sminv_d, int64_t, int64_t, INT64_MAX, DO_MIN) DO_VPZ(sve_uminv_b, uint8_t, uint8_t, uint8_t, H1, -1, DO_MIN) DO_VPZ(sve_uminv_h, uint16_t, uint16_t, uint16_t, H1_2, -1, DO_MIN) DO_VPZ(sve_uminv_s, uint32_t, uint32_t, uint32_t, H1_4, -1, DO_MIN) DO_VPZ_D(sve_uminv_d, uint64_t, uint64_t, -1, DO_MIN) #undef DO_VPZ #undef DO_VPZ_D /* Two vector operand, one scalar operand, unpredicated. */ #define DO_ZZI(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, uint64_t s64, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / sizeof(TYPE); \ TYPE s = s64, *d = vd, *n = vn; \ for (i = 0; i < opr_sz; ++i) { \ d[i] = OP(n[i], s); \ } \ } #define DO_SUBR(X, Y) (Y - X) DO_ZZI(sve_subri_b, uint8_t, DO_SUBR) DO_ZZI(sve_subri_h, uint16_t, DO_SUBR) DO_ZZI(sve_subri_s, uint32_t, DO_SUBR) DO_ZZI(sve_subri_d, uint64_t, DO_SUBR) DO_ZZI(sve_smaxi_b, int8_t, DO_MAX) DO_ZZI(sve_smaxi_h, int16_t, DO_MAX) DO_ZZI(sve_smaxi_s, int32_t, DO_MAX) DO_ZZI(sve_smaxi_d, int64_t, DO_MAX) DO_ZZI(sve_smini_b, int8_t, DO_MIN) DO_ZZI(sve_smini_h, int16_t, DO_MIN) DO_ZZI(sve_smini_s, int32_t, DO_MIN) DO_ZZI(sve_smini_d, int64_t, DO_MIN) DO_ZZI(sve_umaxi_b, uint8_t, DO_MAX) DO_ZZI(sve_umaxi_h, uint16_t, DO_MAX) DO_ZZI(sve_umaxi_s, uint32_t, DO_MAX) DO_ZZI(sve_umaxi_d, uint64_t, DO_MAX) DO_ZZI(sve_umini_b, uint8_t, DO_MIN) DO_ZZI(sve_umini_h, uint16_t, DO_MIN) DO_ZZI(sve_umini_s, uint32_t, DO_MIN) DO_ZZI(sve_umini_d, uint64_t, DO_MIN) #undef DO_ZZI #undef DO_AND #undef DO_ORR #undef DO_EOR #undef DO_BIC #undef DO_ADD #undef DO_SUB #undef DO_MAX #undef DO_MIN #undef DO_ABD #undef DO_MUL #undef DO_DIV #undef DO_ASR #undef DO_LSR #undef DO_LSL #undef DO_SUBR /* Similar to the ARM LastActiveElement pseudocode function, except the result is multiplied by the element size. This includes the not found indication; e.g. not found for esz=3 is -8. */ static intptr_t last_active_element(uint64_t *g, intptr_t words, intptr_t esz) { uint64_t mask = pred_esz_masks[esz]; intptr_t i = words; do { uint64_t this_g = g[--i] & mask; if (this_g) { return i * 64 + (63 - clz64(this_g)); } } while (i > 0); return (intptr_t)-1 << esz; } uint32_t HELPER(sve_pfirst)(void *vd, void *vg, uint32_t pred_desc) { intptr_t words = DIV_ROUND_UP(FIELD_EX32(pred_desc, PREDDESC, OPRSZ), 8); uint32_t flags = PREDTEST_INIT; uint64_t *d = vd, *g = vg; intptr_t i = 0; do { uint64_t this_d = d[i]; uint64_t this_g = g[i]; if (this_g) { if (!(flags & 4)) { /* Set in D the first bit of G. */ this_d |= this_g & -this_g; d[i] = this_d; } flags = iter_predtest_fwd(this_d, this_g, flags); } } while (++i < words); return flags; } uint32_t HELPER(sve_pnext)(void *vd, void *vg, uint32_t pred_desc) { intptr_t words = DIV_ROUND_UP(FIELD_EX32(pred_desc, PREDDESC, OPRSZ), 8); intptr_t esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); uint32_t flags = PREDTEST_INIT; uint64_t *d = vd, *g = vg, esz_mask; intptr_t i, next; next = last_active_element(vd, words, esz) + (1 << esz); esz_mask = pred_esz_masks[esz]; /* Similar to the pseudocode for pnext, but scaled by ESZ so that we find the correct bit. */ if (next < words * 64) { uint64_t mask = -1; if (next & 63) { mask = ~((1ull << (next & 63)) - 1); next &= -64; } do { uint64_t this_g = g[next / 64] & esz_mask & mask; if (this_g != 0) { next = (next & -64) + ctz64(this_g); break; } next += 64; mask = -1; } while (next < words * 64); } i = 0; do { uint64_t this_d = 0; if (i == next / 64) { this_d = 1ull << (next & 63); } d[i] = this_d; flags = iter_predtest_fwd(this_d, g[i] & esz_mask, flags); } while (++i < words); return flags; } /* * Copy Zn into Zd, and store zero into inactive elements. * If inv, store zeros into the active elements. */ void HELPER(sve_movz_b)(void *vd, void *vn, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t inv = -(uint64_t)(simd_data(desc) & 1); uint64_t *d = vd, *n = vn; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] & (expand_pred_b(pg[H1(i)]) ^ inv); } } void HELPER(sve_movz_h)(void *vd, void *vn, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t inv = -(uint64_t)(simd_data(desc) & 1); uint64_t *d = vd, *n = vn; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] & (expand_pred_h(pg[H1(i)]) ^ inv); } } void HELPER(sve_movz_s)(void *vd, void *vn, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t inv = -(uint64_t)(simd_data(desc) & 1); uint64_t *d = vd, *n = vn; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] & (expand_pred_s(pg[H1(i)]) ^ inv); } } void HELPER(sve_movz_d)(void *vd, void *vn, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; uint8_t *pg = vg; uint8_t inv = simd_data(desc); for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] & -(uint64_t)((pg[H1(i)] ^ inv) & 1); } } /* Three-operand expander, immediate operand, controlled by a predicate. */ #define DO_ZPZI(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ TYPE imm = simd_data(desc); \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ if (pg & 1) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, imm); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ } /* Similarly, specialized for 64-bit operands. */ #define DO_ZPZI_D(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / 8; \ TYPE *d = vd, *n = vn; \ TYPE imm = simd_data(desc); \ uint8_t *pg = vg; \ for (i = 0; i < opr_sz; i += 1) { \ if (pg[H1(i)] & 1) { \ TYPE nn = n[i]; \ d[i] = OP(nn, imm); \ } \ } \ } #define DO_SHR(N, M) (N >> M) #define DO_SHL(N, M) (N << M) /* Arithmetic shift right for division. This rounds negative numbers toward zero as per signed division. Therefore before shifting, when N is negative, add 2**M-1. */ #define DO_ASRD(N, M) ((N + (N < 0 ? ((__typeof(N))1 << M) - 1 : 0)) >> M) static inline uint64_t do_urshr(uint64_t x, unsigned sh) { if (likely(sh < 64)) { return (x >> sh) + ((x >> (sh - 1)) & 1); } else if (sh == 64) { return x >> 63; } else { return 0; } } static inline int64_t do_srshr(int64_t x, unsigned sh) { if (likely(sh < 64)) { return (x >> sh) + ((x >> (sh - 1)) & 1); } else { /* Rounding the sign bit always produces 0. */ return 0; } } DO_ZPZI(sve_asr_zpzi_b, int8_t, H1, DO_SHR) DO_ZPZI(sve_asr_zpzi_h, int16_t, H1_2, DO_SHR) DO_ZPZI(sve_asr_zpzi_s, int32_t, H1_4, DO_SHR) DO_ZPZI_D(sve_asr_zpzi_d, int64_t, DO_SHR) DO_ZPZI(sve_lsr_zpzi_b, uint8_t, H1, DO_SHR) DO_ZPZI(sve_lsr_zpzi_h, uint16_t, H1_2, DO_SHR) DO_ZPZI(sve_lsr_zpzi_s, uint32_t, H1_4, DO_SHR) DO_ZPZI_D(sve_lsr_zpzi_d, uint64_t, DO_SHR) DO_ZPZI(sve_lsl_zpzi_b, uint8_t, H1, DO_SHL) DO_ZPZI(sve_lsl_zpzi_h, uint16_t, H1_2, DO_SHL) DO_ZPZI(sve_lsl_zpzi_s, uint32_t, H1_4, DO_SHL) DO_ZPZI_D(sve_lsl_zpzi_d, uint64_t, DO_SHL) DO_ZPZI(sve_asrd_b, int8_t, H1, DO_ASRD) DO_ZPZI(sve_asrd_h, int16_t, H1_2, DO_ASRD) DO_ZPZI(sve_asrd_s, int32_t, H1_4, DO_ASRD) DO_ZPZI_D(sve_asrd_d, int64_t, DO_ASRD) /* SVE2 bitwise shift by immediate */ DO_ZPZI(sve2_sqshl_zpzi_b, int8_t, H1, do_sqshl_b) DO_ZPZI(sve2_sqshl_zpzi_h, int16_t, H1_2, do_sqshl_h) DO_ZPZI(sve2_sqshl_zpzi_s, int32_t, H1_4, do_sqshl_s) DO_ZPZI_D(sve2_sqshl_zpzi_d, int64_t, do_sqshl_d) DO_ZPZI(sve2_uqshl_zpzi_b, uint8_t, H1, do_uqshl_b) DO_ZPZI(sve2_uqshl_zpzi_h, uint16_t, H1_2, do_uqshl_h) DO_ZPZI(sve2_uqshl_zpzi_s, uint32_t, H1_4, do_uqshl_s) DO_ZPZI_D(sve2_uqshl_zpzi_d, uint64_t, do_uqshl_d) DO_ZPZI(sve2_srshr_b, int8_t, H1, do_srshr) DO_ZPZI(sve2_srshr_h, int16_t, H1_2, do_srshr) DO_ZPZI(sve2_srshr_s, int32_t, H1_4, do_srshr) DO_ZPZI_D(sve2_srshr_d, int64_t, do_srshr) DO_ZPZI(sve2_urshr_b, uint8_t, H1, do_urshr) DO_ZPZI(sve2_urshr_h, uint16_t, H1_2, do_urshr) DO_ZPZI(sve2_urshr_s, uint32_t, H1_4, do_urshr) DO_ZPZI_D(sve2_urshr_d, uint64_t, do_urshr) #define do_suqrshl_b(n, m) \ ({ uint32_t discard; do_suqrshl_bhs(n, (int8_t)m, 8, false, &discard); }) #define do_suqrshl_h(n, m) \ ({ uint32_t discard; do_suqrshl_bhs(n, (int16_t)m, 16, false, &discard); }) #define do_suqrshl_s(n, m) \ ({ uint32_t discard; do_suqrshl_bhs(n, m, 32, false, &discard); }) #define do_suqrshl_d(n, m) \ ({ uint32_t discard; do_suqrshl_d(n, m, false, &discard); }) DO_ZPZI(sve2_sqshlu_b, int8_t, H1, do_suqrshl_b) DO_ZPZI(sve2_sqshlu_h, int16_t, H1_2, do_suqrshl_h) DO_ZPZI(sve2_sqshlu_s, int32_t, H1_4, do_suqrshl_s) DO_ZPZI_D(sve2_sqshlu_d, int64_t, do_suqrshl_d) #undef DO_ASRD #undef DO_ZPZI #undef DO_ZPZI_D #define DO_SHRNB(NAME, TYPEW, TYPEN, OP) \ void HELPER(NAME)(void *vd, void *vn, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ int shift = simd_data(desc); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEW *)(vn + i); \ *(TYPEW *)(vd + i) = (TYPEN)OP(nn, shift); \ } \ } #define DO_SHRNT(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ int shift = simd_data(desc); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEW *)(vn + HW(i)); \ *(TYPEN *)(vd + HN(i + sizeof(TYPEN))) = OP(nn, shift); \ } \ } DO_SHRNB(sve2_shrnb_h, uint16_t, uint8_t, DO_SHR) DO_SHRNB(sve2_shrnb_s, uint32_t, uint16_t, DO_SHR) DO_SHRNB(sve2_shrnb_d, uint64_t, uint32_t, DO_SHR) DO_SHRNT(sve2_shrnt_h, uint16_t, uint8_t, H1_2, H1, DO_SHR) DO_SHRNT(sve2_shrnt_s, uint32_t, uint16_t, H1_4, H1_2, DO_SHR) DO_SHRNT(sve2_shrnt_d, uint64_t, uint32_t, H1_8, H1_4, DO_SHR) DO_SHRNB(sve2_rshrnb_h, uint16_t, uint8_t, do_urshr) DO_SHRNB(sve2_rshrnb_s, uint32_t, uint16_t, do_urshr) DO_SHRNB(sve2_rshrnb_d, uint64_t, uint32_t, do_urshr) DO_SHRNT(sve2_rshrnt_h, uint16_t, uint8_t, H1_2, H1, do_urshr) DO_SHRNT(sve2_rshrnt_s, uint32_t, uint16_t, H1_4, H1_2, do_urshr) DO_SHRNT(sve2_rshrnt_d, uint64_t, uint32_t, H1_8, H1_4, do_urshr) #define DO_SQSHRUN_H(x, sh) do_sat_bhs((int64_t)(x) >> sh, 0, UINT8_MAX) #define DO_SQSHRUN_S(x, sh) do_sat_bhs((int64_t)(x) >> sh, 0, UINT16_MAX) #define DO_SQSHRUN_D(x, sh) \ do_sat_bhs((int64_t)(x) >> (sh < 64 ? sh : 63), 0, UINT32_MAX) DO_SHRNB(sve2_sqshrunb_h, int16_t, uint8_t, DO_SQSHRUN_H) DO_SHRNB(sve2_sqshrunb_s, int32_t, uint16_t, DO_SQSHRUN_S) DO_SHRNB(sve2_sqshrunb_d, int64_t, uint32_t, DO_SQSHRUN_D) DO_SHRNT(sve2_sqshrunt_h, int16_t, uint8_t, H1_2, H1, DO_SQSHRUN_H) DO_SHRNT(sve2_sqshrunt_s, int32_t, uint16_t, H1_4, H1_2, DO_SQSHRUN_S) DO_SHRNT(sve2_sqshrunt_d, int64_t, uint32_t, H1_8, H1_4, DO_SQSHRUN_D) #define DO_SQRSHRUN_H(x, sh) do_sat_bhs(do_srshr(x, sh), 0, UINT8_MAX) #define DO_SQRSHRUN_S(x, sh) do_sat_bhs(do_srshr(x, sh), 0, UINT16_MAX) #define DO_SQRSHRUN_D(x, sh) do_sat_bhs(do_srshr(x, sh), 0, UINT32_MAX) DO_SHRNB(sve2_sqrshrunb_h, int16_t, uint8_t, DO_SQRSHRUN_H) DO_SHRNB(sve2_sqrshrunb_s, int32_t, uint16_t, DO_SQRSHRUN_S) DO_SHRNB(sve2_sqrshrunb_d, int64_t, uint32_t, DO_SQRSHRUN_D) DO_SHRNT(sve2_sqrshrunt_h, int16_t, uint8_t, H1_2, H1, DO_SQRSHRUN_H) DO_SHRNT(sve2_sqrshrunt_s, int32_t, uint16_t, H1_4, H1_2, DO_SQRSHRUN_S) DO_SHRNT(sve2_sqrshrunt_d, int64_t, uint32_t, H1_8, H1_4, DO_SQRSHRUN_D) #define DO_SQSHRN_H(x, sh) do_sat_bhs(x >> sh, INT8_MIN, INT8_MAX) #define DO_SQSHRN_S(x, sh) do_sat_bhs(x >> sh, INT16_MIN, INT16_MAX) #define DO_SQSHRN_D(x, sh) do_sat_bhs(x >> sh, INT32_MIN, INT32_MAX) DO_SHRNB(sve2_sqshrnb_h, int16_t, uint8_t, DO_SQSHRN_H) DO_SHRNB(sve2_sqshrnb_s, int32_t, uint16_t, DO_SQSHRN_S) DO_SHRNB(sve2_sqshrnb_d, int64_t, uint32_t, DO_SQSHRN_D) DO_SHRNT(sve2_sqshrnt_h, int16_t, uint8_t, H1_2, H1, DO_SQSHRN_H) DO_SHRNT(sve2_sqshrnt_s, int32_t, uint16_t, H1_4, H1_2, DO_SQSHRN_S) DO_SHRNT(sve2_sqshrnt_d, int64_t, uint32_t, H1_8, H1_4, DO_SQSHRN_D) #define DO_SQRSHRN_H(x, sh) do_sat_bhs(do_srshr(x, sh), INT8_MIN, INT8_MAX) #define DO_SQRSHRN_S(x, sh) do_sat_bhs(do_srshr(x, sh), INT16_MIN, INT16_MAX) #define DO_SQRSHRN_D(x, sh) do_sat_bhs(do_srshr(x, sh), INT32_MIN, INT32_MAX) DO_SHRNB(sve2_sqrshrnb_h, int16_t, uint8_t, DO_SQRSHRN_H) DO_SHRNB(sve2_sqrshrnb_s, int32_t, uint16_t, DO_SQRSHRN_S) DO_SHRNB(sve2_sqrshrnb_d, int64_t, uint32_t, DO_SQRSHRN_D) DO_SHRNT(sve2_sqrshrnt_h, int16_t, uint8_t, H1_2, H1, DO_SQRSHRN_H) DO_SHRNT(sve2_sqrshrnt_s, int32_t, uint16_t, H1_4, H1_2, DO_SQRSHRN_S) DO_SHRNT(sve2_sqrshrnt_d, int64_t, uint32_t, H1_8, H1_4, DO_SQRSHRN_D) #define DO_UQSHRN_H(x, sh) MIN(x >> sh, UINT8_MAX) #define DO_UQSHRN_S(x, sh) MIN(x >> sh, UINT16_MAX) #define DO_UQSHRN_D(x, sh) MIN(x >> sh, UINT32_MAX) DO_SHRNB(sve2_uqshrnb_h, uint16_t, uint8_t, DO_UQSHRN_H) DO_SHRNB(sve2_uqshrnb_s, uint32_t, uint16_t, DO_UQSHRN_S) DO_SHRNB(sve2_uqshrnb_d, uint64_t, uint32_t, DO_UQSHRN_D) DO_SHRNT(sve2_uqshrnt_h, uint16_t, uint8_t, H1_2, H1, DO_UQSHRN_H) DO_SHRNT(sve2_uqshrnt_s, uint32_t, uint16_t, H1_4, H1_2, DO_UQSHRN_S) DO_SHRNT(sve2_uqshrnt_d, uint64_t, uint32_t, H1_8, H1_4, DO_UQSHRN_D) #define DO_UQRSHRN_H(x, sh) MIN(do_urshr(x, sh), UINT8_MAX) #define DO_UQRSHRN_S(x, sh) MIN(do_urshr(x, sh), UINT16_MAX) #define DO_UQRSHRN_D(x, sh) MIN(do_urshr(x, sh), UINT32_MAX) DO_SHRNB(sve2_uqrshrnb_h, uint16_t, uint8_t, DO_UQRSHRN_H) DO_SHRNB(sve2_uqrshrnb_s, uint32_t, uint16_t, DO_UQRSHRN_S) DO_SHRNB(sve2_uqrshrnb_d, uint64_t, uint32_t, DO_UQRSHRN_D) DO_SHRNT(sve2_uqrshrnt_h, uint16_t, uint8_t, H1_2, H1, DO_UQRSHRN_H) DO_SHRNT(sve2_uqrshrnt_s, uint32_t, uint16_t, H1_4, H1_2, DO_UQRSHRN_S) DO_SHRNT(sve2_uqrshrnt_d, uint64_t, uint32_t, H1_8, H1_4, DO_UQRSHRN_D) #undef DO_SHRNB #undef DO_SHRNT #define DO_BINOPNB(NAME, TYPEW, TYPEN, SHIFT, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEW *)(vn + i); \ TYPEW mm = *(TYPEW *)(vm + i); \ *(TYPEW *)(vd + i) = (TYPEN)OP(nn, mm, SHIFT); \ } \ } #define DO_BINOPNT(NAME, TYPEW, TYPEN, SHIFT, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; i += sizeof(TYPEW)) { \ TYPEW nn = *(TYPEW *)(vn + HW(i)); \ TYPEW mm = *(TYPEW *)(vm + HW(i)); \ *(TYPEN *)(vd + HN(i + sizeof(TYPEN))) = OP(nn, mm, SHIFT); \ } \ } #define DO_ADDHN(N, M, SH) ((N + M) >> SH) #define DO_RADDHN(N, M, SH) ((N + M + ((__typeof(N))1 << (SH - 1))) >> SH) #define DO_SUBHN(N, M, SH) ((N - M) >> SH) #define DO_RSUBHN(N, M, SH) ((N - M + ((__typeof(N))1 << (SH - 1))) >> SH) DO_BINOPNB(sve2_addhnb_h, uint16_t, uint8_t, 8, DO_ADDHN) DO_BINOPNB(sve2_addhnb_s, uint32_t, uint16_t, 16, DO_ADDHN) DO_BINOPNB(sve2_addhnb_d, uint64_t, uint32_t, 32, DO_ADDHN) DO_BINOPNT(sve2_addhnt_h, uint16_t, uint8_t, 8, H1_2, H1, DO_ADDHN) DO_BINOPNT(sve2_addhnt_s, uint32_t, uint16_t, 16, H1_4, H1_2, DO_ADDHN) DO_BINOPNT(sve2_addhnt_d, uint64_t, uint32_t, 32, H1_8, H1_4, DO_ADDHN) DO_BINOPNB(sve2_raddhnb_h, uint16_t, uint8_t, 8, DO_RADDHN) DO_BINOPNB(sve2_raddhnb_s, uint32_t, uint16_t, 16, DO_RADDHN) DO_BINOPNB(sve2_raddhnb_d, uint64_t, uint32_t, 32, DO_RADDHN) DO_BINOPNT(sve2_raddhnt_h, uint16_t, uint8_t, 8, H1_2, H1, DO_RADDHN) DO_BINOPNT(sve2_raddhnt_s, uint32_t, uint16_t, 16, H1_4, H1_2, DO_RADDHN) DO_BINOPNT(sve2_raddhnt_d, uint64_t, uint32_t, 32, H1_8, H1_4, DO_RADDHN) DO_BINOPNB(sve2_subhnb_h, uint16_t, uint8_t, 8, DO_SUBHN) DO_BINOPNB(sve2_subhnb_s, uint32_t, uint16_t, 16, DO_SUBHN) DO_BINOPNB(sve2_subhnb_d, uint64_t, uint32_t, 32, DO_SUBHN) DO_BINOPNT(sve2_subhnt_h, uint16_t, uint8_t, 8, H1_2, H1, DO_SUBHN) DO_BINOPNT(sve2_subhnt_s, uint32_t, uint16_t, 16, H1_4, H1_2, DO_SUBHN) DO_BINOPNT(sve2_subhnt_d, uint64_t, uint32_t, 32, H1_8, H1_4, DO_SUBHN) DO_BINOPNB(sve2_rsubhnb_h, uint16_t, uint8_t, 8, DO_RSUBHN) DO_BINOPNB(sve2_rsubhnb_s, uint32_t, uint16_t, 16, DO_RSUBHN) DO_BINOPNB(sve2_rsubhnb_d, uint64_t, uint32_t, 32, DO_RSUBHN) DO_BINOPNT(sve2_rsubhnt_h, uint16_t, uint8_t, 8, H1_2, H1, DO_RSUBHN) DO_BINOPNT(sve2_rsubhnt_s, uint32_t, uint16_t, 16, H1_4, H1_2, DO_RSUBHN) DO_BINOPNT(sve2_rsubhnt_d, uint64_t, uint32_t, 32, H1_8, H1_4, DO_RSUBHN) #undef DO_RSUBHN #undef DO_SUBHN #undef DO_RADDHN #undef DO_ADDHN #undef DO_BINOPNB /* Fully general four-operand expander, controlled by a predicate. */ #define DO_ZPZZZ(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *va, void *vn, void *vm, \ void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ for (i = 0; i < opr_sz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ if (pg & 1) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ TYPE mm = *(TYPE *)(vm + H(i)); \ TYPE aa = *(TYPE *)(va + H(i)); \ *(TYPE *)(vd + H(i)) = OP(aa, nn, mm); \ } \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ } /* Similarly, specialized for 64-bit operands. */ #define DO_ZPZZZ_D(NAME, TYPE, OP) \ void HELPER(NAME)(void *vd, void *va, void *vn, void *vm, \ void *vg, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc) / 8; \ TYPE *d = vd, *a = va, *n = vn, *m = vm; \ uint8_t *pg = vg; \ for (i = 0; i < opr_sz; i += 1) { \ if (pg[H1(i)] & 1) { \ TYPE aa = a[i], nn = n[i], mm = m[i]; \ d[i] = OP(aa, nn, mm); \ } \ } \ } #define DO_MLA(A, N, M) (A + N * M) #define DO_MLS(A, N, M) (A - N * M) DO_ZPZZZ(sve_mla_b, uint8_t, H1, DO_MLA) DO_ZPZZZ(sve_mls_b, uint8_t, H1, DO_MLS) DO_ZPZZZ(sve_mla_h, uint16_t, H1_2, DO_MLA) DO_ZPZZZ(sve_mls_h, uint16_t, H1_2, DO_MLS) DO_ZPZZZ(sve_mla_s, uint32_t, H1_4, DO_MLA) DO_ZPZZZ(sve_mls_s, uint32_t, H1_4, DO_MLS) DO_ZPZZZ_D(sve_mla_d, uint64_t, DO_MLA) DO_ZPZZZ_D(sve_mls_d, uint64_t, DO_MLS) #undef DO_MLA #undef DO_MLS #undef DO_ZPZZZ #undef DO_ZPZZZ_D void HELPER(sve_index_b)(void *vd, uint32_t start, uint32_t incr, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc); uint8_t *d = vd; for (i = 0; i < opr_sz; i += 1) { d[H1(i)] = start + i * incr; } } void HELPER(sve_index_h)(void *vd, uint32_t start, uint32_t incr, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 2; uint16_t *d = vd; for (i = 0; i < opr_sz; i += 1) { d[H2(i)] = start + i * incr; } } void HELPER(sve_index_s)(void *vd, uint32_t start, uint32_t incr, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 4; uint32_t *d = vd; for (i = 0; i < opr_sz; i += 1) { d[H4(i)] = start + i * incr; } } void HELPER(sve_index_d)(void *vd, uint64_t start, uint64_t incr, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd; for (i = 0; i < opr_sz; i += 1) { d[i] = start + i * incr; } } void HELPER(sve_adr_p32)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 4; uint32_t sh = simd_data(desc); uint32_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] + (m[i] << sh); } } void HELPER(sve_adr_p64)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t sh = simd_data(desc); uint64_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] + (m[i] << sh); } } void HELPER(sve_adr_s32)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t sh = simd_data(desc); uint64_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] + ((uint64_t)(int32_t)m[i] << sh); } } void HELPER(sve_adr_u32)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t sh = simd_data(desc); uint64_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { d[i] = n[i] + ((uint64_t)(uint32_t)m[i] << sh); } } void HELPER(sve_fexpa_h)(void *vd, void *vn, uint32_t desc) { /* These constants are cut-and-paste directly from the ARM pseudocode. */ static const uint16_t coeff[] = { 0x0000, 0x0016, 0x002d, 0x0045, 0x005d, 0x0075, 0x008e, 0x00a8, 0x00c2, 0x00dc, 0x00f8, 0x0114, 0x0130, 0x014d, 0x016b, 0x0189, 0x01a8, 0x01c8, 0x01e8, 0x0209, 0x022b, 0x024e, 0x0271, 0x0295, 0x02ba, 0x02e0, 0x0306, 0x032e, 0x0356, 0x037f, 0x03a9, 0x03d4, }; intptr_t i, opr_sz = simd_oprsz(desc) / 2; uint16_t *d = vd, *n = vn; for (i = 0; i < opr_sz; i++) { uint16_t nn = n[i]; intptr_t idx = extract32(nn, 0, 5); uint16_t exp = extract32(nn, 5, 5); d[i] = coeff[idx] | (exp << 10); } } void HELPER(sve_fexpa_s)(void *vd, void *vn, uint32_t desc) { /* These constants are cut-and-paste directly from the ARM pseudocode. */ static const uint32_t coeff[] = { 0x000000, 0x0164d2, 0x02cd87, 0x043a29, 0x05aac3, 0x071f62, 0x08980f, 0x0a14d5, 0x0b95c2, 0x0d1adf, 0x0ea43a, 0x1031dc, 0x11c3d3, 0x135a2b, 0x14f4f0, 0x16942d, 0x1837f0, 0x19e046, 0x1b8d3a, 0x1d3eda, 0x1ef532, 0x20b051, 0x227043, 0x243516, 0x25fed7, 0x27cd94, 0x29a15b, 0x2b7a3a, 0x2d583f, 0x2f3b79, 0x3123f6, 0x3311c4, 0x3504f3, 0x36fd92, 0x38fbaf, 0x3aff5b, 0x3d08a4, 0x3f179a, 0x412c4d, 0x4346cd, 0x45672a, 0x478d75, 0x49b9be, 0x4bec15, 0x4e248c, 0x506334, 0x52a81e, 0x54f35b, 0x5744fd, 0x599d16, 0x5bfbb8, 0x5e60f5, 0x60ccdf, 0x633f89, 0x65b907, 0x68396a, 0x6ac0c7, 0x6d4f30, 0x6fe4ba, 0x728177, 0x75257d, 0x77d0df, 0x7a83b3, 0x7d3e0c, }; intptr_t i, opr_sz = simd_oprsz(desc) / 4; uint32_t *d = vd, *n = vn; for (i = 0; i < opr_sz; i++) { uint32_t nn = n[i]; intptr_t idx = extract32(nn, 0, 6); uint32_t exp = extract32(nn, 6, 8); d[i] = coeff[idx] | (exp << 23); } } void HELPER(sve_fexpa_d)(void *vd, void *vn, uint32_t desc) { /* These constants are cut-and-paste directly from the ARM pseudocode. */ static const uint64_t coeff[] = { 0x0000000000000ull, 0x02C9A3E778061ull, 0x059B0D3158574ull, 0x0874518759BC8ull, 0x0B5586CF9890Full, 0x0E3EC32D3D1A2ull, 0x11301D0125B51ull, 0x1429AAEA92DE0ull, 0x172B83C7D517Bull, 0x1A35BEB6FCB75ull, 0x1D4873168B9AAull, 0x2063B88628CD6ull, 0x2387A6E756238ull, 0x26B4565E27CDDull, 0x29E9DF51FDEE1ull, 0x2D285A6E4030Bull, 0x306FE0A31B715ull, 0x33C08B26416FFull, 0x371A7373AA9CBull, 0x3A7DB34E59FF7ull, 0x3DEA64C123422ull, 0x4160A21F72E2Aull, 0x44E086061892Dull, 0x486A2B5C13CD0ull, 0x4BFDAD5362A27ull, 0x4F9B2769D2CA7ull, 0x5342B569D4F82ull, 0x56F4736B527DAull, 0x5AB07DD485429ull, 0x5E76F15AD2148ull, 0x6247EB03A5585ull, 0x6623882552225ull, 0x6A09E667F3BCDull, 0x6DFB23C651A2Full, 0x71F75E8EC5F74ull, 0x75FEB564267C9ull, 0x7A11473EB0187ull, 0x7E2F336CF4E62ull, 0x82589994CCE13ull, 0x868D99B4492EDull, 0x8ACE5422AA0DBull, 0x8F1AE99157736ull, 0x93737B0CDC5E5ull, 0x97D829FDE4E50ull, 0x9C49182A3F090ull, 0xA0C667B5DE565ull, 0xA5503B23E255Dull, 0xA9E6B5579FDBFull, 0xAE89F995AD3ADull, 0xB33A2B84F15FBull, 0xB7F76F2FB5E47ull, 0xBCC1E904BC1D2ull, 0xC199BDD85529Cull, 0xC67F12E57D14Bull, 0xCB720DCEF9069ull, 0xD072D4A07897Cull, 0xD5818DCFBA487ull, 0xDA9E603DB3285ull, 0xDFC97337B9B5Full, 0xE502EE78B3FF6ull, 0xEA4AFA2A490DAull, 0xEFA1BEE615A27ull, 0xF50765B6E4540ull, 0xFA7C1819E90D8ull, }; intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; for (i = 0; i < opr_sz; i++) { uint64_t nn = n[i]; intptr_t idx = extract32(nn, 0, 6); uint64_t exp = extract32(nn, 6, 11); d[i] = coeff[idx] | (exp << 52); } } void HELPER(sve_ftssel_h)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 2; uint16_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { uint16_t nn = n[i]; uint16_t mm = m[i]; if (mm & 1) { nn = float16_one; } d[i] = nn ^ (mm & 2) << 14; } } void HELPER(sve_ftssel_s)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 4; uint32_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { uint32_t nn = n[i]; uint32_t mm = m[i]; if (mm & 1) { nn = float32_one; } d[i] = nn ^ (mm & 2) << 30; } } void HELPER(sve_ftssel_d)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i]; uint64_t mm = m[i]; if (mm & 1) { nn = float64_one; } d[i] = nn ^ (mm & 2) << 62; } } /* * Signed saturating addition with scalar operand. */ void HELPER(sve_sqaddi_b)(void *d, void *a, int32_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(int8_t)) { *(int8_t *)(d + i) = DO_SQADD_B(b, *(int8_t *)(a + i)); } } void HELPER(sve_sqaddi_h)(void *d, void *a, int32_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(int16_t)) { *(int16_t *)(d + i) = DO_SQADD_H(b, *(int16_t *)(a + i)); } } void HELPER(sve_sqaddi_s)(void *d, void *a, int64_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(int32_t)) { *(int32_t *)(d + i) = DO_SQADD_S(b, *(int32_t *)(a + i)); } } void HELPER(sve_sqaddi_d)(void *d, void *a, int64_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(int64_t)) { *(int64_t *)(d + i) = do_sqadd_d(b, *(int64_t *)(a + i)); } } /* * Unsigned saturating addition with scalar operand. */ void HELPER(sve_uqaddi_b)(void *d, void *a, int32_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(uint8_t)) { *(uint8_t *)(d + i) = DO_UQADD_B(b, *(uint8_t *)(a + i)); } } void HELPER(sve_uqaddi_h)(void *d, void *a, int32_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(uint16_t)) { *(uint16_t *)(d + i) = DO_UQADD_H(b, *(uint16_t *)(a + i)); } } void HELPER(sve_uqaddi_s)(void *d, void *a, int64_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(uint32_t)) { *(uint32_t *)(d + i) = DO_UQADD_S(b, *(uint32_t *)(a + i)); } } void HELPER(sve_uqaddi_d)(void *d, void *a, uint64_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(uint64_t)) { *(uint64_t *)(d + i) = do_uqadd_d(b, *(uint64_t *)(a + i)); } } void HELPER(sve_uqsubi_d)(void *d, void *a, uint64_t b, uint32_t desc) { intptr_t i, oprsz = simd_oprsz(desc); for (i = 0; i < oprsz; i += sizeof(uint64_t)) { *(uint64_t *)(d + i) = do_uqsub_d(*(uint64_t *)(a + i), b); } } /* Two operand predicated copy immediate with merge. All valid immediates * can fit within 17 signed bits in the simd_data field. */ void HELPER(sve_cpy_m_b)(void *vd, void *vn, void *vg, uint64_t mm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; uint8_t *pg = vg; mm = dup_const(MO_8, mm); for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i]; uint64_t pp = expand_pred_b(pg[H1(i)]); d[i] = (mm & pp) | (nn & ~pp); } } void HELPER(sve_cpy_m_h)(void *vd, void *vn, void *vg, uint64_t mm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; uint8_t *pg = vg; mm = dup_const(MO_16, mm); for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i]; uint64_t pp = expand_pred_h(pg[H1(i)]); d[i] = (mm & pp) | (nn & ~pp); } } void HELPER(sve_cpy_m_s)(void *vd, void *vn, void *vg, uint64_t mm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; uint8_t *pg = vg; mm = dup_const(MO_32, mm); for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i]; uint64_t pp = expand_pred_s(pg[H1(i)]); d[i] = (mm & pp) | (nn & ~pp); } } void HELPER(sve_cpy_m_d)(void *vd, void *vn, void *vg, uint64_t mm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i]; d[i] = (pg[H1(i)] & 1 ? mm : nn); } } void HELPER(sve_cpy_z_b)(void *vd, void *vg, uint64_t val, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd; uint8_t *pg = vg; val = dup_const(MO_8, val); for (i = 0; i < opr_sz; i += 1) { d[i] = val & expand_pred_b(pg[H1(i)]); } } void HELPER(sve_cpy_z_h)(void *vd, void *vg, uint64_t val, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd; uint8_t *pg = vg; val = dup_const(MO_16, val); for (i = 0; i < opr_sz; i += 1) { d[i] = val & expand_pred_h(pg[H1(i)]); } } void HELPER(sve_cpy_z_s)(void *vd, void *vg, uint64_t val, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd; uint8_t *pg = vg; val = dup_const(MO_32, val); for (i = 0; i < opr_sz; i += 1) { d[i] = val & expand_pred_s(pg[H1(i)]); } } void HELPER(sve_cpy_z_d)(void *vd, void *vg, uint64_t val, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { d[i] = (pg[H1(i)] & 1 ? val : 0); } } /* Big-endian hosts need to frob the byte indices. If the copy * happens to be 8-byte aligned, then no frobbing necessary. */ static void swap_memmove(void *vd, void *vs, size_t n) { uintptr_t d = (uintptr_t)vd; uintptr_t s = (uintptr_t)vs; uintptr_t o = (d | s | n) & 7; size_t i; #if !HOST_BIG_ENDIAN o = 0; #endif switch (o) { case 0: memmove(vd, vs, n); break; case 4: if (d < s || d >= s + n) { for (i = 0; i < n; i += 4) { *(uint32_t *)H1_4(d + i) = *(uint32_t *)H1_4(s + i); } } else { for (i = n; i > 0; ) { i -= 4; *(uint32_t *)H1_4(d + i) = *(uint32_t *)H1_4(s + i); } } break; case 2: case 6: if (d < s || d >= s + n) { for (i = 0; i < n; i += 2) { *(uint16_t *)H1_2(d + i) = *(uint16_t *)H1_2(s + i); } } else { for (i = n; i > 0; ) { i -= 2; *(uint16_t *)H1_2(d + i) = *(uint16_t *)H1_2(s + i); } } break; default: if (d < s || d >= s + n) { for (i = 0; i < n; i++) { *(uint8_t *)H1(d + i) = *(uint8_t *)H1(s + i); } } else { for (i = n; i > 0; ) { i -= 1; *(uint8_t *)H1(d + i) = *(uint8_t *)H1(s + i); } } break; } } /* Similarly for memset of 0. */ static void swap_memzero(void *vd, size_t n) { uintptr_t d = (uintptr_t)vd; uintptr_t o = (d | n) & 7; size_t i; /* Usually, the first bit of a predicate is set, so N is 0. */ if (likely(n == 0)) { return; } #if !HOST_BIG_ENDIAN o = 0; #endif switch (o) { case 0: memset(vd, 0, n); break; case 4: for (i = 0; i < n; i += 4) { *(uint32_t *)H1_4(d + i) = 0; } break; case 2: case 6: for (i = 0; i < n; i += 2) { *(uint16_t *)H1_2(d + i) = 0; } break; default: for (i = 0; i < n; i++) { *(uint8_t *)H1(d + i) = 0; } break; } } void HELPER(sve_ext)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t opr_sz = simd_oprsz(desc); size_t n_ofs = simd_data(desc); size_t n_siz = opr_sz - n_ofs; if (vd != vm) { swap_memmove(vd, vn + n_ofs, n_siz); swap_memmove(vd + n_siz, vm, n_ofs); } else if (vd != vn) { swap_memmove(vd + n_siz, vd, n_ofs); swap_memmove(vd, vn + n_ofs, n_siz); } else { /* vd == vn == vm. Need temp space. */ ARMVectorReg tmp; swap_memmove(&tmp, vm, n_ofs); swap_memmove(vd, vd + n_ofs, n_siz); memcpy(vd + n_siz, &tmp, n_ofs); } } #define DO_INSR(NAME, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, uint64_t val, uint32_t desc) \ { \ intptr_t opr_sz = simd_oprsz(desc); \ swap_memmove(vd + sizeof(TYPE), vn, opr_sz - sizeof(TYPE)); \ *(TYPE *)(vd + H(0)) = val; \ } DO_INSR(sve_insr_b, uint8_t, H1) DO_INSR(sve_insr_h, uint16_t, H1_2) DO_INSR(sve_insr_s, uint32_t, H1_4) DO_INSR(sve_insr_d, uint64_t, H1_8) #undef DO_INSR void HELPER(sve_rev_b)(void *vd, void *vn, uint32_t desc) { intptr_t i, j, opr_sz = simd_oprsz(desc); for (i = 0, j = opr_sz - 8; i < opr_sz / 2; i += 8, j -= 8) { uint64_t f = *(uint64_t *)(vn + i); uint64_t b = *(uint64_t *)(vn + j); *(uint64_t *)(vd + i) = bswap64(b); *(uint64_t *)(vd + j) = bswap64(f); } } void HELPER(sve_rev_h)(void *vd, void *vn, uint32_t desc) { intptr_t i, j, opr_sz = simd_oprsz(desc); for (i = 0, j = opr_sz - 8; i < opr_sz / 2; i += 8, j -= 8) { uint64_t f = *(uint64_t *)(vn + i); uint64_t b = *(uint64_t *)(vn + j); *(uint64_t *)(vd + i) = hswap64(b); *(uint64_t *)(vd + j) = hswap64(f); } } void HELPER(sve_rev_s)(void *vd, void *vn, uint32_t desc) { intptr_t i, j, opr_sz = simd_oprsz(desc); for (i = 0, j = opr_sz - 8; i < opr_sz / 2; i += 8, j -= 8) { uint64_t f = *(uint64_t *)(vn + i); uint64_t b = *(uint64_t *)(vn + j); *(uint64_t *)(vd + i) = rol64(b, 32); *(uint64_t *)(vd + j) = rol64(f, 32); } } void HELPER(sve_rev_d)(void *vd, void *vn, uint32_t desc) { intptr_t i, j, opr_sz = simd_oprsz(desc); for (i = 0, j = opr_sz - 8; i < opr_sz / 2; i += 8, j -= 8) { uint64_t f = *(uint64_t *)(vn + i); uint64_t b = *(uint64_t *)(vn + j); *(uint64_t *)(vd + i) = b; *(uint64_t *)(vd + j) = f; } } typedef void tb_impl_fn(void *, void *, void *, void *, uintptr_t, bool); static inline void do_tbl1(void *vd, void *vn, void *vm, uint32_t desc, bool is_tbx, tb_impl_fn *fn) { ARMVectorReg scratch; uintptr_t oprsz = simd_oprsz(desc); if (unlikely(vd == vn)) { vn = memcpy(&scratch, vn, oprsz); } fn(vd, vn, NULL, vm, oprsz, is_tbx); } static inline void do_tbl2(void *vd, void *vn0, void *vn1, void *vm, uint32_t desc, bool is_tbx, tb_impl_fn *fn) { ARMVectorReg scratch; uintptr_t oprsz = simd_oprsz(desc); if (unlikely(vd == vn0)) { vn0 = memcpy(&scratch, vn0, oprsz); if (vd == vn1) { vn1 = vn0; } } else if (unlikely(vd == vn1)) { vn1 = memcpy(&scratch, vn1, oprsz); } fn(vd, vn0, vn1, vm, oprsz, is_tbx); } #define DO_TB(SUFF, TYPE, H) \ static inline void do_tb_##SUFF(void *vd, void *vt0, void *vt1, \ void *vm, uintptr_t oprsz, bool is_tbx) \ { \ TYPE *d = vd, *tbl0 = vt0, *tbl1 = vt1, *indexes = vm; \ uintptr_t i, nelem = oprsz / sizeof(TYPE); \ for (i = 0; i < nelem; ++i) { \ TYPE index = indexes[H1(i)], val = 0; \ if (index < nelem) { \ val = tbl0[H(index)]; \ } else { \ index -= nelem; \ if (tbl1 && index < nelem) { \ val = tbl1[H(index)]; \ } else if (is_tbx) { \ continue; \ } \ } \ d[H(i)] = val; \ } \ } \ void HELPER(sve_tbl_##SUFF)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ do_tbl1(vd, vn, vm, desc, false, do_tb_##SUFF); \ } \ void HELPER(sve2_tbl_##SUFF)(void *vd, void *vn0, void *vn1, \ void *vm, uint32_t desc) \ { \ do_tbl2(vd, vn0, vn1, vm, desc, false, do_tb_##SUFF); \ } \ void HELPER(sve2_tbx_##SUFF)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ do_tbl1(vd, vn, vm, desc, true, do_tb_##SUFF); \ } DO_TB(b, uint8_t, H1) DO_TB(h, uint16_t, H2) DO_TB(s, uint32_t, H4) DO_TB(d, uint64_t, H8) #undef DO_TB #define DO_UNPK(NAME, TYPED, TYPES, HD, HS) \ void HELPER(NAME)(void *vd, void *vn, uint32_t desc) \ { \ intptr_t i, opr_sz = simd_oprsz(desc); \ TYPED *d = vd; \ TYPES *n = vn; \ ARMVectorReg tmp; \ if (unlikely(vn - vd < opr_sz)) { \ n = memcpy(&tmp, n, opr_sz / 2); \ } \ for (i = 0; i < opr_sz / sizeof(TYPED); i++) { \ d[HD(i)] = n[HS(i)]; \ } \ } DO_UNPK(sve_sunpk_h, int16_t, int8_t, H2, H1) DO_UNPK(sve_sunpk_s, int32_t, int16_t, H4, H2) DO_UNPK(sve_sunpk_d, int64_t, int32_t, H8, H4) DO_UNPK(sve_uunpk_h, uint16_t, uint8_t, H2, H1) DO_UNPK(sve_uunpk_s, uint32_t, uint16_t, H4, H2) DO_UNPK(sve_uunpk_d, uint64_t, uint32_t, H8, H4) #undef DO_UNPK /* Mask of bits included in the even numbered predicates of width esz. * We also use this for expand_bits/compress_bits, and so extend the * same pattern out to 16-bit units. */ static const uint64_t even_bit_esz_masks[5] = { 0x5555555555555555ull, 0x3333333333333333ull, 0x0f0f0f0f0f0f0f0full, 0x00ff00ff00ff00ffull, 0x0000ffff0000ffffull, }; /* Zero-extend units of 2**N bits to units of 2**(N+1) bits. * For N==0, this corresponds to the operation that in qemu/bitops.h * we call half_shuffle64; this algorithm is from Hacker's Delight, * section 7-2 Shuffling Bits. */ static uint64_t expand_bits(uint64_t x, int n) { int i; x &= 0xffffffffu; for (i = 4; i >= n; i--) { int sh = 1 << i; x = ((x << sh) | x) & even_bit_esz_masks[i]; } return x; } /* Compress units of 2**(N+1) bits to units of 2**N bits. * For N==0, this corresponds to the operation that in qemu/bitops.h * we call half_unshuffle64; this algorithm is from Hacker's Delight, * section 7-2 Shuffling Bits, where it is called an inverse half shuffle. */ static uint64_t compress_bits(uint64_t x, int n) { int i; for (i = n; i <= 4; i++) { int sh = 1 << i; x &= even_bit_esz_masks[i]; x = (x >> sh) | x; } return x & 0xffffffffu; } void HELPER(sve_zip_p)(void *vd, void *vn, void *vm, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); int esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); intptr_t high = FIELD_EX32(pred_desc, PREDDESC, DATA); int esize = 1 << esz; uint64_t *d = vd; intptr_t i; if (oprsz <= 8) { uint64_t nn = *(uint64_t *)vn; uint64_t mm = *(uint64_t *)vm; int half = 4 * oprsz; nn = extract64(nn, high * half, half); mm = extract64(mm, high * half, half); nn = expand_bits(nn, esz); mm = expand_bits(mm, esz); d[0] = nn | (mm << esize); } else { ARMPredicateReg tmp; /* We produce output faster than we consume input. Therefore we must be mindful of possible overlap. */ if (vd == vn) { vn = memcpy(&tmp, vn, oprsz); if (vd == vm) { vm = vn; } } else if (vd == vm) { vm = memcpy(&tmp, vm, oprsz); } if (high) { high = oprsz >> 1; } if ((oprsz & 7) == 0) { uint32_t *n = vn, *m = vm; high >>= 2; for (i = 0; i < oprsz / 8; i++) { uint64_t nn = n[H4(high + i)]; uint64_t mm = m[H4(high + i)]; nn = expand_bits(nn, esz); mm = expand_bits(mm, esz); d[i] = nn | (mm << esize); } } else { uint8_t *n = vn, *m = vm; uint16_t *d16 = vd; for (i = 0; i < oprsz / 2; i++) { uint16_t nn = n[H1(high + i)]; uint16_t mm = m[H1(high + i)]; nn = expand_bits(nn, esz); mm = expand_bits(mm, esz); d16[H2(i)] = nn | (mm << esize); } } } } void HELPER(sve_uzp_p)(void *vd, void *vn, void *vm, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); int esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); int odd = FIELD_EX32(pred_desc, PREDDESC, DATA) << esz; uint64_t *d = vd, *n = vn, *m = vm; uint64_t l, h; intptr_t i; if (oprsz <= 8) { l = compress_bits(n[0] >> odd, esz); h = compress_bits(m[0] >> odd, esz); d[0] = l | (h << (4 * oprsz)); } else { ARMPredicateReg tmp_m; intptr_t oprsz_16 = oprsz / 16; if ((vm - vd) < (uintptr_t)oprsz) { m = memcpy(&tmp_m, vm, oprsz); } for (i = 0; i < oprsz_16; i++) { l = n[2 * i + 0]; h = n[2 * i + 1]; l = compress_bits(l >> odd, esz); h = compress_bits(h >> odd, esz); d[i] = l | (h << 32); } /* * For VL which is not a multiple of 512, the results from M do not * align nicely with the uint64_t for D. Put the aligned results * from M into TMP_M and then copy it into place afterward. */ if (oprsz & 15) { int final_shift = (oprsz & 15) * 2; l = n[2 * i + 0]; h = n[2 * i + 1]; l = compress_bits(l >> odd, esz); h = compress_bits(h >> odd, esz); d[i] = l | (h << final_shift); for (i = 0; i < oprsz_16; i++) { l = m[2 * i + 0]; h = m[2 * i + 1]; l = compress_bits(l >> odd, esz); h = compress_bits(h >> odd, esz); tmp_m.p[i] = l | (h << 32); } l = m[2 * i + 0]; h = m[2 * i + 1]; l = compress_bits(l >> odd, esz); h = compress_bits(h >> odd, esz); tmp_m.p[i] = l | (h << final_shift); swap_memmove(vd + oprsz / 2, &tmp_m, oprsz / 2); } else { for (i = 0; i < oprsz_16; i++) { l = m[2 * i + 0]; h = m[2 * i + 1]; l = compress_bits(l >> odd, esz); h = compress_bits(h >> odd, esz); d[oprsz_16 + i] = l | (h << 32); } } } } void HELPER(sve_trn_p)(void *vd, void *vn, void *vm, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); int esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); int odd = FIELD_EX32(pred_desc, PREDDESC, DATA); uint64_t *d = vd, *n = vn, *m = vm; uint64_t mask; int shr, shl; intptr_t i; shl = 1 << esz; shr = 0; mask = even_bit_esz_masks[esz]; if (odd) { mask <<= shl; shr = shl; shl = 0; } for (i = 0; i < DIV_ROUND_UP(oprsz, 8); i++) { uint64_t nn = (n[i] & mask) >> shr; uint64_t mm = (m[i] & mask) << shl; d[i] = nn + mm; } } /* Reverse units of 2**N bits. */ static uint64_t reverse_bits_64(uint64_t x, int n) { int i, sh; x = bswap64(x); for (i = 2, sh = 4; i >= n; i--, sh >>= 1) { uint64_t mask = even_bit_esz_masks[i]; x = ((x & mask) << sh) | ((x >> sh) & mask); } return x; } static uint8_t reverse_bits_8(uint8_t x, int n) { static const uint8_t mask[3] = { 0x55, 0x33, 0x0f }; int i, sh; for (i = 2, sh = 4; i >= n; i--, sh >>= 1) { x = ((x & mask[i]) << sh) | ((x >> sh) & mask[i]); } return x; } void HELPER(sve_rev_p)(void *vd, void *vn, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); int esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); intptr_t i, oprsz_2 = oprsz / 2; if (oprsz <= 8) { uint64_t l = *(uint64_t *)vn; l = reverse_bits_64(l << (64 - 8 * oprsz), esz); *(uint64_t *)vd = l; } else if ((oprsz & 15) == 0) { for (i = 0; i < oprsz_2; i += 8) { intptr_t ih = oprsz - 8 - i; uint64_t l = reverse_bits_64(*(uint64_t *)(vn + i), esz); uint64_t h = reverse_bits_64(*(uint64_t *)(vn + ih), esz); *(uint64_t *)(vd + i) = h; *(uint64_t *)(vd + ih) = l; } } else { for (i = 0; i < oprsz_2; i += 1) { intptr_t il = H1(i); intptr_t ih = H1(oprsz - 1 - i); uint8_t l = reverse_bits_8(*(uint8_t *)(vn + il), esz); uint8_t h = reverse_bits_8(*(uint8_t *)(vn + ih), esz); *(uint8_t *)(vd + il) = h; *(uint8_t *)(vd + ih) = l; } } } void HELPER(sve_punpk_p)(void *vd, void *vn, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); intptr_t high = FIELD_EX32(pred_desc, PREDDESC, DATA); uint64_t *d = vd; intptr_t i; if (oprsz <= 8) { uint64_t nn = *(uint64_t *)vn; int half = 4 * oprsz; nn = extract64(nn, high * half, half); nn = expand_bits(nn, 0); d[0] = nn; } else { ARMPredicateReg tmp_n; /* We produce output faster than we consume input. Therefore we must be mindful of possible overlap. */ if ((vn - vd) < (uintptr_t)oprsz) { vn = memcpy(&tmp_n, vn, oprsz); } if (high) { high = oprsz >> 1; } if ((oprsz & 7) == 0) { uint32_t *n = vn; high >>= 2; for (i = 0; i < oprsz / 8; i++) { uint64_t nn = n[H4(high + i)]; d[i] = expand_bits(nn, 0); } } else { uint16_t *d16 = vd; uint8_t *n = vn; for (i = 0; i < oprsz / 2; i++) { uint16_t nn = n[H1(high + i)]; d16[H2(i)] = expand_bits(nn, 0); } } } } #define DO_ZIP(NAME, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t oprsz = simd_oprsz(desc); \ intptr_t i, oprsz_2 = oprsz / 2; \ ARMVectorReg tmp_n, tmp_m; \ /* We produce output faster than we consume input. \ Therefore we must be mindful of possible overlap. */ \ if (unlikely((vn - vd) < (uintptr_t)oprsz)) { \ vn = memcpy(&tmp_n, vn, oprsz_2); \ } \ if (unlikely((vm - vd) < (uintptr_t)oprsz)) { \ vm = memcpy(&tmp_m, vm, oprsz_2); \ } \ for (i = 0; i < oprsz_2; i += sizeof(TYPE)) { \ *(TYPE *)(vd + H(2 * i + 0)) = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(2 * i + sizeof(TYPE))) = *(TYPE *)(vm + H(i)); \ } \ if (sizeof(TYPE) == 16 && unlikely(oprsz & 16)) { \ memset(vd + oprsz - 16, 0, 16); \ } \ } DO_ZIP(sve_zip_b, uint8_t, H1) DO_ZIP(sve_zip_h, uint16_t, H1_2) DO_ZIP(sve_zip_s, uint32_t, H1_4) DO_ZIP(sve_zip_d, uint64_t, H1_8) DO_ZIP(sve2_zip_q, Int128, ) #define DO_UZP(NAME, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t oprsz = simd_oprsz(desc); \ intptr_t odd_ofs = simd_data(desc); \ intptr_t i, p; \ ARMVectorReg tmp_m; \ if (unlikely((vm - vd) < (uintptr_t)oprsz)) { \ vm = memcpy(&tmp_m, vm, oprsz); \ } \ i = 0, p = odd_ofs; \ do { \ *(TYPE *)(vd + H(i)) = *(TYPE *)(vn + H(p)); \ i += sizeof(TYPE), p += 2 * sizeof(TYPE); \ } while (p < oprsz); \ p -= oprsz; \ do { \ *(TYPE *)(vd + H(i)) = *(TYPE *)(vm + H(p)); \ i += sizeof(TYPE), p += 2 * sizeof(TYPE); \ } while (p < oprsz); \ tcg_debug_assert(i == oprsz); \ } DO_UZP(sve_uzp_b, uint8_t, H1) DO_UZP(sve_uzp_h, uint16_t, H1_2) DO_UZP(sve_uzp_s, uint32_t, H1_4) DO_UZP(sve_uzp_d, uint64_t, H1_8) DO_UZP(sve2_uzp_q, Int128, ) #define DO_TRN(NAME, TYPE, H) \ void HELPER(NAME)(void *vd, void *vn, void *vm, uint32_t desc) \ { \ intptr_t oprsz = simd_oprsz(desc); \ intptr_t odd_ofs = simd_data(desc); \ intptr_t i; \ for (i = 0; i < oprsz; i += 2 * sizeof(TYPE)) { \ TYPE ae = *(TYPE *)(vn + H(i + odd_ofs)); \ TYPE be = *(TYPE *)(vm + H(i + odd_ofs)); \ *(TYPE *)(vd + H(i + 0)) = ae; \ *(TYPE *)(vd + H(i + sizeof(TYPE))) = be; \ } \ if (sizeof(TYPE) == 16 && unlikely(oprsz & 16)) { \ memset(vd + oprsz - 16, 0, 16); \ } \ } DO_TRN(sve_trn_b, uint8_t, H1) DO_TRN(sve_trn_h, uint16_t, H1_2) DO_TRN(sve_trn_s, uint32_t, H1_4) DO_TRN(sve_trn_d, uint64_t, H1_8) DO_TRN(sve2_trn_q, Int128, ) #undef DO_ZIP #undef DO_UZP #undef DO_TRN void HELPER(sve_compact_s)(void *vd, void *vn, void *vg, uint32_t desc) { intptr_t i, j, opr_sz = simd_oprsz(desc) / 4; uint32_t *d = vd, *n = vn; uint8_t *pg = vg; for (i = j = 0; i < opr_sz; i++) { if (pg[H1(i / 2)] & (i & 1 ? 0x10 : 0x01)) { d[H4(j)] = n[H4(i)]; j++; } } for (; j < opr_sz; j++) { d[H4(j)] = 0; } } void HELPER(sve_compact_d)(void *vd, void *vn, void *vg, uint32_t desc) { intptr_t i, j, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn; uint8_t *pg = vg; for (i = j = 0; i < opr_sz; i++) { if (pg[H1(i)] & 1) { d[j] = n[i]; j++; } } for (; j < opr_sz; j++) { d[j] = 0; } } /* Similar to the ARM LastActiveElement pseudocode function, except the * result is multiplied by the element size. This includes the not found * indication; e.g. not found for esz=3 is -8. */ int32_t HELPER(sve_last_active_element)(void *vg, uint32_t pred_desc) { intptr_t words = DIV_ROUND_UP(FIELD_EX32(pred_desc, PREDDESC, OPRSZ), 8); intptr_t esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); return last_active_element(vg, words, esz); } void HELPER(sve_splice)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { intptr_t opr_sz = simd_oprsz(desc) / 8; int esz = simd_data(desc); uint64_t pg, first_g, last_g, len, mask = pred_esz_masks[esz]; intptr_t i, first_i, last_i; ARMVectorReg tmp; first_i = last_i = 0; first_g = last_g = 0; /* Find the extent of the active elements within VG. */ for (i = QEMU_ALIGN_UP(opr_sz, 8) - 8; i >= 0; i -= 8) { pg = *(uint64_t *)(vg + i) & mask; if (pg) { if (last_g == 0) { last_g = pg; last_i = i; } first_g = pg; first_i = i; } } len = 0; if (first_g != 0) { first_i = first_i * 8 + ctz64(first_g); last_i = last_i * 8 + 63 - clz64(last_g); len = last_i - first_i + (1 << esz); if (vd == vm) { vm = memcpy(&tmp, vm, opr_sz * 8); } swap_memmove(vd, vn + first_i, len); } swap_memmove(vd + len, vm, opr_sz * 8 - len); } void HELPER(sve_sel_zpzz_b)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i], mm = m[i]; uint64_t pp = expand_pred_b(pg[H1(i)]); d[i] = (nn & pp) | (mm & ~pp); } } void HELPER(sve_sel_zpzz_h)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i], mm = m[i]; uint64_t pp = expand_pred_h(pg[H1(i)]); d[i] = (nn & pp) | (mm & ~pp); } } void HELPER(sve_sel_zpzz_s)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i], mm = m[i]; uint64_t pp = expand_pred_s(pg[H1(i)]); d[i] = (nn & pp) | (mm & ~pp); } } void HELPER(sve_sel_zpzz_d)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm; uint8_t *pg = vg; for (i = 0; i < opr_sz; i += 1) { uint64_t nn = n[i], mm = m[i]; d[i] = (pg[H1(i)] & 1 ? nn : mm); } } /* Two operand comparison controlled by a predicate. * ??? It is very tempting to want to be able to expand this inline * with x86 instructions, e.g. * * vcmpeqw zm, zn, %ymm0 * vpmovmskb %ymm0, %eax * and $0x5555, %eax * and pg, %eax * * or even aarch64, e.g. * * // mask = 4000 1000 0400 0100 0040 0010 0004 0001 * cmeq v0.8h, zn, zm * and v0.8h, v0.8h, mask * addv h0, v0.8h * and v0.8b, pg * * However, coming up with an abstraction that allows vector inputs and * a scalar output, and also handles the byte-ordering of sub-uint64_t * scalar outputs, is tricky. */ #define DO_CMP_PPZZ(NAME, TYPE, OP, H, MASK) \ uint32_t HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t opr_sz = simd_oprsz(desc); \ uint32_t flags = PREDTEST_INIT; \ intptr_t i = opr_sz; \ do { \ uint64_t out = 0, pg; \ do { \ i -= sizeof(TYPE), out <<= sizeof(TYPE); \ TYPE nn = *(TYPE *)(vn + H(i)); \ TYPE mm = *(TYPE *)(vm + H(i)); \ out |= nn OP mm; \ } while (i & 63); \ pg = *(uint64_t *)(vg + (i >> 3)) & MASK; \ out &= pg; \ *(uint64_t *)(vd + (i >> 3)) = out; \ flags = iter_predtest_bwd(out, pg, flags); \ } while (i > 0); \ return flags; \ } #define DO_CMP_PPZZ_B(NAME, TYPE, OP) \ DO_CMP_PPZZ(NAME, TYPE, OP, H1, 0xffffffffffffffffull) #define DO_CMP_PPZZ_H(NAME, TYPE, OP) \ DO_CMP_PPZZ(NAME, TYPE, OP, H1_2, 0x5555555555555555ull) #define DO_CMP_PPZZ_S(NAME, TYPE, OP) \ DO_CMP_PPZZ(NAME, TYPE, OP, H1_4, 0x1111111111111111ull) #define DO_CMP_PPZZ_D(NAME, TYPE, OP) \ DO_CMP_PPZZ(NAME, TYPE, OP, H1_8, 0x0101010101010101ull) DO_CMP_PPZZ_B(sve_cmpeq_ppzz_b, uint8_t, ==) DO_CMP_PPZZ_H(sve_cmpeq_ppzz_h, uint16_t, ==) DO_CMP_PPZZ_S(sve_cmpeq_ppzz_s, uint32_t, ==) DO_CMP_PPZZ_D(sve_cmpeq_ppzz_d, uint64_t, ==) DO_CMP_PPZZ_B(sve_cmpne_ppzz_b, uint8_t, !=) DO_CMP_PPZZ_H(sve_cmpne_ppzz_h, uint16_t, !=) DO_CMP_PPZZ_S(sve_cmpne_ppzz_s, uint32_t, !=) DO_CMP_PPZZ_D(sve_cmpne_ppzz_d, uint64_t, !=) DO_CMP_PPZZ_B(sve_cmpgt_ppzz_b, int8_t, >) DO_CMP_PPZZ_H(sve_cmpgt_ppzz_h, int16_t, >) DO_CMP_PPZZ_S(sve_cmpgt_ppzz_s, int32_t, >) DO_CMP_PPZZ_D(sve_cmpgt_ppzz_d, int64_t, >) DO_CMP_PPZZ_B(sve_cmpge_ppzz_b, int8_t, >=) DO_CMP_PPZZ_H(sve_cmpge_ppzz_h, int16_t, >=) DO_CMP_PPZZ_S(sve_cmpge_ppzz_s, int32_t, >=) DO_CMP_PPZZ_D(sve_cmpge_ppzz_d, int64_t, >=) DO_CMP_PPZZ_B(sve_cmphi_ppzz_b, uint8_t, >) DO_CMP_PPZZ_H(sve_cmphi_ppzz_h, uint16_t, >) DO_CMP_PPZZ_S(sve_cmphi_ppzz_s, uint32_t, >) DO_CMP_PPZZ_D(sve_cmphi_ppzz_d, uint64_t, >) DO_CMP_PPZZ_B(sve_cmphs_ppzz_b, uint8_t, >=) DO_CMP_PPZZ_H(sve_cmphs_ppzz_h, uint16_t, >=) DO_CMP_PPZZ_S(sve_cmphs_ppzz_s, uint32_t, >=) DO_CMP_PPZZ_D(sve_cmphs_ppzz_d, uint64_t, >=) #undef DO_CMP_PPZZ_B #undef DO_CMP_PPZZ_H #undef DO_CMP_PPZZ_S #undef DO_CMP_PPZZ_D #undef DO_CMP_PPZZ /* Similar, but the second source is "wide". */ #define DO_CMP_PPZW(NAME, TYPE, TYPEW, OP, H, MASK) \ uint32_t HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ intptr_t opr_sz = simd_oprsz(desc); \ uint32_t flags = PREDTEST_INIT; \ intptr_t i = opr_sz; \ do { \ uint64_t out = 0, pg; \ do { \ TYPEW mm = *(TYPEW *)(vm + i - 8); \ do { \ i -= sizeof(TYPE), out <<= sizeof(TYPE); \ TYPE nn = *(TYPE *)(vn + H(i)); \ out |= nn OP mm; \ } while (i & 7); \ } while (i & 63); \ pg = *(uint64_t *)(vg + (i >> 3)) & MASK; \ out &= pg; \ *(uint64_t *)(vd + (i >> 3)) = out; \ flags = iter_predtest_bwd(out, pg, flags); \ } while (i > 0); \ return flags; \ } #define DO_CMP_PPZW_B(NAME, TYPE, TYPEW, OP) \ DO_CMP_PPZW(NAME, TYPE, TYPEW, OP, H1, 0xffffffffffffffffull) #define DO_CMP_PPZW_H(NAME, TYPE, TYPEW, OP) \ DO_CMP_PPZW(NAME, TYPE, TYPEW, OP, H1_2, 0x5555555555555555ull) #define DO_CMP_PPZW_S(NAME, TYPE, TYPEW, OP) \ DO_CMP_PPZW(NAME, TYPE, TYPEW, OP, H1_4, 0x1111111111111111ull) DO_CMP_PPZW_B(sve_cmpeq_ppzw_b, int8_t, uint64_t, ==) DO_CMP_PPZW_H(sve_cmpeq_ppzw_h, int16_t, uint64_t, ==) DO_CMP_PPZW_S(sve_cmpeq_ppzw_s, int32_t, uint64_t, ==) DO_CMP_PPZW_B(sve_cmpne_ppzw_b, int8_t, uint64_t, !=) DO_CMP_PPZW_H(sve_cmpne_ppzw_h, int16_t, uint64_t, !=) DO_CMP_PPZW_S(sve_cmpne_ppzw_s, int32_t, uint64_t, !=) DO_CMP_PPZW_B(sve_cmpgt_ppzw_b, int8_t, int64_t, >) DO_CMP_PPZW_H(sve_cmpgt_ppzw_h, int16_t, int64_t, >) DO_CMP_PPZW_S(sve_cmpgt_ppzw_s, int32_t, int64_t, >) DO_CMP_PPZW_B(sve_cmpge_ppzw_b, int8_t, int64_t, >=) DO_CMP_PPZW_H(sve_cmpge_ppzw_h, int16_t, int64_t, >=) DO_CMP_PPZW_S(sve_cmpge_ppzw_s, int32_t, int64_t, >=) DO_CMP_PPZW_B(sve_cmphi_ppzw_b, uint8_t, uint64_t, >) DO_CMP_PPZW_H(sve_cmphi_ppzw_h, uint16_t, uint64_t, >) DO_CMP_PPZW_S(sve_cmphi_ppzw_s, uint32_t, uint64_t, >) DO_CMP_PPZW_B(sve_cmphs_ppzw_b, uint8_t, uint64_t, >=) DO_CMP_PPZW_H(sve_cmphs_ppzw_h, uint16_t, uint64_t, >=) DO_CMP_PPZW_S(sve_cmphs_ppzw_s, uint32_t, uint64_t, >=) DO_CMP_PPZW_B(sve_cmplt_ppzw_b, int8_t, int64_t, <) DO_CMP_PPZW_H(sve_cmplt_ppzw_h, int16_t, int64_t, <) DO_CMP_PPZW_S(sve_cmplt_ppzw_s, int32_t, int64_t, <) DO_CMP_PPZW_B(sve_cmple_ppzw_b, int8_t, int64_t, <=) DO_CMP_PPZW_H(sve_cmple_ppzw_h, int16_t, int64_t, <=) DO_CMP_PPZW_S(sve_cmple_ppzw_s, int32_t, int64_t, <=) DO_CMP_PPZW_B(sve_cmplo_ppzw_b, uint8_t, uint64_t, <) DO_CMP_PPZW_H(sve_cmplo_ppzw_h, uint16_t, uint64_t, <) DO_CMP_PPZW_S(sve_cmplo_ppzw_s, uint32_t, uint64_t, <) DO_CMP_PPZW_B(sve_cmpls_ppzw_b, uint8_t, uint64_t, <=) DO_CMP_PPZW_H(sve_cmpls_ppzw_h, uint16_t, uint64_t, <=) DO_CMP_PPZW_S(sve_cmpls_ppzw_s, uint32_t, uint64_t, <=) #undef DO_CMP_PPZW_B #undef DO_CMP_PPZW_H #undef DO_CMP_PPZW_S #undef DO_CMP_PPZW /* Similar, but the second source is immediate. */ #define DO_CMP_PPZI(NAME, TYPE, OP, H, MASK) \ uint32_t HELPER(NAME)(void *vd, void *vn, void *vg, uint32_t desc) \ { \ intptr_t opr_sz = simd_oprsz(desc); \ uint32_t flags = PREDTEST_INIT; \ TYPE mm = simd_data(desc); \ intptr_t i = opr_sz; \ do { \ uint64_t out = 0, pg; \ do { \ i -= sizeof(TYPE), out <<= sizeof(TYPE); \ TYPE nn = *(TYPE *)(vn + H(i)); \ out |= nn OP mm; \ } while (i & 63); \ pg = *(uint64_t *)(vg + (i >> 3)) & MASK; \ out &= pg; \ *(uint64_t *)(vd + (i >> 3)) = out; \ flags = iter_predtest_bwd(out, pg, flags); \ } while (i > 0); \ return flags; \ } #define DO_CMP_PPZI_B(NAME, TYPE, OP) \ DO_CMP_PPZI(NAME, TYPE, OP, H1, 0xffffffffffffffffull) #define DO_CMP_PPZI_H(NAME, TYPE, OP) \ DO_CMP_PPZI(NAME, TYPE, OP, H1_2, 0x5555555555555555ull) #define DO_CMP_PPZI_S(NAME, TYPE, OP) \ DO_CMP_PPZI(NAME, TYPE, OP, H1_4, 0x1111111111111111ull) #define DO_CMP_PPZI_D(NAME, TYPE, OP) \ DO_CMP_PPZI(NAME, TYPE, OP, H1_8, 0x0101010101010101ull) DO_CMP_PPZI_B(sve_cmpeq_ppzi_b, uint8_t, ==) DO_CMP_PPZI_H(sve_cmpeq_ppzi_h, uint16_t, ==) DO_CMP_PPZI_S(sve_cmpeq_ppzi_s, uint32_t, ==) DO_CMP_PPZI_D(sve_cmpeq_ppzi_d, uint64_t, ==) DO_CMP_PPZI_B(sve_cmpne_ppzi_b, uint8_t, !=) DO_CMP_PPZI_H(sve_cmpne_ppzi_h, uint16_t, !=) DO_CMP_PPZI_S(sve_cmpne_ppzi_s, uint32_t, !=) DO_CMP_PPZI_D(sve_cmpne_ppzi_d, uint64_t, !=) DO_CMP_PPZI_B(sve_cmpgt_ppzi_b, int8_t, >) DO_CMP_PPZI_H(sve_cmpgt_ppzi_h, int16_t, >) DO_CMP_PPZI_S(sve_cmpgt_ppzi_s, int32_t, >) DO_CMP_PPZI_D(sve_cmpgt_ppzi_d, int64_t, >) DO_CMP_PPZI_B(sve_cmpge_ppzi_b, int8_t, >=) DO_CMP_PPZI_H(sve_cmpge_ppzi_h, int16_t, >=) DO_CMP_PPZI_S(sve_cmpge_ppzi_s, int32_t, >=) DO_CMP_PPZI_D(sve_cmpge_ppzi_d, int64_t, >=) DO_CMP_PPZI_B(sve_cmphi_ppzi_b, uint8_t, >) DO_CMP_PPZI_H(sve_cmphi_ppzi_h, uint16_t, >) DO_CMP_PPZI_S(sve_cmphi_ppzi_s, uint32_t, >) DO_CMP_PPZI_D(sve_cmphi_ppzi_d, uint64_t, >) DO_CMP_PPZI_B(sve_cmphs_ppzi_b, uint8_t, >=) DO_CMP_PPZI_H(sve_cmphs_ppzi_h, uint16_t, >=) DO_CMP_PPZI_S(sve_cmphs_ppzi_s, uint32_t, >=) DO_CMP_PPZI_D(sve_cmphs_ppzi_d, uint64_t, >=) DO_CMP_PPZI_B(sve_cmplt_ppzi_b, int8_t, <) DO_CMP_PPZI_H(sve_cmplt_ppzi_h, int16_t, <) DO_CMP_PPZI_S(sve_cmplt_ppzi_s, int32_t, <) DO_CMP_PPZI_D(sve_cmplt_ppzi_d, int64_t, <) DO_CMP_PPZI_B(sve_cmple_ppzi_b, int8_t, <=) DO_CMP_PPZI_H(sve_cmple_ppzi_h, int16_t, <=) DO_CMP_PPZI_S(sve_cmple_ppzi_s, int32_t, <=) DO_CMP_PPZI_D(sve_cmple_ppzi_d, int64_t, <=) DO_CMP_PPZI_B(sve_cmplo_ppzi_b, uint8_t, <) DO_CMP_PPZI_H(sve_cmplo_ppzi_h, uint16_t, <) DO_CMP_PPZI_S(sve_cmplo_ppzi_s, uint32_t, <) DO_CMP_PPZI_D(sve_cmplo_ppzi_d, uint64_t, <) DO_CMP_PPZI_B(sve_cmpls_ppzi_b, uint8_t, <=) DO_CMP_PPZI_H(sve_cmpls_ppzi_h, uint16_t, <=) DO_CMP_PPZI_S(sve_cmpls_ppzi_s, uint32_t, <=) DO_CMP_PPZI_D(sve_cmpls_ppzi_d, uint64_t, <=) #undef DO_CMP_PPZI_B #undef DO_CMP_PPZI_H #undef DO_CMP_PPZI_S #undef DO_CMP_PPZI_D #undef DO_CMP_PPZI /* Similar to the ARM LastActive pseudocode function. */ static bool last_active_pred(void *vd, void *vg, intptr_t oprsz) { intptr_t i; for (i = QEMU_ALIGN_UP(oprsz, 8) - 8; i >= 0; i -= 8) { uint64_t pg = *(uint64_t *)(vg + i); if (pg) { return (pow2floor(pg) & *(uint64_t *)(vd + i)) != 0; } } return 0; } /* Compute a mask into RETB that is true for all G, up to and including * (if after) or excluding (if !after) the first G & N. * Return true if BRK found. */ static bool compute_brk(uint64_t *retb, uint64_t n, uint64_t g, bool brk, bool after) { uint64_t b; if (brk) { b = 0; } else if ((g & n) == 0) { /* For all G, no N are set; break not found. */ b = g; } else { /* Break somewhere in N. Locate it. */ b = g & n; /* guard true, pred true */ b = b & -b; /* first such */ if (after) { b = b | (b - 1); /* break after same */ } else { b = b - 1; /* break before same */ } brk = true; } *retb = b; return brk; } /* Compute a zeroing BRK. */ static void compute_brk_z(uint64_t *d, uint64_t *n, uint64_t *g, intptr_t oprsz, bool after) { bool brk = false; intptr_t i; for (i = 0; i < DIV_ROUND_UP(oprsz, 8); ++i) { uint64_t this_b, this_g = g[i]; brk = compute_brk(&this_b, n[i], this_g, brk, after); d[i] = this_b & this_g; } } /* Likewise, but also compute flags. */ static uint32_t compute_brks_z(uint64_t *d, uint64_t *n, uint64_t *g, intptr_t oprsz, bool after) { uint32_t flags = PREDTEST_INIT; bool brk = false; intptr_t i; for (i = 0; i < DIV_ROUND_UP(oprsz, 8); ++i) { uint64_t this_b, this_d, this_g = g[i]; brk = compute_brk(&this_b, n[i], this_g, brk, after); d[i] = this_d = this_b & this_g; flags = iter_predtest_fwd(this_d, this_g, flags); } return flags; } /* Compute a merging BRK. */ static void compute_brk_m(uint64_t *d, uint64_t *n, uint64_t *g, intptr_t oprsz, bool after) { bool brk = false; intptr_t i; for (i = 0; i < DIV_ROUND_UP(oprsz, 8); ++i) { uint64_t this_b, this_g = g[i]; brk = compute_brk(&this_b, n[i], this_g, brk, after); d[i] = (this_b & this_g) | (d[i] & ~this_g); } } /* Likewise, but also compute flags. */ static uint32_t compute_brks_m(uint64_t *d, uint64_t *n, uint64_t *g, intptr_t oprsz, bool after) { uint32_t flags = PREDTEST_INIT; bool brk = false; intptr_t i; for (i = 0; i < oprsz / 8; ++i) { uint64_t this_b, this_d = d[i], this_g = g[i]; brk = compute_brk(&this_b, n[i], this_g, brk, after); d[i] = this_d = (this_b & this_g) | (this_d & ~this_g); flags = iter_predtest_fwd(this_d, this_g, flags); } return flags; } static uint32_t do_zero(ARMPredicateReg *d, intptr_t oprsz) { /* It is quicker to zero the whole predicate than loop on OPRSZ. * The compiler should turn this into 4 64-bit integer stores. */ memset(d, 0, sizeof(ARMPredicateReg)); return PREDTEST_INIT; } void HELPER(sve_brkpa)(void *vd, void *vn, void *vm, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); if (last_active_pred(vn, vg, oprsz)) { compute_brk_z(vd, vm, vg, oprsz, true); } else { do_zero(vd, oprsz); } } uint32_t HELPER(sve_brkpas)(void *vd, void *vn, void *vm, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); if (last_active_pred(vn, vg, oprsz)) { return compute_brks_z(vd, vm, vg, oprsz, true); } else { return do_zero(vd, oprsz); } } void HELPER(sve_brkpb)(void *vd, void *vn, void *vm, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); if (last_active_pred(vn, vg, oprsz)) { compute_brk_z(vd, vm, vg, oprsz, false); } else { do_zero(vd, oprsz); } } uint32_t HELPER(sve_brkpbs)(void *vd, void *vn, void *vm, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); if (last_active_pred(vn, vg, oprsz)) { return compute_brks_z(vd, vm, vg, oprsz, false); } else { return do_zero(vd, oprsz); } } void HELPER(sve_brka_z)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); compute_brk_z(vd, vn, vg, oprsz, true); } uint32_t HELPER(sve_brkas_z)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); return compute_brks_z(vd, vn, vg, oprsz, true); } void HELPER(sve_brkb_z)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); compute_brk_z(vd, vn, vg, oprsz, false); } uint32_t HELPER(sve_brkbs_z)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); return compute_brks_z(vd, vn, vg, oprsz, false); } void HELPER(sve_brka_m)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); compute_brk_m(vd, vn, vg, oprsz, true); } uint32_t HELPER(sve_brkas_m)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); return compute_brks_m(vd, vn, vg, oprsz, true); } void HELPER(sve_brkb_m)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); compute_brk_m(vd, vn, vg, oprsz, false); } uint32_t HELPER(sve_brkbs_m)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); return compute_brks_m(vd, vn, vg, oprsz, false); } void HELPER(sve_brkn)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); if (!last_active_pred(vn, vg, oprsz)) { do_zero(vd, oprsz); } } /* As if PredTest(Ones(PL), D, esz). */ static uint32_t predtest_ones(ARMPredicateReg *d, intptr_t oprsz, uint64_t esz_mask) { uint32_t flags = PREDTEST_INIT; intptr_t i; for (i = 0; i < oprsz / 8; i++) { flags = iter_predtest_fwd(d->p[i], esz_mask, flags); } if (oprsz & 7) { uint64_t mask = ~(-1ULL << (8 * (oprsz & 7))); flags = iter_predtest_fwd(d->p[i], esz_mask & mask, flags); } return flags; } uint32_t HELPER(sve_brkns)(void *vd, void *vn, void *vg, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); if (last_active_pred(vn, vg, oprsz)) { return predtest_ones(vd, oprsz, -1); } else { return do_zero(vd, oprsz); } } uint64_t HELPER(sve_cntp)(void *vn, void *vg, uint32_t pred_desc) { intptr_t words = DIV_ROUND_UP(FIELD_EX32(pred_desc, PREDDESC, OPRSZ), 8); intptr_t esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); uint64_t *n = vn, *g = vg, sum = 0, mask = pred_esz_masks[esz]; intptr_t i; for (i = 0; i < words; ++i) { uint64_t t = n[i] & g[i] & mask; sum += ctpop64(t); } return sum; } uint32_t HELPER(sve_whilel)(void *vd, uint32_t count, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); intptr_t esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); uint64_t esz_mask = pred_esz_masks[esz]; ARMPredicateReg *d = vd; uint32_t flags; intptr_t i; /* Begin with a zero predicate register. */ flags = do_zero(d, oprsz); if (count == 0) { return flags; } /* Set all of the requested bits. */ for (i = 0; i < count / 64; ++i) { d->p[i] = esz_mask; } if (count & 63) { d->p[i] = MAKE_64BIT_MASK(0, count & 63) & esz_mask; } return predtest_ones(d, oprsz, esz_mask); } uint32_t HELPER(sve_whileg)(void *vd, uint32_t count, uint32_t pred_desc) { intptr_t oprsz = FIELD_EX32(pred_desc, PREDDESC, OPRSZ); intptr_t esz = FIELD_EX32(pred_desc, PREDDESC, ESZ); uint64_t esz_mask = pred_esz_masks[esz]; ARMPredicateReg *d = vd; intptr_t i, invcount, oprbits; uint64_t bits; if (count == 0) { return do_zero(d, oprsz); } oprbits = oprsz * 8; tcg_debug_assert(count <= oprbits); bits = esz_mask; if (oprbits & 63) { bits &= MAKE_64BIT_MASK(0, oprbits & 63); } invcount = oprbits - count; for (i = (oprsz - 1) / 8; i > invcount / 64; --i) { d->p[i] = bits; bits = esz_mask; } d->p[i] = bits & MAKE_64BIT_MASK(invcount & 63, 64); while (--i >= 0) { d->p[i] = 0; } return predtest_ones(d, oprsz, esz_mask); } /* Recursive reduction on a function; * C.f. the ARM ARM function ReducePredicated. * * While it would be possible to write this without the DATA temporary, * it is much simpler to process the predicate register this way. * The recursion is bounded to depth 7 (128 fp16 elements), so there's * little to gain with a more complex non-recursive form. */ #define DO_REDUCE(NAME, TYPE, H, FUNC, IDENT) \ static TYPE NAME##_reduce(TYPE *data, float_status *status, uintptr_t n) \ { \ if (n == 1) { \ return *data; \ } else { \ uintptr_t half = n / 2; \ TYPE lo = NAME##_reduce(data, status, half); \ TYPE hi = NAME##_reduce(data + half, status, half); \ return TYPE##_##FUNC(lo, hi, status); \ } \ } \ uint64_t HELPER(NAME)(void *vn, void *vg, void *vs, uint32_t desc) \ { \ uintptr_t i, oprsz = simd_oprsz(desc), maxsz = simd_data(desc); \ TYPE data[sizeof(ARMVectorReg) / sizeof(TYPE)]; \ for (i = 0; i < oprsz; ) { \ uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); \ do { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)((void *)data + i) = (pg & 1 ? nn : IDENT); \ i += sizeof(TYPE), pg >>= sizeof(TYPE); \ } while (i & 15); \ } \ for (; i < maxsz; i += sizeof(TYPE)) { \ *(TYPE *)((void *)data + i) = IDENT; \ } \ return NAME##_reduce(data, vs, maxsz / sizeof(TYPE)); \ } DO_REDUCE(sve_faddv_h, float16, H1_2, add, float16_zero) DO_REDUCE(sve_faddv_s, float32, H1_4, add, float32_zero) DO_REDUCE(sve_faddv_d, float64, H1_8, add, float64_zero) /* Identity is floatN_default_nan, without the function call. */ DO_REDUCE(sve_fminnmv_h, float16, H1_2, minnum, 0x7E00) DO_REDUCE(sve_fminnmv_s, float32, H1_4, minnum, 0x7FC00000) DO_REDUCE(sve_fminnmv_d, float64, H1_8, minnum, 0x7FF8000000000000ULL) DO_REDUCE(sve_fmaxnmv_h, float16, H1_2, maxnum, 0x7E00) DO_REDUCE(sve_fmaxnmv_s, float32, H1_4, maxnum, 0x7FC00000) DO_REDUCE(sve_fmaxnmv_d, float64, H1_8, maxnum, 0x7FF8000000000000ULL) DO_REDUCE(sve_fminv_h, float16, H1_2, min, float16_infinity) DO_REDUCE(sve_fminv_s, float32, H1_4, min, float32_infinity) DO_REDUCE(sve_fminv_d, float64, H1_8, min, float64_infinity) DO_REDUCE(sve_fmaxv_h, float16, H1_2, max, float16_chs(float16_infinity)) DO_REDUCE(sve_fmaxv_s, float32, H1_4, max, float32_chs(float32_infinity)) DO_REDUCE(sve_fmaxv_d, float64, H1_8, max, float64_chs(float64_infinity)) #undef DO_REDUCE uint64_t HELPER(sve_fadda_h)(uint64_t nn, void *vm, void *vg, void *status, uint32_t desc) { intptr_t i = 0, opr_sz = simd_oprsz(desc); float16 result = nn; do { uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); do { if (pg & 1) { float16 mm = *(float16 *)(vm + H1_2(i)); result = float16_add(result, mm, status); } i += sizeof(float16), pg >>= sizeof(float16); } while (i & 15); } while (i < opr_sz); return result; } uint64_t HELPER(sve_fadda_s)(uint64_t nn, void *vm, void *vg, void *status, uint32_t desc) { intptr_t i = 0, opr_sz = simd_oprsz(desc); float32 result = nn; do { uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)); do { if (pg & 1) { float32 mm = *(float32 *)(vm + H1_2(i)); result = float32_add(result, mm, status); } i += sizeof(float32), pg >>= sizeof(float32); } while (i & 15); } while (i < opr_sz); return result; } uint64_t HELPER(sve_fadda_d)(uint64_t nn, void *vm, void *vg, void *status, uint32_t desc) { intptr_t i = 0, opr_sz = simd_oprsz(desc) / 8; uint64_t *m = vm; uint8_t *pg = vg; for (i = 0; i < opr_sz; i++) { if (pg[H1(i)] & 1) { nn = float64_add(nn, m[i], status); } } return nn; } /* Fully general three-operand expander, controlled by a predicate, * With the extra float_status parameter. */ #define DO_ZPZZ_FP(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, \ void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc); \ uint64_t *g = vg; \ do { \ uint64_t pg = g[(i - 1) >> 6]; \ do { \ i -= sizeof(TYPE); \ if (likely((pg >> (i & 63)) & 1)) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ TYPE mm = *(TYPE *)(vm + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, mm, status); \ } \ } while (i & 63); \ } while (i != 0); \ } DO_ZPZZ_FP(sve_fadd_h, uint16_t, H1_2, float16_add) DO_ZPZZ_FP(sve_fadd_s, uint32_t, H1_4, float32_add) DO_ZPZZ_FP(sve_fadd_d, uint64_t, H1_8, float64_add) DO_ZPZZ_FP(sve_fsub_h, uint16_t, H1_2, float16_sub) DO_ZPZZ_FP(sve_fsub_s, uint32_t, H1_4, float32_sub) DO_ZPZZ_FP(sve_fsub_d, uint64_t, H1_8, float64_sub) DO_ZPZZ_FP(sve_fmul_h, uint16_t, H1_2, float16_mul) DO_ZPZZ_FP(sve_fmul_s, uint32_t, H1_4, float32_mul) DO_ZPZZ_FP(sve_fmul_d, uint64_t, H1_8, float64_mul) DO_ZPZZ_FP(sve_fdiv_h, uint16_t, H1_2, float16_div) DO_ZPZZ_FP(sve_fdiv_s, uint32_t, H1_4, float32_div) DO_ZPZZ_FP(sve_fdiv_d, uint64_t, H1_8, float64_div) DO_ZPZZ_FP(sve_fmin_h, uint16_t, H1_2, float16_min) DO_ZPZZ_FP(sve_fmin_s, uint32_t, H1_4, float32_min) DO_ZPZZ_FP(sve_fmin_d, uint64_t, H1_8, float64_min) DO_ZPZZ_FP(sve_fmax_h, uint16_t, H1_2, float16_max) DO_ZPZZ_FP(sve_fmax_s, uint32_t, H1_4, float32_max) DO_ZPZZ_FP(sve_fmax_d, uint64_t, H1_8, float64_max) DO_ZPZZ_FP(sve_fminnum_h, uint16_t, H1_2, float16_minnum) DO_ZPZZ_FP(sve_fminnum_s, uint32_t, H1_4, float32_minnum) DO_ZPZZ_FP(sve_fminnum_d, uint64_t, H1_8, float64_minnum) DO_ZPZZ_FP(sve_fmaxnum_h, uint16_t, H1_2, float16_maxnum) DO_ZPZZ_FP(sve_fmaxnum_s, uint32_t, H1_4, float32_maxnum) DO_ZPZZ_FP(sve_fmaxnum_d, uint64_t, H1_8, float64_maxnum) static inline float16 abd_h(float16 a, float16 b, float_status *s) { return float16_abs(float16_sub(a, b, s)); } static inline float32 abd_s(float32 a, float32 b, float_status *s) { return float32_abs(float32_sub(a, b, s)); } static inline float64 abd_d(float64 a, float64 b, float_status *s) { return float64_abs(float64_sub(a, b, s)); } DO_ZPZZ_FP(sve_fabd_h, uint16_t, H1_2, abd_h) DO_ZPZZ_FP(sve_fabd_s, uint32_t, H1_4, abd_s) DO_ZPZZ_FP(sve_fabd_d, uint64_t, H1_8, abd_d) static inline float64 scalbn_d(float64 a, int64_t b, float_status *s) { int b_int = MIN(MAX(b, INT_MIN), INT_MAX); return float64_scalbn(a, b_int, s); } DO_ZPZZ_FP(sve_fscalbn_h, int16_t, H1_2, float16_scalbn) DO_ZPZZ_FP(sve_fscalbn_s, int32_t, H1_4, float32_scalbn) DO_ZPZZ_FP(sve_fscalbn_d, int64_t, H1_8, scalbn_d) DO_ZPZZ_FP(sve_fmulx_h, uint16_t, H1_2, helper_advsimd_mulxh) DO_ZPZZ_FP(sve_fmulx_s, uint32_t, H1_4, helper_vfp_mulxs) DO_ZPZZ_FP(sve_fmulx_d, uint64_t, H1_8, helper_vfp_mulxd) #undef DO_ZPZZ_FP /* Three-operand expander, with one scalar operand, controlled by * a predicate, with the extra float_status parameter. */ #define DO_ZPZS_FP(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, uint64_t scalar, \ void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc); \ uint64_t *g = vg; \ TYPE mm = scalar; \ do { \ uint64_t pg = g[(i - 1) >> 6]; \ do { \ i -= sizeof(TYPE); \ if (likely((pg >> (i & 63)) & 1)) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, mm, status); \ } \ } while (i & 63); \ } while (i != 0); \ } DO_ZPZS_FP(sve_fadds_h, float16, H1_2, float16_add) DO_ZPZS_FP(sve_fadds_s, float32, H1_4, float32_add) DO_ZPZS_FP(sve_fadds_d, float64, H1_8, float64_add) DO_ZPZS_FP(sve_fsubs_h, float16, H1_2, float16_sub) DO_ZPZS_FP(sve_fsubs_s, float32, H1_4, float32_sub) DO_ZPZS_FP(sve_fsubs_d, float64, H1_8, float64_sub) DO_ZPZS_FP(sve_fmuls_h, float16, H1_2, float16_mul) DO_ZPZS_FP(sve_fmuls_s, float32, H1_4, float32_mul) DO_ZPZS_FP(sve_fmuls_d, float64, H1_8, float64_mul) static inline float16 subr_h(float16 a, float16 b, float_status *s) { return float16_sub(b, a, s); } static inline float32 subr_s(float32 a, float32 b, float_status *s) { return float32_sub(b, a, s); } static inline float64 subr_d(float64 a, float64 b, float_status *s) { return float64_sub(b, a, s); } DO_ZPZS_FP(sve_fsubrs_h, float16, H1_2, subr_h) DO_ZPZS_FP(sve_fsubrs_s, float32, H1_4, subr_s) DO_ZPZS_FP(sve_fsubrs_d, float64, H1_8, subr_d) DO_ZPZS_FP(sve_fmaxnms_h, float16, H1_2, float16_maxnum) DO_ZPZS_FP(sve_fmaxnms_s, float32, H1_4, float32_maxnum) DO_ZPZS_FP(sve_fmaxnms_d, float64, H1_8, float64_maxnum) DO_ZPZS_FP(sve_fminnms_h, float16, H1_2, float16_minnum) DO_ZPZS_FP(sve_fminnms_s, float32, H1_4, float32_minnum) DO_ZPZS_FP(sve_fminnms_d, float64, H1_8, float64_minnum) DO_ZPZS_FP(sve_fmaxs_h, float16, H1_2, float16_max) DO_ZPZS_FP(sve_fmaxs_s, float32, H1_4, float32_max) DO_ZPZS_FP(sve_fmaxs_d, float64, H1_8, float64_max) DO_ZPZS_FP(sve_fmins_h, float16, H1_2, float16_min) DO_ZPZS_FP(sve_fmins_s, float32, H1_4, float32_min) DO_ZPZS_FP(sve_fmins_d, float64, H1_8, float64_min) /* Fully general two-operand expander, controlled by a predicate, * With the extra float_status parameter. */ #define DO_ZPZ_FP(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc); \ uint64_t *g = vg; \ do { \ uint64_t pg = g[(i - 1) >> 6]; \ do { \ i -= sizeof(TYPE); \ if (likely((pg >> (i & 63)) & 1)) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ *(TYPE *)(vd + H(i)) = OP(nn, status); \ } \ } while (i & 63); \ } while (i != 0); \ } /* SVE fp16 conversions always use IEEE mode. Like AdvSIMD, they ignore * FZ16. When converting from fp16, this affects flushing input denormals; * when converting to fp16, this affects flushing output denormals. */ static inline float32 sve_f16_to_f32(float16 f, float_status *fpst) { bool save = get_flush_inputs_to_zero(fpst); float32 ret; set_flush_inputs_to_zero(false, fpst); ret = float16_to_float32(f, true, fpst); set_flush_inputs_to_zero(save, fpst); return ret; } static inline float64 sve_f16_to_f64(float16 f, float_status *fpst) { bool save = get_flush_inputs_to_zero(fpst); float64 ret; set_flush_inputs_to_zero(false, fpst); ret = float16_to_float64(f, true, fpst); set_flush_inputs_to_zero(save, fpst); return ret; } static inline float16 sve_f32_to_f16(float32 f, float_status *fpst) { bool save = get_flush_to_zero(fpst); float16 ret; set_flush_to_zero(false, fpst); ret = float32_to_float16(f, true, fpst); set_flush_to_zero(save, fpst); return ret; } static inline float16 sve_f64_to_f16(float64 f, float_status *fpst) { bool save = get_flush_to_zero(fpst); float16 ret; set_flush_to_zero(false, fpst); ret = float64_to_float16(f, true, fpst); set_flush_to_zero(save, fpst); return ret; } static inline int16_t vfp_float16_to_int16_rtz(float16 f, float_status *s) { if (float16_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float16_to_int16_round_to_zero(f, s); } static inline int64_t vfp_float16_to_int64_rtz(float16 f, float_status *s) { if (float16_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float16_to_int64_round_to_zero(f, s); } static inline int64_t vfp_float32_to_int64_rtz(float32 f, float_status *s) { if (float32_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float32_to_int64_round_to_zero(f, s); } static inline int64_t vfp_float64_to_int64_rtz(float64 f, float_status *s) { if (float64_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float64_to_int64_round_to_zero(f, s); } static inline uint16_t vfp_float16_to_uint16_rtz(float16 f, float_status *s) { if (float16_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float16_to_uint16_round_to_zero(f, s); } static inline uint64_t vfp_float16_to_uint64_rtz(float16 f, float_status *s) { if (float16_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float16_to_uint64_round_to_zero(f, s); } static inline uint64_t vfp_float32_to_uint64_rtz(float32 f, float_status *s) { if (float32_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float32_to_uint64_round_to_zero(f, s); } static inline uint64_t vfp_float64_to_uint64_rtz(float64 f, float_status *s) { if (float64_is_any_nan(f)) { float_raise(float_flag_invalid, s); return 0; } return float64_to_uint64_round_to_zero(f, s); } DO_ZPZ_FP(sve_fcvt_sh, uint32_t, H1_4, sve_f32_to_f16) DO_ZPZ_FP(sve_fcvt_hs, uint32_t, H1_4, sve_f16_to_f32) DO_ZPZ_FP(sve_bfcvt, uint32_t, H1_4, float32_to_bfloat16) DO_ZPZ_FP(sve_fcvt_dh, uint64_t, H1_8, sve_f64_to_f16) DO_ZPZ_FP(sve_fcvt_hd, uint64_t, H1_8, sve_f16_to_f64) DO_ZPZ_FP(sve_fcvt_ds, uint64_t, H1_8, float64_to_float32) DO_ZPZ_FP(sve_fcvt_sd, uint64_t, H1_8, float32_to_float64) DO_ZPZ_FP(sve_fcvtzs_hh, uint16_t, H1_2, vfp_float16_to_int16_rtz) DO_ZPZ_FP(sve_fcvtzs_hs, uint32_t, H1_4, helper_vfp_tosizh) DO_ZPZ_FP(sve_fcvtzs_ss, uint32_t, H1_4, helper_vfp_tosizs) DO_ZPZ_FP(sve_fcvtzs_hd, uint64_t, H1_8, vfp_float16_to_int64_rtz) DO_ZPZ_FP(sve_fcvtzs_sd, uint64_t, H1_8, vfp_float32_to_int64_rtz) DO_ZPZ_FP(sve_fcvtzs_ds, uint64_t, H1_8, helper_vfp_tosizd) DO_ZPZ_FP(sve_fcvtzs_dd, uint64_t, H1_8, vfp_float64_to_int64_rtz) DO_ZPZ_FP(sve_fcvtzu_hh, uint16_t, H1_2, vfp_float16_to_uint16_rtz) DO_ZPZ_FP(sve_fcvtzu_hs, uint32_t, H1_4, helper_vfp_touizh) DO_ZPZ_FP(sve_fcvtzu_ss, uint32_t, H1_4, helper_vfp_touizs) DO_ZPZ_FP(sve_fcvtzu_hd, uint64_t, H1_8, vfp_float16_to_uint64_rtz) DO_ZPZ_FP(sve_fcvtzu_sd, uint64_t, H1_8, vfp_float32_to_uint64_rtz) DO_ZPZ_FP(sve_fcvtzu_ds, uint64_t, H1_8, helper_vfp_touizd) DO_ZPZ_FP(sve_fcvtzu_dd, uint64_t, H1_8, vfp_float64_to_uint64_rtz) DO_ZPZ_FP(sve_frint_h, uint16_t, H1_2, helper_advsimd_rinth) DO_ZPZ_FP(sve_frint_s, uint32_t, H1_4, helper_rints) DO_ZPZ_FP(sve_frint_d, uint64_t, H1_8, helper_rintd) DO_ZPZ_FP(sve_frintx_h, uint16_t, H1_2, float16_round_to_int) DO_ZPZ_FP(sve_frintx_s, uint32_t, H1_4, float32_round_to_int) DO_ZPZ_FP(sve_frintx_d, uint64_t, H1_8, float64_round_to_int) DO_ZPZ_FP(sve_frecpx_h, uint16_t, H1_2, helper_frecpx_f16) DO_ZPZ_FP(sve_frecpx_s, uint32_t, H1_4, helper_frecpx_f32) DO_ZPZ_FP(sve_frecpx_d, uint64_t, H1_8, helper_frecpx_f64) DO_ZPZ_FP(sve_fsqrt_h, uint16_t, H1_2, float16_sqrt) DO_ZPZ_FP(sve_fsqrt_s, uint32_t, H1_4, float32_sqrt) DO_ZPZ_FP(sve_fsqrt_d, uint64_t, H1_8, float64_sqrt) DO_ZPZ_FP(sve_scvt_hh, uint16_t, H1_2, int16_to_float16) DO_ZPZ_FP(sve_scvt_sh, uint32_t, H1_4, int32_to_float16) DO_ZPZ_FP(sve_scvt_ss, uint32_t, H1_4, int32_to_float32) DO_ZPZ_FP(sve_scvt_sd, uint64_t, H1_8, int32_to_float64) DO_ZPZ_FP(sve_scvt_dh, uint64_t, H1_8, int64_to_float16) DO_ZPZ_FP(sve_scvt_ds, uint64_t, H1_8, int64_to_float32) DO_ZPZ_FP(sve_scvt_dd, uint64_t, H1_8, int64_to_float64) DO_ZPZ_FP(sve_ucvt_hh, uint16_t, H1_2, uint16_to_float16) DO_ZPZ_FP(sve_ucvt_sh, uint32_t, H1_4, uint32_to_float16) DO_ZPZ_FP(sve_ucvt_ss, uint32_t, H1_4, uint32_to_float32) DO_ZPZ_FP(sve_ucvt_sd, uint64_t, H1_8, uint32_to_float64) DO_ZPZ_FP(sve_ucvt_dh, uint64_t, H1_8, uint64_to_float16) DO_ZPZ_FP(sve_ucvt_ds, uint64_t, H1_8, uint64_to_float32) DO_ZPZ_FP(sve_ucvt_dd, uint64_t, H1_8, uint64_to_float64) static int16_t do_float16_logb_as_int(float16 a, float_status *s) { /* Extract frac to the top of the uint32_t. */ uint32_t frac = (uint32_t)a << (16 + 6); int16_t exp = extract32(a, 10, 5); if (unlikely(exp == 0)) { if (frac != 0) { if (!get_flush_inputs_to_zero(s)) { /* denormal: bias - fractional_zeros */ return -15 - clz32(frac); } /* flush to zero */ float_raise(float_flag_input_denormal, s); } } else if (unlikely(exp == 0x1f)) { if (frac == 0) { return INT16_MAX; /* infinity */ } } else { /* normal: exp - bias */ return exp - 15; } /* nan or zero */ float_raise(float_flag_invalid, s); return INT16_MIN; } static int32_t do_float32_logb_as_int(float32 a, float_status *s) { /* Extract frac to the top of the uint32_t. */ uint32_t frac = a << 9; int32_t exp = extract32(a, 23, 8); if (unlikely(exp == 0)) { if (frac != 0) { if (!get_flush_inputs_to_zero(s)) { /* denormal: bias - fractional_zeros */ return -127 - clz32(frac); } /* flush to zero */ float_raise(float_flag_input_denormal, s); } } else if (unlikely(exp == 0xff)) { if (frac == 0) { return INT32_MAX; /* infinity */ } } else { /* normal: exp - bias */ return exp - 127; } /* nan or zero */ float_raise(float_flag_invalid, s); return INT32_MIN; } static int64_t do_float64_logb_as_int(float64 a, float_status *s) { /* Extract frac to the top of the uint64_t. */ uint64_t frac = a << 12; int64_t exp = extract64(a, 52, 11); if (unlikely(exp == 0)) { if (frac != 0) { if (!get_flush_inputs_to_zero(s)) { /* denormal: bias - fractional_zeros */ return -1023 - clz64(frac); } /* flush to zero */ float_raise(float_flag_input_denormal, s); } } else if (unlikely(exp == 0x7ff)) { if (frac == 0) { return INT64_MAX; /* infinity */ } } else { /* normal: exp - bias */ return exp - 1023; } /* nan or zero */ float_raise(float_flag_invalid, s); return INT64_MIN; } DO_ZPZ_FP(flogb_h, float16, H1_2, do_float16_logb_as_int) DO_ZPZ_FP(flogb_s, float32, H1_4, do_float32_logb_as_int) DO_ZPZ_FP(flogb_d, float64, H1_8, do_float64_logb_as_int) #undef DO_ZPZ_FP static void do_fmla_zpzzz_h(void *vd, void *vn, void *vm, void *va, void *vg, float_status *status, uint32_t desc, uint16_t neg1, uint16_t neg3) { intptr_t i = simd_oprsz(desc); uint64_t *g = vg; do { uint64_t pg = g[(i - 1) >> 6]; do { i -= 2; if (likely((pg >> (i & 63)) & 1)) { float16 e1, e2, e3, r; e1 = *(uint16_t *)(vn + H1_2(i)) ^ neg1; e2 = *(uint16_t *)(vm + H1_2(i)); e3 = *(uint16_t *)(va + H1_2(i)) ^ neg3; r = float16_muladd(e1, e2, e3, 0, status); *(uint16_t *)(vd + H1_2(i)) = r; } } while (i & 63); } while (i != 0); } void HELPER(sve_fmla_zpzzz_h)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_h(vd, vn, vm, va, vg, status, desc, 0, 0); } void HELPER(sve_fmls_zpzzz_h)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_h(vd, vn, vm, va, vg, status, desc, 0x8000, 0); } void HELPER(sve_fnmla_zpzzz_h)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_h(vd, vn, vm, va, vg, status, desc, 0x8000, 0x8000); } void HELPER(sve_fnmls_zpzzz_h)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_h(vd, vn, vm, va, vg, status, desc, 0, 0x8000); } static void do_fmla_zpzzz_s(void *vd, void *vn, void *vm, void *va, void *vg, float_status *status, uint32_t desc, uint32_t neg1, uint32_t neg3) { intptr_t i = simd_oprsz(desc); uint64_t *g = vg; do { uint64_t pg = g[(i - 1) >> 6]; do { i -= 4; if (likely((pg >> (i & 63)) & 1)) { float32 e1, e2, e3, r; e1 = *(uint32_t *)(vn + H1_4(i)) ^ neg1; e2 = *(uint32_t *)(vm + H1_4(i)); e3 = *(uint32_t *)(va + H1_4(i)) ^ neg3; r = float32_muladd(e1, e2, e3, 0, status); *(uint32_t *)(vd + H1_4(i)) = r; } } while (i & 63); } while (i != 0); } void HELPER(sve_fmla_zpzzz_s)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_s(vd, vn, vm, va, vg, status, desc, 0, 0); } void HELPER(sve_fmls_zpzzz_s)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_s(vd, vn, vm, va, vg, status, desc, 0x80000000, 0); } void HELPER(sve_fnmla_zpzzz_s)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_s(vd, vn, vm, va, vg, status, desc, 0x80000000, 0x80000000); } void HELPER(sve_fnmls_zpzzz_s)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_s(vd, vn, vm, va, vg, status, desc, 0, 0x80000000); } static void do_fmla_zpzzz_d(void *vd, void *vn, void *vm, void *va, void *vg, float_status *status, uint32_t desc, uint64_t neg1, uint64_t neg3) { intptr_t i = simd_oprsz(desc); uint64_t *g = vg; do { uint64_t pg = g[(i - 1) >> 6]; do { i -= 8; if (likely((pg >> (i & 63)) & 1)) { float64 e1, e2, e3, r; e1 = *(uint64_t *)(vn + i) ^ neg1; e2 = *(uint64_t *)(vm + i); e3 = *(uint64_t *)(va + i) ^ neg3; r = float64_muladd(e1, e2, e3, 0, status); *(uint64_t *)(vd + i) = r; } } while (i & 63); } while (i != 0); } void HELPER(sve_fmla_zpzzz_d)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_d(vd, vn, vm, va, vg, status, desc, 0, 0); } void HELPER(sve_fmls_zpzzz_d)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_d(vd, vn, vm, va, vg, status, desc, INT64_MIN, 0); } void HELPER(sve_fnmla_zpzzz_d)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_d(vd, vn, vm, va, vg, status, desc, INT64_MIN, INT64_MIN); } void HELPER(sve_fnmls_zpzzz_d)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { do_fmla_zpzzz_d(vd, vn, vm, va, vg, status, desc, 0, INT64_MIN); } /* Two operand floating-point comparison controlled by a predicate. * Unlike the integer version, we are not allowed to optimistically * compare operands, since the comparison may have side effects wrt * the FPSR. */ #define DO_FPCMP_PPZZ(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, \ void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc), j = (i - 1) >> 6; \ uint64_t *d = vd, *g = vg; \ do { \ uint64_t out = 0, pg = g[j]; \ do { \ i -= sizeof(TYPE), out <<= sizeof(TYPE); \ if (likely((pg >> (i & 63)) & 1)) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ TYPE mm = *(TYPE *)(vm + H(i)); \ out |= OP(TYPE, nn, mm, status); \ } \ } while (i & 63); \ d[j--] = out; \ } while (i > 0); \ } #define DO_FPCMP_PPZZ_H(NAME, OP) \ DO_FPCMP_PPZZ(NAME##_h, float16, H1_2, OP) #define DO_FPCMP_PPZZ_S(NAME, OP) \ DO_FPCMP_PPZZ(NAME##_s, float32, H1_4, OP) #define DO_FPCMP_PPZZ_D(NAME, OP) \ DO_FPCMP_PPZZ(NAME##_d, float64, H1_8, OP) #define DO_FPCMP_PPZZ_ALL(NAME, OP) \ DO_FPCMP_PPZZ_H(NAME, OP) \ DO_FPCMP_PPZZ_S(NAME, OP) \ DO_FPCMP_PPZZ_D(NAME, OP) #define DO_FCMGE(TYPE, X, Y, ST) TYPE##_compare(Y, X, ST) <= 0 #define DO_FCMGT(TYPE, X, Y, ST) TYPE##_compare(Y, X, ST) < 0 #define DO_FCMLE(TYPE, X, Y, ST) TYPE##_compare(X, Y, ST) <= 0 #define DO_FCMLT(TYPE, X, Y, ST) TYPE##_compare(X, Y, ST) < 0 #define DO_FCMEQ(TYPE, X, Y, ST) TYPE##_compare_quiet(X, Y, ST) == 0 #define DO_FCMNE(TYPE, X, Y, ST) TYPE##_compare_quiet(X, Y, ST) != 0 #define DO_FCMUO(TYPE, X, Y, ST) \ TYPE##_compare_quiet(X, Y, ST) == float_relation_unordered #define DO_FACGE(TYPE, X, Y, ST) \ TYPE##_compare(TYPE##_abs(Y), TYPE##_abs(X), ST) <= 0 #define DO_FACGT(TYPE, X, Y, ST) \ TYPE##_compare(TYPE##_abs(Y), TYPE##_abs(X), ST) < 0 DO_FPCMP_PPZZ_ALL(sve_fcmge, DO_FCMGE) DO_FPCMP_PPZZ_ALL(sve_fcmgt, DO_FCMGT) DO_FPCMP_PPZZ_ALL(sve_fcmeq, DO_FCMEQ) DO_FPCMP_PPZZ_ALL(sve_fcmne, DO_FCMNE) DO_FPCMP_PPZZ_ALL(sve_fcmuo, DO_FCMUO) DO_FPCMP_PPZZ_ALL(sve_facge, DO_FACGE) DO_FPCMP_PPZZ_ALL(sve_facgt, DO_FACGT) #undef DO_FPCMP_PPZZ_ALL #undef DO_FPCMP_PPZZ_D #undef DO_FPCMP_PPZZ_S #undef DO_FPCMP_PPZZ_H #undef DO_FPCMP_PPZZ /* One operand floating-point comparison against zero, controlled * by a predicate. */ #define DO_FPCMP_PPZ0(NAME, TYPE, H, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, \ void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc), j = (i - 1) >> 6; \ uint64_t *d = vd, *g = vg; \ do { \ uint64_t out = 0, pg = g[j]; \ do { \ i -= sizeof(TYPE), out <<= sizeof(TYPE); \ if ((pg >> (i & 63)) & 1) { \ TYPE nn = *(TYPE *)(vn + H(i)); \ out |= OP(TYPE, nn, 0, status); \ } \ } while (i & 63); \ d[j--] = out; \ } while (i > 0); \ } #define DO_FPCMP_PPZ0_H(NAME, OP) \ DO_FPCMP_PPZ0(NAME##_h, float16, H1_2, OP) #define DO_FPCMP_PPZ0_S(NAME, OP) \ DO_FPCMP_PPZ0(NAME##_s, float32, H1_4, OP) #define DO_FPCMP_PPZ0_D(NAME, OP) \ DO_FPCMP_PPZ0(NAME##_d, float64, H1_8, OP) #define DO_FPCMP_PPZ0_ALL(NAME, OP) \ DO_FPCMP_PPZ0_H(NAME, OP) \ DO_FPCMP_PPZ0_S(NAME, OP) \ DO_FPCMP_PPZ0_D(NAME, OP) DO_FPCMP_PPZ0_ALL(sve_fcmge0, DO_FCMGE) DO_FPCMP_PPZ0_ALL(sve_fcmgt0, DO_FCMGT) DO_FPCMP_PPZ0_ALL(sve_fcmle0, DO_FCMLE) DO_FPCMP_PPZ0_ALL(sve_fcmlt0, DO_FCMLT) DO_FPCMP_PPZ0_ALL(sve_fcmeq0, DO_FCMEQ) DO_FPCMP_PPZ0_ALL(sve_fcmne0, DO_FCMNE) /* FP Trig Multiply-Add. */ void HELPER(sve_ftmad_h)(void *vd, void *vn, void *vm, void *vs, uint32_t desc) { static const float16 coeff[16] = { 0x3c00, 0xb155, 0x2030, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, 0x3c00, 0xb800, 0x293a, 0x0000, 0x0000, 0x0000, 0x0000, 0x0000, }; intptr_t i, opr_sz = simd_oprsz(desc) / sizeof(float16); intptr_t x = simd_data(desc); float16 *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i++) { float16 mm = m[i]; intptr_t xx = x; if (float16_is_neg(mm)) { mm = float16_abs(mm); xx += 8; } d[i] = float16_muladd(n[i], mm, coeff[xx], 0, vs); } } void HELPER(sve_ftmad_s)(void *vd, void *vn, void *vm, void *vs, uint32_t desc) { static const float32 coeff[16] = { 0x3f800000, 0xbe2aaaab, 0x3c088886, 0xb95008b9, 0x36369d6d, 0x00000000, 0x00000000, 0x00000000, 0x3f800000, 0xbf000000, 0x3d2aaaa6, 0xbab60705, 0x37cd37cc, 0x00000000, 0x00000000, 0x00000000, }; intptr_t i, opr_sz = simd_oprsz(desc) / sizeof(float32); intptr_t x = simd_data(desc); float32 *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i++) { float32 mm = m[i]; intptr_t xx = x; if (float32_is_neg(mm)) { mm = float32_abs(mm); xx += 8; } d[i] = float32_muladd(n[i], mm, coeff[xx], 0, vs); } } void HELPER(sve_ftmad_d)(void *vd, void *vn, void *vm, void *vs, uint32_t desc) { static const float64 coeff[16] = { 0x3ff0000000000000ull, 0xbfc5555555555543ull, 0x3f8111111110f30cull, 0xbf2a01a019b92fc6ull, 0x3ec71de351f3d22bull, 0xbe5ae5e2b60f7b91ull, 0x3de5d8408868552full, 0x0000000000000000ull, 0x3ff0000000000000ull, 0xbfe0000000000000ull, 0x3fa5555555555536ull, 0xbf56c16c16c13a0bull, 0x3efa01a019b1e8d8ull, 0xbe927e4f7282f468ull, 0x3e21ee96d2641b13ull, 0xbda8f76380fbb401ull, }; intptr_t i, opr_sz = simd_oprsz(desc) / sizeof(float64); intptr_t x = simd_data(desc); float64 *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; i++) { float64 mm = m[i]; intptr_t xx = x; if (float64_is_neg(mm)) { mm = float64_abs(mm); xx += 8; } d[i] = float64_muladd(n[i], mm, coeff[xx], 0, vs); } } /* * FP Complex Add */ void HELPER(sve_fcadd_h)(void *vd, void *vn, void *vm, void *vg, void *vs, uint32_t desc) { intptr_t j, i = simd_oprsz(desc); uint64_t *g = vg; float16 neg_imag = float16_set_sign(0, simd_data(desc)); float16 neg_real = float16_chs(neg_imag); do { uint64_t pg = g[(i - 1) >> 6]; do { float16 e0, e1, e2, e3; /* I holds the real index; J holds the imag index. */ j = i - sizeof(float16); i -= 2 * sizeof(float16); e0 = *(float16 *)(vn + H1_2(i)); e1 = *(float16 *)(vm + H1_2(j)) ^ neg_real; e2 = *(float16 *)(vn + H1_2(j)); e3 = *(float16 *)(vm + H1_2(i)) ^ neg_imag; if (likely((pg >> (i & 63)) & 1)) { *(float16 *)(vd + H1_2(i)) = float16_add(e0, e1, vs); } if (likely((pg >> (j & 63)) & 1)) { *(float16 *)(vd + H1_2(j)) = float16_add(e2, e3, vs); } } while (i & 63); } while (i != 0); } void HELPER(sve_fcadd_s)(void *vd, void *vn, void *vm, void *vg, void *vs, uint32_t desc) { intptr_t j, i = simd_oprsz(desc); uint64_t *g = vg; float32 neg_imag = float32_set_sign(0, simd_data(desc)); float32 neg_real = float32_chs(neg_imag); do { uint64_t pg = g[(i - 1) >> 6]; do { float32 e0, e1, e2, e3; /* I holds the real index; J holds the imag index. */ j = i - sizeof(float32); i -= 2 * sizeof(float32); e0 = *(float32 *)(vn + H1_2(i)); e1 = *(float32 *)(vm + H1_2(j)) ^ neg_real; e2 = *(float32 *)(vn + H1_2(j)); e3 = *(float32 *)(vm + H1_2(i)) ^ neg_imag; if (likely((pg >> (i & 63)) & 1)) { *(float32 *)(vd + H1_2(i)) = float32_add(e0, e1, vs); } if (likely((pg >> (j & 63)) & 1)) { *(float32 *)(vd + H1_2(j)) = float32_add(e2, e3, vs); } } while (i & 63); } while (i != 0); } void HELPER(sve_fcadd_d)(void *vd, void *vn, void *vm, void *vg, void *vs, uint32_t desc) { intptr_t j, i = simd_oprsz(desc); uint64_t *g = vg; float64 neg_imag = float64_set_sign(0, simd_data(desc)); float64 neg_real = float64_chs(neg_imag); do { uint64_t pg = g[(i - 1) >> 6]; do { float64 e0, e1, e2, e3; /* I holds the real index; J holds the imag index. */ j = i - sizeof(float64); i -= 2 * sizeof(float64); e0 = *(float64 *)(vn + H1_2(i)); e1 = *(float64 *)(vm + H1_2(j)) ^ neg_real; e2 = *(float64 *)(vn + H1_2(j)); e3 = *(float64 *)(vm + H1_2(i)) ^ neg_imag; if (likely((pg >> (i & 63)) & 1)) { *(float64 *)(vd + H1_2(i)) = float64_add(e0, e1, vs); } if (likely((pg >> (j & 63)) & 1)) { *(float64 *)(vd + H1_2(j)) = float64_add(e2, e3, vs); } } while (i & 63); } while (i != 0); } /* * FP Complex Multiply */ void HELPER(sve_fcmla_zpzzz_h)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { intptr_t j, i = simd_oprsz(desc); unsigned rot = simd_data(desc); bool flip = rot & 1; float16 neg_imag, neg_real; uint64_t *g = vg; neg_imag = float16_set_sign(0, (rot & 2) != 0); neg_real = float16_set_sign(0, rot == 1 || rot == 2); do { uint64_t pg = g[(i - 1) >> 6]; do { float16 e1, e2, e3, e4, nr, ni, mr, mi, d; /* I holds the real index; J holds the imag index. */ j = i - sizeof(float16); i -= 2 * sizeof(float16); nr = *(float16 *)(vn + H1_2(i)); ni = *(float16 *)(vn + H1_2(j)); mr = *(float16 *)(vm + H1_2(i)); mi = *(float16 *)(vm + H1_2(j)); e2 = (flip ? ni : nr); e1 = (flip ? mi : mr) ^ neg_real; e4 = e2; e3 = (flip ? mr : mi) ^ neg_imag; if (likely((pg >> (i & 63)) & 1)) { d = *(float16 *)(va + H1_2(i)); d = float16_muladd(e2, e1, d, 0, status); *(float16 *)(vd + H1_2(i)) = d; } if (likely((pg >> (j & 63)) & 1)) { d = *(float16 *)(va + H1_2(j)); d = float16_muladd(e4, e3, d, 0, status); *(float16 *)(vd + H1_2(j)) = d; } } while (i & 63); } while (i != 0); } void HELPER(sve_fcmla_zpzzz_s)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { intptr_t j, i = simd_oprsz(desc); unsigned rot = simd_data(desc); bool flip = rot & 1; float32 neg_imag, neg_real; uint64_t *g = vg; neg_imag = float32_set_sign(0, (rot & 2) != 0); neg_real = float32_set_sign(0, rot == 1 || rot == 2); do { uint64_t pg = g[(i - 1) >> 6]; do { float32 e1, e2, e3, e4, nr, ni, mr, mi, d; /* I holds the real index; J holds the imag index. */ j = i - sizeof(float32); i -= 2 * sizeof(float32); nr = *(float32 *)(vn + H1_2(i)); ni = *(float32 *)(vn + H1_2(j)); mr = *(float32 *)(vm + H1_2(i)); mi = *(float32 *)(vm + H1_2(j)); e2 = (flip ? ni : nr); e1 = (flip ? mi : mr) ^ neg_real; e4 = e2; e3 = (flip ? mr : mi) ^ neg_imag; if (likely((pg >> (i & 63)) & 1)) { d = *(float32 *)(va + H1_2(i)); d = float32_muladd(e2, e1, d, 0, status); *(float32 *)(vd + H1_2(i)) = d; } if (likely((pg >> (j & 63)) & 1)) { d = *(float32 *)(va + H1_2(j)); d = float32_muladd(e4, e3, d, 0, status); *(float32 *)(vd + H1_2(j)) = d; } } while (i & 63); } while (i != 0); } void HELPER(sve_fcmla_zpzzz_d)(void *vd, void *vn, void *vm, void *va, void *vg, void *status, uint32_t desc) { intptr_t j, i = simd_oprsz(desc); unsigned rot = simd_data(desc); bool flip = rot & 1; float64 neg_imag, neg_real; uint64_t *g = vg; neg_imag = float64_set_sign(0, (rot & 2) != 0); neg_real = float64_set_sign(0, rot == 1 || rot == 2); do { uint64_t pg = g[(i - 1) >> 6]; do { float64 e1, e2, e3, e4, nr, ni, mr, mi, d; /* I holds the real index; J holds the imag index. */ j = i - sizeof(float64); i -= 2 * sizeof(float64); nr = *(float64 *)(vn + H1_2(i)); ni = *(float64 *)(vn + H1_2(j)); mr = *(float64 *)(vm + H1_2(i)); mi = *(float64 *)(vm + H1_2(j)); e2 = (flip ? ni : nr); e1 = (flip ? mi : mr) ^ neg_real; e4 = e2; e3 = (flip ? mr : mi) ^ neg_imag; if (likely((pg >> (i & 63)) & 1)) { d = *(float64 *)(va + H1_2(i)); d = float64_muladd(e2, e1, d, 0, status); *(float64 *)(vd + H1_2(i)) = d; } if (likely((pg >> (j & 63)) & 1)) { d = *(float64 *)(va + H1_2(j)); d = float64_muladd(e4, e3, d, 0, status); *(float64 *)(vd + H1_2(j)) = d; } } while (i & 63); } while (i != 0); } /* * Load contiguous data, protected by a governing predicate. */ /* * Load one element into @vd + @reg_off from @host. * The controlling predicate is known to be true. */ typedef void sve_ldst1_host_fn(void *vd, intptr_t reg_off, void *host); /* * Load one element into @vd + @reg_off from (@env, @vaddr, @ra). * The controlling predicate is known to be true. */ typedef void sve_ldst1_tlb_fn(CPUARMState *env, void *vd, intptr_t reg_off, target_ulong vaddr, uintptr_t retaddr); /* * Generate the above primitives. */ #define DO_LD_HOST(NAME, H, TYPEE, TYPEM, HOST) \ static void sve_##NAME##_host(void *vd, intptr_t reg_off, void *host) \ { \ TYPEM val = HOST(host); \ *(TYPEE *)(vd + H(reg_off)) = val; \ } #define DO_ST_HOST(NAME, H, TYPEE, TYPEM, HOST) \ static void sve_##NAME##_host(void *vd, intptr_t reg_off, void *host) \ { HOST(host, (TYPEM)*(TYPEE *)(vd + H(reg_off))); } #define DO_LD_TLB(NAME, H, TYPEE, TYPEM, TLB) \ static void sve_##NAME##_tlb(CPUARMState *env, void *vd, intptr_t reg_off, \ target_ulong addr, uintptr_t ra) \ { \ *(TYPEE *)(vd + H(reg_off)) = \ (TYPEM)TLB(env, useronly_clean_ptr(addr), ra); \ } #define DO_ST_TLB(NAME, H, TYPEE, TYPEM, TLB) \ static void sve_##NAME##_tlb(CPUARMState *env, void *vd, intptr_t reg_off, \ target_ulong addr, uintptr_t ra) \ { \ TLB(env, useronly_clean_ptr(addr), \ (TYPEM)*(TYPEE *)(vd + H(reg_off)), ra); \ } #define DO_LD_PRIM_1(NAME, H, TE, TM) \ DO_LD_HOST(NAME, H, TE, TM, ldub_p) \ DO_LD_TLB(NAME, H, TE, TM, cpu_ldub_data_ra) DO_LD_PRIM_1(ld1bb, H1, uint8_t, uint8_t) DO_LD_PRIM_1(ld1bhu, H1_2, uint16_t, uint8_t) DO_LD_PRIM_1(ld1bhs, H1_2, uint16_t, int8_t) DO_LD_PRIM_1(ld1bsu, H1_4, uint32_t, uint8_t) DO_LD_PRIM_1(ld1bss, H1_4, uint32_t, int8_t) DO_LD_PRIM_1(ld1bdu, H1_8, uint64_t, uint8_t) DO_LD_PRIM_1(ld1bds, H1_8, uint64_t, int8_t) #define DO_ST_PRIM_1(NAME, H, TE, TM) \ DO_ST_HOST(st1##NAME, H, TE, TM, stb_p) \ DO_ST_TLB(st1##NAME, H, TE, TM, cpu_stb_data_ra) DO_ST_PRIM_1(bb, H1, uint8_t, uint8_t) DO_ST_PRIM_1(bh, H1_2, uint16_t, uint8_t) DO_ST_PRIM_1(bs, H1_4, uint32_t, uint8_t) DO_ST_PRIM_1(bd, H1_8, uint64_t, uint8_t) #define DO_LD_PRIM_2(NAME, H, TE, TM, LD) \ DO_LD_HOST(ld1##NAME##_be, H, TE, TM, LD##_be_p) \ DO_LD_HOST(ld1##NAME##_le, H, TE, TM, LD##_le_p) \ DO_LD_TLB(ld1##NAME##_be, H, TE, TM, cpu_##LD##_be_data_ra) \ DO_LD_TLB(ld1##NAME##_le, H, TE, TM, cpu_##LD##_le_data_ra) #define DO_ST_PRIM_2(NAME, H, TE, TM, ST) \ DO_ST_HOST(st1##NAME##_be, H, TE, TM, ST##_be_p) \ DO_ST_HOST(st1##NAME##_le, H, TE, TM, ST##_le_p) \ DO_ST_TLB(st1##NAME##_be, H, TE, TM, cpu_##ST##_be_data_ra) \ DO_ST_TLB(st1##NAME##_le, H, TE, TM, cpu_##ST##_le_data_ra) DO_LD_PRIM_2(hh, H1_2, uint16_t, uint16_t, lduw) DO_LD_PRIM_2(hsu, H1_4, uint32_t, uint16_t, lduw) DO_LD_PRIM_2(hss, H1_4, uint32_t, int16_t, lduw) DO_LD_PRIM_2(hdu, H1_8, uint64_t, uint16_t, lduw) DO_LD_PRIM_2(hds, H1_8, uint64_t, int16_t, lduw) DO_ST_PRIM_2(hh, H1_2, uint16_t, uint16_t, stw) DO_ST_PRIM_2(hs, H1_4, uint32_t, uint16_t, stw) DO_ST_PRIM_2(hd, H1_8, uint64_t, uint16_t, stw) DO_LD_PRIM_2(ss, H1_4, uint32_t, uint32_t, ldl) DO_LD_PRIM_2(sdu, H1_8, uint64_t, uint32_t, ldl) DO_LD_PRIM_2(sds, H1_8, uint64_t, int32_t, ldl) DO_ST_PRIM_2(ss, H1_4, uint32_t, uint32_t, stl) DO_ST_PRIM_2(sd, H1_8, uint64_t, uint32_t, stl) DO_LD_PRIM_2(dd, H1_8, uint64_t, uint64_t, ldq) DO_ST_PRIM_2(dd, H1_8, uint64_t, uint64_t, stq) #undef DO_LD_TLB #undef DO_ST_TLB #undef DO_LD_HOST #undef DO_LD_PRIM_1 #undef DO_ST_PRIM_1 #undef DO_LD_PRIM_2 #undef DO_ST_PRIM_2 /* * Skip through a sequence of inactive elements in the guarding predicate @vg, * beginning at @reg_off bounded by @reg_max. Return the offset of the active * element >= @reg_off, or @reg_max if there were no active elements at all. */ static intptr_t find_next_active(uint64_t *vg, intptr_t reg_off, intptr_t reg_max, int esz) { uint64_t pg_mask = pred_esz_masks[esz]; uint64_t pg = (vg[reg_off >> 6] & pg_mask) >> (reg_off & 63); /* In normal usage, the first element is active. */ if (likely(pg & 1)) { return reg_off; } if (pg == 0) { reg_off &= -64; do { reg_off += 64; if (unlikely(reg_off >= reg_max)) { /* The entire predicate was false. */ return reg_max; } pg = vg[reg_off >> 6] & pg_mask; } while (pg == 0); } reg_off += ctz64(pg); /* We should never see an out of range predicate bit set. */ tcg_debug_assert(reg_off < reg_max); return reg_off; } /* * Resolve the guest virtual address to info->host and info->flags. * If @nofault, return false if the page is invalid, otherwise * exit via page fault exception. */ typedef struct { void *host; int flags; MemTxAttrs attrs; } SVEHostPage; static bool sve_probe_page(SVEHostPage *info, bool nofault, CPUARMState *env, target_ulong addr, int mem_off, MMUAccessType access_type, int mmu_idx, uintptr_t retaddr) { int flags; addr += mem_off; /* * User-only currently always issues with TBI. See the comment * above useronly_clean_ptr. Usually we clean this top byte away * during translation, but we can't do that for e.g. vector + imm * addressing modes. * * We currently always enable TBI for user-only, and do not provide * a way to turn it off. So clean the pointer unconditionally here, * rather than look it up here, or pass it down from above. */ addr = useronly_clean_ptr(addr); flags = probe_access_flags(env, addr, access_type, mmu_idx, nofault, &info->host, retaddr); info->flags = flags; if (flags & TLB_INVALID_MASK) { g_assert(nofault); return false; } /* Ensure that info->host[] is relative to addr, not addr + mem_off. */ info->host -= mem_off; #ifdef CONFIG_USER_ONLY memset(&info->attrs, 0, sizeof(info->attrs)); #else /* * Find the iotlbentry for addr and return the transaction attributes. * This *must* be present in the TLB because we just found the mapping. */ { uintptr_t index = tlb_index(env, mmu_idx, addr); # ifdef CONFIG_DEBUG_TCG CPUTLBEntry *entry = tlb_entry(env, mmu_idx, addr); target_ulong comparator = (access_type == MMU_DATA_LOAD ? entry->addr_read : tlb_addr_write(entry)); g_assert(tlb_hit(comparator, addr)); # endif CPUIOTLBEntry *iotlbentry = &env_tlb(env)->d[mmu_idx].iotlb[index]; info->attrs = iotlbentry->attrs; } #endif return true; } /* * Analyse contiguous data, protected by a governing predicate. */ typedef enum { FAULT_NO, FAULT_FIRST, FAULT_ALL, } SVEContFault; typedef struct { /* * First and last element wholly contained within the two pages. * mem_off_first[0] and reg_off_first[0] are always set >= 0. * reg_off_last[0] may be < 0 if the first element crosses pages. * All of mem_off_first[1], reg_off_first[1] and reg_off_last[1] * are set >= 0 only if there are complete elements on a second page. * * The reg_off_* offsets are relative to the internal vector register. * The mem_off_first offset is relative to the memory address; the * two offsets are different when a load operation extends, a store * operation truncates, or for multi-register operations. */ int16_t mem_off_first[2]; int16_t reg_off_first[2]; int16_t reg_off_last[2]; /* * One element that is misaligned and spans both pages, * or -1 if there is no such active element. */ int16_t mem_off_split; int16_t reg_off_split; /* * The byte offset at which the entire operation crosses a page boundary. * Set >= 0 if and only if the entire operation spans two pages. */ int16_t page_split; /* TLB data for the two pages. */ SVEHostPage page[2]; } SVEContLdSt; /* * Find first active element on each page, and a loose bound for the * final element on each page. Identify any single element that spans * the page boundary. Return true if there are any active elements. */ static bool sve_cont_ldst_elements(SVEContLdSt *info, target_ulong addr, uint64_t *vg, intptr_t reg_max, int esz, int msize) { const int esize = 1 << esz; const uint64_t pg_mask = pred_esz_masks[esz]; intptr_t reg_off_first = -1, reg_off_last = -1, reg_off_split; intptr_t mem_off_last, mem_off_split; intptr_t page_split, elt_split; intptr_t i; /* Set all of the element indices to -1, and the TLB data to 0. */ memset(info, -1, offsetof(SVEContLdSt, page)); memset(info->page, 0, sizeof(info->page)); /* Gross scan over the entire predicate to find bounds. */ i = 0; do { uint64_t pg = vg[i] & pg_mask; if (pg) { reg_off_last = i * 64 + 63 - clz64(pg); if (reg_off_first < 0) { reg_off_first = i * 64 + ctz64(pg); } } } while (++i * 64 < reg_max); if (unlikely(reg_off_first < 0)) { /* No active elements, no pages touched. */ return false; } tcg_debug_assert(reg_off_last >= 0 && reg_off_last < reg_max); info->reg_off_first[0] = reg_off_first; info->mem_off_first[0] = (reg_off_first >> esz) * msize; mem_off_last = (reg_off_last >> esz) * msize; page_split = -(addr | TARGET_PAGE_MASK); if (likely(mem_off_last + msize <= page_split)) { /* The entire operation fits within a single page. */ info->reg_off_last[0] = reg_off_last; return true; } info->page_split = page_split; elt_split = page_split / msize; reg_off_split = elt_split << esz; mem_off_split = elt_split * msize; /* * This is the last full element on the first page, but it is not * necessarily active. If there is no full element, i.e. the first * active element is the one that's split, this value remains -1. * It is useful as iteration bounds. */ if (elt_split != 0) { info->reg_off_last[0] = reg_off_split - esize; } /* Determine if an unaligned element spans the pages. */ if (page_split % msize != 0) { /* It is helpful to know if the split element is active. */ if ((vg[reg_off_split >> 6] >> (reg_off_split & 63)) & 1) { info->reg_off_split = reg_off_split; info->mem_off_split = mem_off_split; if (reg_off_split == reg_off_last) { /* The page crossing element is last. */ return true; } } reg_off_split += esize; mem_off_split += msize; } /* * We do want the first active element on the second page, because * this may affect the address reported in an exception. */ reg_off_split = find_next_active(vg, reg_off_split, reg_max, esz); tcg_debug_assert(reg_off_split <= reg_off_last); info->reg_off_first[1] = reg_off_split; info->mem_off_first[1] = (reg_off_split >> esz) * msize; info->reg_off_last[1] = reg_off_last; return true; } /* * Resolve the guest virtual addresses to info->page[]. * Control the generation of page faults with @fault. Return false if * there is no work to do, which can only happen with @fault == FAULT_NO. */ static bool sve_cont_ldst_pages(SVEContLdSt *info, SVEContFault fault, CPUARMState *env, target_ulong addr, MMUAccessType access_type, uintptr_t retaddr) { int mmu_idx = cpu_mmu_index(env, false); int mem_off = info->mem_off_first[0]; bool nofault = fault == FAULT_NO; bool have_work = true; if (!sve_probe_page(&info->page[0], nofault, env, addr, mem_off, access_type, mmu_idx, retaddr)) { /* No work to be done. */ return false; } if (likely(info->page_split < 0)) { /* The entire operation was on the one page. */ return true; } /* * If the second page is invalid, then we want the fault address to be * the first byte on that page which is accessed. */ if (info->mem_off_split >= 0) { /* * There is an element split across the pages. The fault address * should be the first byte of the second page. */ mem_off = info->page_split; /* * If the split element is also the first active element * of the vector, then: For first-fault we should continue * to generate faults for the second page. For no-fault, * we have work only if the second page is valid. */ if (info->mem_off_first[0] < info->mem_off_split) { nofault = FAULT_FIRST; have_work = false; } } else { /* * There is no element split across the pages. The fault address * should be the first active element on the second page. */ mem_off = info->mem_off_first[1]; /* * There must have been one active element on the first page, * so we're out of first-fault territory. */ nofault = fault != FAULT_ALL; } have_work |= sve_probe_page(&info->page[1], nofault, env, addr, mem_off, access_type, mmu_idx, retaddr); return have_work; } static void sve_cont_ldst_watchpoints(SVEContLdSt *info, CPUARMState *env, uint64_t *vg, target_ulong addr, int esize, int msize, int wp_access, uintptr_t retaddr) { #ifndef CONFIG_USER_ONLY intptr_t mem_off, reg_off, reg_last; int flags0 = info->page[0].flags; int flags1 = info->page[1].flags; if (likely(!((flags0 | flags1) & TLB_WATCHPOINT))) { return; } /* Indicate that watchpoints are handled. */ info->page[0].flags = flags0 & ~TLB_WATCHPOINT; info->page[1].flags = flags1 & ~TLB_WATCHPOINT; if (flags0 & TLB_WATCHPOINT) { mem_off = info->mem_off_first[0]; reg_off = info->reg_off_first[0]; reg_last = info->reg_off_last[0]; while (reg_off <= reg_last) { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { cpu_check_watchpoint(env_cpu(env), addr + mem_off, msize, info->page[0].attrs, wp_access, retaddr); } reg_off += esize; mem_off += msize; } while (reg_off <= reg_last && (reg_off & 63)); } } mem_off = info->mem_off_split; if (mem_off >= 0) { cpu_check_watchpoint(env_cpu(env), addr + mem_off, msize, info->page[0].attrs, wp_access, retaddr); } mem_off = info->mem_off_first[1]; if ((flags1 & TLB_WATCHPOINT) && mem_off >= 0) { reg_off = info->reg_off_first[1]; reg_last = info->reg_off_last[1]; do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { cpu_check_watchpoint(env_cpu(env), addr + mem_off, msize, info->page[1].attrs, wp_access, retaddr); } reg_off += esize; mem_off += msize; } while (reg_off & 63); } while (reg_off <= reg_last); } #endif } static void sve_cont_ldst_mte_check(SVEContLdSt *info, CPUARMState *env, uint64_t *vg, target_ulong addr, int esize, int msize, uint32_t mtedesc, uintptr_t ra) { intptr_t mem_off, reg_off, reg_last; /* Process the page only if MemAttr == Tagged. */ if (arm_tlb_mte_tagged(&info->page[0].attrs)) { mem_off = info->mem_off_first[0]; reg_off = info->reg_off_first[0]; reg_last = info->reg_off_split; if (reg_last < 0) { reg_last = info->reg_off_last[0]; } do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { mte_check(env, mtedesc, addr, ra); } reg_off += esize; mem_off += msize; } while (reg_off <= reg_last && (reg_off & 63)); } while (reg_off <= reg_last); } mem_off = info->mem_off_first[1]; if (mem_off >= 0 && arm_tlb_mte_tagged(&info->page[1].attrs)) { reg_off = info->reg_off_first[1]; reg_last = info->reg_off_last[1]; do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { mte_check(env, mtedesc, addr, ra); } reg_off += esize; mem_off += msize; } while (reg_off & 63); } while (reg_off <= reg_last); } } /* * Common helper for all contiguous 1,2,3,4-register predicated stores. */ static inline QEMU_ALWAYS_INLINE void sve_ldN_r(CPUARMState *env, uint64_t *vg, const target_ulong addr, uint32_t desc, const uintptr_t retaddr, const int esz, const int msz, const int N, uint32_t mtedesc, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { const unsigned rd = simd_data(desc); const intptr_t reg_max = simd_oprsz(desc); intptr_t reg_off, reg_last, mem_off; SVEContLdSt info; void *host; int flags, i; /* Find the active elements. */ if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, N << msz)) { /* The entire predicate was false; no load occurs. */ for (i = 0; i < N; ++i) { memset(&env->vfp.zregs[(rd + i) & 31], 0, reg_max); } return; } /* Probe the page(s). Exit with exception for any invalid page. */ sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_LOAD, retaddr); /* Handle watchpoints for all active elements. */ sve_cont_ldst_watchpoints(&info, env, vg, addr, 1 << esz, N << msz, BP_MEM_READ, retaddr); /* * Handle mte checks for all active elements. * Since TBI must be set for MTE, !mtedesc => !mte_active. */ if (mtedesc) { sve_cont_ldst_mte_check(&info, env, vg, addr, 1 << esz, N << msz, mtedesc, retaddr); } flags = info.page[0].flags | info.page[1].flags; if (unlikely(flags != 0)) { #ifdef CONFIG_USER_ONLY g_assert_not_reached(); #else /* * At least one page includes MMIO. * Any bus operation can fail with cpu_transaction_failed, * which for ARM will raise SyncExternal. Perform the load * into scratch memory to preserve register state until the end. */ ARMVectorReg scratch[4] = { }; mem_off = info.mem_off_first[0]; reg_off = info.reg_off_first[0]; reg_last = info.reg_off_last[1]; if (reg_last < 0) { reg_last = info.reg_off_split; if (reg_last < 0) { reg_last = info.reg_off_last[0]; } } do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { for (i = 0; i < N; ++i) { tlb_fn(env, &scratch[i], reg_off, addr + mem_off + (i << msz), retaddr); } } reg_off += 1 << esz; mem_off += N << msz; } while (reg_off & 63); } while (reg_off <= reg_last); for (i = 0; i < N; ++i) { memcpy(&env->vfp.zregs[(rd + i) & 31], &scratch[i], reg_max); } return; #endif } /* The entire operation is in RAM, on valid pages. */ for (i = 0; i < N; ++i) { memset(&env->vfp.zregs[(rd + i) & 31], 0, reg_max); } mem_off = info.mem_off_first[0]; reg_off = info.reg_off_first[0]; reg_last = info.reg_off_last[0]; host = info.page[0].host; while (reg_off <= reg_last) { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { for (i = 0; i < N; ++i) { host_fn(&env->vfp.zregs[(rd + i) & 31], reg_off, host + mem_off + (i << msz)); } } reg_off += 1 << esz; mem_off += N << msz; } while (reg_off <= reg_last && (reg_off & 63)); } /* * Use the slow path to manage the cross-page misalignment. * But we know this is RAM and cannot trap. */ mem_off = info.mem_off_split; if (unlikely(mem_off >= 0)) { reg_off = info.reg_off_split; for (i = 0; i < N; ++i) { tlb_fn(env, &env->vfp.zregs[(rd + i) & 31], reg_off, addr + mem_off + (i << msz), retaddr); } } mem_off = info.mem_off_first[1]; if (unlikely(mem_off >= 0)) { reg_off = info.reg_off_first[1]; reg_last = info.reg_off_last[1]; host = info.page[1].host; do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { for (i = 0; i < N; ++i) { host_fn(&env->vfp.zregs[(rd + i) & 31], reg_off, host + mem_off + (i << msz)); } } reg_off += 1 << esz; mem_off += N << msz; } while (reg_off & 63); } while (reg_off <= reg_last); } } static inline QEMU_ALWAYS_INLINE void sve_ldN_r_mte(CPUARMState *env, uint64_t *vg, target_ulong addr, uint32_t desc, const uintptr_t ra, const int esz, const int msz, const int N, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); int bit55 = extract64(addr, 55, 1); /* Remove mtedesc from the normal sve descriptor. */ desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* Perform gross MTE suppression early. */ if (!tbi_check(desc, bit55) || tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { mtedesc = 0; } sve_ldN_r(env, vg, addr, desc, ra, esz, msz, N, mtedesc, host_fn, tlb_fn); } #define DO_LD1_1(NAME, ESZ) \ void HELPER(sve_##NAME##_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, 1, 0, \ sve_##NAME##_host, sve_##NAME##_tlb); \ } \ void HELPER(sve_##NAME##_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, 1, \ sve_##NAME##_host, sve_##NAME##_tlb); \ } #define DO_LD1_2(NAME, ESZ, MSZ) \ void HELPER(sve_##NAME##_le_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1, 0, \ sve_##NAME##_le_host, sve_##NAME##_le_tlb); \ } \ void HELPER(sve_##NAME##_be_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1, 0, \ sve_##NAME##_be_host, sve_##NAME##_be_tlb); \ } \ void HELPER(sve_##NAME##_le_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1, \ sve_##NAME##_le_host, sve_##NAME##_le_tlb); \ } \ void HELPER(sve_##NAME##_be_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, 1, \ sve_##NAME##_be_host, sve_##NAME##_be_tlb); \ } DO_LD1_1(ld1bb, MO_8) DO_LD1_1(ld1bhu, MO_16) DO_LD1_1(ld1bhs, MO_16) DO_LD1_1(ld1bsu, MO_32) DO_LD1_1(ld1bss, MO_32) DO_LD1_1(ld1bdu, MO_64) DO_LD1_1(ld1bds, MO_64) DO_LD1_2(ld1hh, MO_16, MO_16) DO_LD1_2(ld1hsu, MO_32, MO_16) DO_LD1_2(ld1hss, MO_32, MO_16) DO_LD1_2(ld1hdu, MO_64, MO_16) DO_LD1_2(ld1hds, MO_64, MO_16) DO_LD1_2(ld1ss, MO_32, MO_32) DO_LD1_2(ld1sdu, MO_64, MO_32) DO_LD1_2(ld1sds, MO_64, MO_32) DO_LD1_2(ld1dd, MO_64, MO_64) #undef DO_LD1_1 #undef DO_LD1_2 #define DO_LDN_1(N) \ void HELPER(sve_ld##N##bb_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r(env, vg, addr, desc, GETPC(), MO_8, MO_8, N, 0, \ sve_ld1bb_host, sve_ld1bb_tlb); \ } \ void HELPER(sve_ld##N##bb_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r_mte(env, vg, addr, desc, GETPC(), MO_8, MO_8, N, \ sve_ld1bb_host, sve_ld1bb_tlb); \ } #define DO_LDN_2(N, SUFF, ESZ) \ void HELPER(sve_ld##N##SUFF##_le_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, ESZ, N, 0, \ sve_ld1##SUFF##_le_host, sve_ld1##SUFF##_le_tlb); \ } \ void HELPER(sve_ld##N##SUFF##_be_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r(env, vg, addr, desc, GETPC(), ESZ, ESZ, N, 0, \ sve_ld1##SUFF##_be_host, sve_ld1##SUFF##_be_tlb); \ } \ void HELPER(sve_ld##N##SUFF##_le_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, ESZ, N, \ sve_ld1##SUFF##_le_host, sve_ld1##SUFF##_le_tlb); \ } \ void HELPER(sve_ld##N##SUFF##_be_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldN_r_mte(env, vg, addr, desc, GETPC(), ESZ, ESZ, N, \ sve_ld1##SUFF##_be_host, sve_ld1##SUFF##_be_tlb); \ } DO_LDN_1(2) DO_LDN_1(3) DO_LDN_1(4) DO_LDN_2(2, hh, MO_16) DO_LDN_2(3, hh, MO_16) DO_LDN_2(4, hh, MO_16) DO_LDN_2(2, ss, MO_32) DO_LDN_2(3, ss, MO_32) DO_LDN_2(4, ss, MO_32) DO_LDN_2(2, dd, MO_64) DO_LDN_2(3, dd, MO_64) DO_LDN_2(4, dd, MO_64) #undef DO_LDN_1 #undef DO_LDN_2 /* * Load contiguous data, first-fault and no-fault. * * For user-only, one could argue that we should hold the mmap_lock during * the operation so that there is no race between page_check_range and the * load operation. However, unmapping pages out from under a running thread * is extraordinarily unlikely. This theoretical race condition also affects * linux-user/ in its get_user/put_user macros. * * TODO: Construct some helpers, written in assembly, that interact with * host_signal_handler to produce memory ops which can properly report errors * without racing. */ /* Fault on byte I. All bits in FFR from I are cleared. The vector * result from I is CONSTRAINED UNPREDICTABLE; we choose the MERGE * option, which leaves subsequent data unchanged. */ static void record_fault(CPUARMState *env, uintptr_t i, uintptr_t oprsz) { uint64_t *ffr = env->vfp.pregs[FFR_PRED_NUM].p; if (i & 63) { ffr[i / 64] &= MAKE_64BIT_MASK(0, i & 63); i = ROUND_UP(i, 64); } for (; i < oprsz; i += 64) { ffr[i / 64] = 0; } } /* * Common helper for all contiguous no-fault and first-fault loads. */ static inline QEMU_ALWAYS_INLINE void sve_ldnfff1_r(CPUARMState *env, void *vg, const target_ulong addr, uint32_t desc, const uintptr_t retaddr, uint32_t mtedesc, const int esz, const int msz, const SVEContFault fault, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { const unsigned rd = simd_data(desc); void *vd = &env->vfp.zregs[rd]; const intptr_t reg_max = simd_oprsz(desc); intptr_t reg_off, mem_off, reg_last; SVEContLdSt info; int flags; void *host; /* Find the active elements. */ if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, 1 << msz)) { /* The entire predicate was false; no load occurs. */ memset(vd, 0, reg_max); return; } reg_off = info.reg_off_first[0]; /* Probe the page(s). */ if (!sve_cont_ldst_pages(&info, fault, env, addr, MMU_DATA_LOAD, retaddr)) { /* Fault on first element. */ tcg_debug_assert(fault == FAULT_NO); memset(vd, 0, reg_max); goto do_fault; } mem_off = info.mem_off_first[0]; flags = info.page[0].flags; /* * Disable MTE checking if the Tagged bit is not set. Since TBI must * be set within MTEDESC for MTE, !mtedesc => !mte_active. */ if (arm_tlb_mte_tagged(&info.page[0].attrs)) { mtedesc = 0; } if (fault == FAULT_FIRST) { /* Trapping mte check for the first-fault element. */ if (mtedesc) { mte_check(env, mtedesc, addr + mem_off, retaddr); } /* * Special handling of the first active element, * if it crosses a page boundary or is MMIO. */ bool is_split = mem_off == info.mem_off_split; if (unlikely(flags != 0) || unlikely(is_split)) { /* * Use the slow path for cross-page handling. * Might trap for MMIO or watchpoints. */ tlb_fn(env, vd, reg_off, addr + mem_off, retaddr); /* After any fault, zero the other elements. */ swap_memzero(vd, reg_off); reg_off += 1 << esz; mem_off += 1 << msz; swap_memzero(vd + reg_off, reg_max - reg_off); if (is_split) { goto second_page; } } else { memset(vd, 0, reg_max); } } else { memset(vd, 0, reg_max); if (unlikely(mem_off == info.mem_off_split)) { /* The first active element crosses a page boundary. */ flags |= info.page[1].flags; if (unlikely(flags & TLB_MMIO)) { /* Some page is MMIO, see below. */ goto do_fault; } if (unlikely(flags & TLB_WATCHPOINT) && (cpu_watchpoint_address_matches (env_cpu(env), addr + mem_off, 1 << msz) & BP_MEM_READ)) { /* Watchpoint hit, see below. */ goto do_fault; } if (mtedesc && !mte_probe(env, mtedesc, addr + mem_off)) { goto do_fault; } /* * Use the slow path for cross-page handling. * This is RAM, without a watchpoint, and will not trap. */ tlb_fn(env, vd, reg_off, addr + mem_off, retaddr); goto second_page; } } /* * From this point on, all memory operations are MemSingleNF. * * Per the MemSingleNF pseudocode, a no-fault load from Device memory * must not actually hit the bus -- it returns (UNKNOWN, FAULT) instead. * * Unfortuately we do not have access to the memory attributes from the * PTE to tell Device memory from Normal memory. So we make a mostly * correct check, and indicate (UNKNOWN, FAULT) for any MMIO. * This gives the right answer for the common cases of "Normal memory, * backed by host RAM" and "Device memory, backed by MMIO". * The architecture allows us to suppress an NF load and return * (UNKNOWN, FAULT) for any reason, so our behaviour for the corner * case of "Normal memory, backed by MMIO" is permitted. The case we * get wrong is "Device memory, backed by host RAM", for which we * should return (UNKNOWN, FAULT) for but do not. * * Similarly, CPU_BP breakpoints would raise exceptions, and so * return (UNKNOWN, FAULT). For simplicity, we consider gdb and * architectural breakpoints the same. */ if (unlikely(flags & TLB_MMIO)) { goto do_fault; } reg_last = info.reg_off_last[0]; host = info.page[0].host; do { uint64_t pg = *(uint64_t *)(vg + (reg_off >> 3)); do { if ((pg >> (reg_off & 63)) & 1) { if (unlikely(flags & TLB_WATCHPOINT) && (cpu_watchpoint_address_matches (env_cpu(env), addr + mem_off, 1 << msz) & BP_MEM_READ)) { goto do_fault; } if (mtedesc && !mte_probe(env, mtedesc, addr + mem_off)) { goto do_fault; } host_fn(vd, reg_off, host + mem_off); } reg_off += 1 << esz; mem_off += 1 << msz; } while (reg_off <= reg_last && (reg_off & 63)); } while (reg_off <= reg_last); /* * MemSingleNF is allowed to fail for any reason. We have special * code above to handle the first element crossing a page boundary. * As an implementation choice, decline to handle a cross-page element * in any other position. */ reg_off = info.reg_off_split; if (reg_off >= 0) { goto do_fault; } second_page: reg_off = info.reg_off_first[1]; if (likely(reg_off < 0)) { /* No active elements on the second page. All done. */ return; } /* * MemSingleNF is allowed to fail for any reason. As an implementation * choice, decline to handle elements on the second page. This should * be low frequency as the guest walks through memory -- the next * iteration of the guest's loop should be aligned on the page boundary, * and then all following iterations will stay aligned. */ do_fault: record_fault(env, reg_off, reg_max); } static inline QEMU_ALWAYS_INLINE void sve_ldnfff1_r_mte(CPUARMState *env, void *vg, target_ulong addr, uint32_t desc, const uintptr_t retaddr, const int esz, const int msz, const SVEContFault fault, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); int bit55 = extract64(addr, 55, 1); /* Remove mtedesc from the normal sve descriptor. */ desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* Perform gross MTE suppression early. */ if (!tbi_check(desc, bit55) || tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { mtedesc = 0; } sve_ldnfff1_r(env, vg, addr, desc, retaddr, mtedesc, esz, msz, fault, host_fn, tlb_fn); } #define DO_LDFF1_LDNF1_1(PART, ESZ) \ void HELPER(sve_ldff1##PART##_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MO_8, FAULT_FIRST, \ sve_ld1##PART##_host, sve_ld1##PART##_tlb); \ } \ void HELPER(sve_ldnf1##PART##_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MO_8, FAULT_NO, \ sve_ld1##PART##_host, sve_ld1##PART##_tlb); \ } \ void HELPER(sve_ldff1##PART##_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, FAULT_FIRST, \ sve_ld1##PART##_host, sve_ld1##PART##_tlb); \ } \ void HELPER(sve_ldnf1##PART##_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, FAULT_NO, \ sve_ld1##PART##_host, sve_ld1##PART##_tlb); \ } #define DO_LDFF1_LDNF1_2(PART, ESZ, MSZ) \ void HELPER(sve_ldff1##PART##_le_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_FIRST, \ sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb); \ } \ void HELPER(sve_ldnf1##PART##_le_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_NO, \ sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb); \ } \ void HELPER(sve_ldff1##PART##_be_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_FIRST, \ sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb); \ } \ void HELPER(sve_ldnf1##PART##_be_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r(env, vg, addr, desc, GETPC(), 0, ESZ, MSZ, FAULT_NO, \ sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb); \ } \ void HELPER(sve_ldff1##PART##_le_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_FIRST, \ sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb); \ } \ void HELPER(sve_ldnf1##PART##_le_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_NO, \ sve_ld1##PART##_le_host, sve_ld1##PART##_le_tlb); \ } \ void HELPER(sve_ldff1##PART##_be_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_FIRST, \ sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb); \ } \ void HELPER(sve_ldnf1##PART##_be_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_ldnfff1_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, FAULT_NO, \ sve_ld1##PART##_be_host, sve_ld1##PART##_be_tlb); \ } DO_LDFF1_LDNF1_1(bb, MO_8) DO_LDFF1_LDNF1_1(bhu, MO_16) DO_LDFF1_LDNF1_1(bhs, MO_16) DO_LDFF1_LDNF1_1(bsu, MO_32) DO_LDFF1_LDNF1_1(bss, MO_32) DO_LDFF1_LDNF1_1(bdu, MO_64) DO_LDFF1_LDNF1_1(bds, MO_64) DO_LDFF1_LDNF1_2(hh, MO_16, MO_16) DO_LDFF1_LDNF1_2(hsu, MO_32, MO_16) DO_LDFF1_LDNF1_2(hss, MO_32, MO_16) DO_LDFF1_LDNF1_2(hdu, MO_64, MO_16) DO_LDFF1_LDNF1_2(hds, MO_64, MO_16) DO_LDFF1_LDNF1_2(ss, MO_32, MO_32) DO_LDFF1_LDNF1_2(sdu, MO_64, MO_32) DO_LDFF1_LDNF1_2(sds, MO_64, MO_32) DO_LDFF1_LDNF1_2(dd, MO_64, MO_64) #undef DO_LDFF1_LDNF1_1 #undef DO_LDFF1_LDNF1_2 /* * Common helper for all contiguous 1,2,3,4-register predicated stores. */ static inline QEMU_ALWAYS_INLINE void sve_stN_r(CPUARMState *env, uint64_t *vg, target_ulong addr, uint32_t desc, const uintptr_t retaddr, const int esz, const int msz, const int N, uint32_t mtedesc, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { const unsigned rd = simd_data(desc); const intptr_t reg_max = simd_oprsz(desc); intptr_t reg_off, reg_last, mem_off; SVEContLdSt info; void *host; int i, flags; /* Find the active elements. */ if (!sve_cont_ldst_elements(&info, addr, vg, reg_max, esz, N << msz)) { /* The entire predicate was false; no store occurs. */ return; } /* Probe the page(s). Exit with exception for any invalid page. */ sve_cont_ldst_pages(&info, FAULT_ALL, env, addr, MMU_DATA_STORE, retaddr); /* Handle watchpoints for all active elements. */ sve_cont_ldst_watchpoints(&info, env, vg, addr, 1 << esz, N << msz, BP_MEM_WRITE, retaddr); /* * Handle mte checks for all active elements. * Since TBI must be set for MTE, !mtedesc => !mte_active. */ if (mtedesc) { sve_cont_ldst_mte_check(&info, env, vg, addr, 1 << esz, N << msz, mtedesc, retaddr); } flags = info.page[0].flags | info.page[1].flags; if (unlikely(flags != 0)) { #ifdef CONFIG_USER_ONLY g_assert_not_reached(); #else /* * At least one page includes MMIO. * Any bus operation can fail with cpu_transaction_failed, * which for ARM will raise SyncExternal. We cannot avoid * this fault and will leave with the store incomplete. */ mem_off = info.mem_off_first[0]; reg_off = info.reg_off_first[0]; reg_last = info.reg_off_last[1]; if (reg_last < 0) { reg_last = info.reg_off_split; if (reg_last < 0) { reg_last = info.reg_off_last[0]; } } do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { for (i = 0; i < N; ++i) { tlb_fn(env, &env->vfp.zregs[(rd + i) & 31], reg_off, addr + mem_off + (i << msz), retaddr); } } reg_off += 1 << esz; mem_off += N << msz; } while (reg_off & 63); } while (reg_off <= reg_last); return; #endif } mem_off = info.mem_off_first[0]; reg_off = info.reg_off_first[0]; reg_last = info.reg_off_last[0]; host = info.page[0].host; while (reg_off <= reg_last) { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { for (i = 0; i < N; ++i) { host_fn(&env->vfp.zregs[(rd + i) & 31], reg_off, host + mem_off + (i << msz)); } } reg_off += 1 << esz; mem_off += N << msz; } while (reg_off <= reg_last && (reg_off & 63)); } /* * Use the slow path to manage the cross-page misalignment. * But we know this is RAM and cannot trap. */ mem_off = info.mem_off_split; if (unlikely(mem_off >= 0)) { reg_off = info.reg_off_split; for (i = 0; i < N; ++i) { tlb_fn(env, &env->vfp.zregs[(rd + i) & 31], reg_off, addr + mem_off + (i << msz), retaddr); } } mem_off = info.mem_off_first[1]; if (unlikely(mem_off >= 0)) { reg_off = info.reg_off_first[1]; reg_last = info.reg_off_last[1]; host = info.page[1].host; do { uint64_t pg = vg[reg_off >> 6]; do { if ((pg >> (reg_off & 63)) & 1) { for (i = 0; i < N; ++i) { host_fn(&env->vfp.zregs[(rd + i) & 31], reg_off, host + mem_off + (i << msz)); } } reg_off += 1 << esz; mem_off += N << msz; } while (reg_off & 63); } while (reg_off <= reg_last); } } static inline QEMU_ALWAYS_INLINE void sve_stN_r_mte(CPUARMState *env, uint64_t *vg, target_ulong addr, uint32_t desc, const uintptr_t ra, const int esz, const int msz, const int N, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); int bit55 = extract64(addr, 55, 1); /* Remove mtedesc from the normal sve descriptor. */ desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* Perform gross MTE suppression early. */ if (!tbi_check(desc, bit55) || tcma_check(desc, bit55, allocation_tag_from_addr(addr))) { mtedesc = 0; } sve_stN_r(env, vg, addr, desc, ra, esz, msz, N, mtedesc, host_fn, tlb_fn); } #define DO_STN_1(N, NAME, ESZ) \ void HELPER(sve_st##N##NAME##_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MO_8, N, 0, \ sve_st1##NAME##_host, sve_st1##NAME##_tlb); \ } \ void HELPER(sve_st##N##NAME##_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_stN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MO_8, N, \ sve_st1##NAME##_host, sve_st1##NAME##_tlb); \ } #define DO_STN_2(N, NAME, ESZ, MSZ) \ void HELPER(sve_st##N##NAME##_le_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, N, 0, \ sve_st1##NAME##_le_host, sve_st1##NAME##_le_tlb); \ } \ void HELPER(sve_st##N##NAME##_be_r)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_stN_r(env, vg, addr, desc, GETPC(), ESZ, MSZ, N, 0, \ sve_st1##NAME##_be_host, sve_st1##NAME##_be_tlb); \ } \ void HELPER(sve_st##N##NAME##_le_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_stN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, N, \ sve_st1##NAME##_le_host, sve_st1##NAME##_le_tlb); \ } \ void HELPER(sve_st##N##NAME##_be_r_mte)(CPUARMState *env, void *vg, \ target_ulong addr, uint32_t desc) \ { \ sve_stN_r_mte(env, vg, addr, desc, GETPC(), ESZ, MSZ, N, \ sve_st1##NAME##_be_host, sve_st1##NAME##_be_tlb); \ } DO_STN_1(1, bb, MO_8) DO_STN_1(1, bh, MO_16) DO_STN_1(1, bs, MO_32) DO_STN_1(1, bd, MO_64) DO_STN_1(2, bb, MO_8) DO_STN_1(3, bb, MO_8) DO_STN_1(4, bb, MO_8) DO_STN_2(1, hh, MO_16, MO_16) DO_STN_2(1, hs, MO_32, MO_16) DO_STN_2(1, hd, MO_64, MO_16) DO_STN_2(2, hh, MO_16, MO_16) DO_STN_2(3, hh, MO_16, MO_16) DO_STN_2(4, hh, MO_16, MO_16) DO_STN_2(1, ss, MO_32, MO_32) DO_STN_2(1, sd, MO_64, MO_32) DO_STN_2(2, ss, MO_32, MO_32) DO_STN_2(3, ss, MO_32, MO_32) DO_STN_2(4, ss, MO_32, MO_32) DO_STN_2(1, dd, MO_64, MO_64) DO_STN_2(2, dd, MO_64, MO_64) DO_STN_2(3, dd, MO_64, MO_64) DO_STN_2(4, dd, MO_64, MO_64) #undef DO_STN_1 #undef DO_STN_2 /* * Loads with a vector index. */ /* * Load the element at @reg + @reg_ofs, sign or zero-extend as needed. */ typedef target_ulong zreg_off_fn(void *reg, intptr_t reg_ofs); static target_ulong off_zsu_s(void *reg, intptr_t reg_ofs) { return *(uint32_t *)(reg + H1_4(reg_ofs)); } static target_ulong off_zss_s(void *reg, intptr_t reg_ofs) { return *(int32_t *)(reg + H1_4(reg_ofs)); } static target_ulong off_zsu_d(void *reg, intptr_t reg_ofs) { return (uint32_t)*(uint64_t *)(reg + reg_ofs); } static target_ulong off_zss_d(void *reg, intptr_t reg_ofs) { return (int32_t)*(uint64_t *)(reg + reg_ofs); } static target_ulong off_zd_d(void *reg, intptr_t reg_ofs) { return *(uint64_t *)(reg + reg_ofs); } static inline QEMU_ALWAYS_INLINE void sve_ld1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm, target_ulong base, uint32_t desc, uintptr_t retaddr, uint32_t mtedesc, int esize, int msize, zreg_off_fn *off_fn, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { const int mmu_idx = cpu_mmu_index(env, false); const intptr_t reg_max = simd_oprsz(desc); const int scale = simd_data(desc); ARMVectorReg scratch; intptr_t reg_off; SVEHostPage info, info2; memset(&scratch, 0, reg_max); reg_off = 0; do { uint64_t pg = vg[reg_off >> 6]; do { if (likely(pg & 1)) { target_ulong addr = base + (off_fn(vm, reg_off) << scale); target_ulong in_page = -(addr | TARGET_PAGE_MASK); sve_probe_page(&info, false, env, addr, 0, MMU_DATA_LOAD, mmu_idx, retaddr); if (likely(in_page >= msize)) { if (unlikely(info.flags & TLB_WATCHPOINT)) { cpu_check_watchpoint(env_cpu(env), addr, msize, info.attrs, BP_MEM_READ, retaddr); } if (mtedesc && arm_tlb_mte_tagged(&info.attrs)) { mte_check(env, mtedesc, addr, retaddr); } if (unlikely(info.flags & TLB_MMIO)) { tlb_fn(env, &scratch, reg_off, addr, retaddr); } else { host_fn(&scratch, reg_off, info.host); } } else { /* Element crosses the page boundary. */ sve_probe_page(&info2, false, env, addr + in_page, 0, MMU_DATA_LOAD, mmu_idx, retaddr); if (unlikely((info.flags | info2.flags) & TLB_WATCHPOINT)) { cpu_check_watchpoint(env_cpu(env), addr, msize, info.attrs, BP_MEM_READ, retaddr); } if (mtedesc && arm_tlb_mte_tagged(&info.attrs)) { mte_check(env, mtedesc, addr, retaddr); } tlb_fn(env, &scratch, reg_off, addr, retaddr); } } reg_off += esize; pg >>= esize; } while (reg_off & 63); } while (reg_off < reg_max); /* Wait until all exceptions have been raised to write back. */ memcpy(vd, &scratch, reg_max); } static inline QEMU_ALWAYS_INLINE void sve_ld1_z_mte(CPUARMState *env, void *vd, uint64_t *vg, void *vm, target_ulong base, uint32_t desc, uintptr_t retaddr, int esize, int msize, zreg_off_fn *off_fn, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* Remove mtedesc from the normal sve descriptor. */ desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* * ??? TODO: For the 32-bit offset extractions, base + ofs cannot * offset base entirely over the address space hole to change the * pointer tag, or change the bit55 selector. So we could here * examine TBI + TCMA like we do for sve_ldN_r_mte(). */ sve_ld1_z(env, vd, vg, vm, base, desc, retaddr, mtedesc, esize, msize, off_fn, host_fn, tlb_fn); } #define DO_LD1_ZPZ_S(MEM, OFS, MSZ) \ void HELPER(sve_ld##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ld1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 4, 1 << MSZ, \ off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } \ void HELPER(sve_ld##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ld1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 4, 1 << MSZ, \ off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } #define DO_LD1_ZPZ_D(MEM, OFS, MSZ) \ void HELPER(sve_ld##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ld1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 8, 1 << MSZ, \ off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } \ void HELPER(sve_ld##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ld1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 8, 1 << MSZ, \ off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } DO_LD1_ZPZ_S(bsu, zsu, MO_8) DO_LD1_ZPZ_S(bsu, zss, MO_8) DO_LD1_ZPZ_D(bdu, zsu, MO_8) DO_LD1_ZPZ_D(bdu, zss, MO_8) DO_LD1_ZPZ_D(bdu, zd, MO_8) DO_LD1_ZPZ_S(bss, zsu, MO_8) DO_LD1_ZPZ_S(bss, zss, MO_8) DO_LD1_ZPZ_D(bds, zsu, MO_8) DO_LD1_ZPZ_D(bds, zss, MO_8) DO_LD1_ZPZ_D(bds, zd, MO_8) DO_LD1_ZPZ_S(hsu_le, zsu, MO_16) DO_LD1_ZPZ_S(hsu_le, zss, MO_16) DO_LD1_ZPZ_D(hdu_le, zsu, MO_16) DO_LD1_ZPZ_D(hdu_le, zss, MO_16) DO_LD1_ZPZ_D(hdu_le, zd, MO_16) DO_LD1_ZPZ_S(hsu_be, zsu, MO_16) DO_LD1_ZPZ_S(hsu_be, zss, MO_16) DO_LD1_ZPZ_D(hdu_be, zsu, MO_16) DO_LD1_ZPZ_D(hdu_be, zss, MO_16) DO_LD1_ZPZ_D(hdu_be, zd, MO_16) DO_LD1_ZPZ_S(hss_le, zsu, MO_16) DO_LD1_ZPZ_S(hss_le, zss, MO_16) DO_LD1_ZPZ_D(hds_le, zsu, MO_16) DO_LD1_ZPZ_D(hds_le, zss, MO_16) DO_LD1_ZPZ_D(hds_le, zd, MO_16) DO_LD1_ZPZ_S(hss_be, zsu, MO_16) DO_LD1_ZPZ_S(hss_be, zss, MO_16) DO_LD1_ZPZ_D(hds_be, zsu, MO_16) DO_LD1_ZPZ_D(hds_be, zss, MO_16) DO_LD1_ZPZ_D(hds_be, zd, MO_16) DO_LD1_ZPZ_S(ss_le, zsu, MO_32) DO_LD1_ZPZ_S(ss_le, zss, MO_32) DO_LD1_ZPZ_D(sdu_le, zsu, MO_32) DO_LD1_ZPZ_D(sdu_le, zss, MO_32) DO_LD1_ZPZ_D(sdu_le, zd, MO_32) DO_LD1_ZPZ_S(ss_be, zsu, MO_32) DO_LD1_ZPZ_S(ss_be, zss, MO_32) DO_LD1_ZPZ_D(sdu_be, zsu, MO_32) DO_LD1_ZPZ_D(sdu_be, zss, MO_32) DO_LD1_ZPZ_D(sdu_be, zd, MO_32) DO_LD1_ZPZ_D(sds_le, zsu, MO_32) DO_LD1_ZPZ_D(sds_le, zss, MO_32) DO_LD1_ZPZ_D(sds_le, zd, MO_32) DO_LD1_ZPZ_D(sds_be, zsu, MO_32) DO_LD1_ZPZ_D(sds_be, zss, MO_32) DO_LD1_ZPZ_D(sds_be, zd, MO_32) DO_LD1_ZPZ_D(dd_le, zsu, MO_64) DO_LD1_ZPZ_D(dd_le, zss, MO_64) DO_LD1_ZPZ_D(dd_le, zd, MO_64) DO_LD1_ZPZ_D(dd_be, zsu, MO_64) DO_LD1_ZPZ_D(dd_be, zss, MO_64) DO_LD1_ZPZ_D(dd_be, zd, MO_64) #undef DO_LD1_ZPZ_S #undef DO_LD1_ZPZ_D /* First fault loads with a vector index. */ /* * Common helpers for all gather first-faulting loads. */ static inline QEMU_ALWAYS_INLINE void sve_ldff1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm, target_ulong base, uint32_t desc, uintptr_t retaddr, uint32_t mtedesc, const int esz, const int msz, zreg_off_fn *off_fn, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { const int mmu_idx = cpu_mmu_index(env, false); const intptr_t reg_max = simd_oprsz(desc); const int scale = simd_data(desc); const int esize = 1 << esz; const int msize = 1 << msz; intptr_t reg_off; SVEHostPage info; target_ulong addr, in_page; /* Skip to the first true predicate. */ reg_off = find_next_active(vg, 0, reg_max, esz); if (unlikely(reg_off >= reg_max)) { /* The entire predicate was false; no load occurs. */ memset(vd, 0, reg_max); return; } /* * Probe the first element, allowing faults. */ addr = base + (off_fn(vm, reg_off) << scale); if (mtedesc) { mte_check(env, mtedesc, addr, retaddr); } tlb_fn(env, vd, reg_off, addr, retaddr); /* After any fault, zero the other elements. */ swap_memzero(vd, reg_off); reg_off += esize; swap_memzero(vd + reg_off, reg_max - reg_off); /* * Probe the remaining elements, not allowing faults. */ while (reg_off < reg_max) { uint64_t pg = vg[reg_off >> 6]; do { if (likely((pg >> (reg_off & 63)) & 1)) { addr = base + (off_fn(vm, reg_off) << scale); in_page = -(addr | TARGET_PAGE_MASK); if (unlikely(in_page < msize)) { /* Stop if the element crosses a page boundary. */ goto fault; } sve_probe_page(&info, true, env, addr, 0, MMU_DATA_LOAD, mmu_idx, retaddr); if (unlikely(info.flags & (TLB_INVALID_MASK | TLB_MMIO))) { goto fault; } if (unlikely(info.flags & TLB_WATCHPOINT) && (cpu_watchpoint_address_matches (env_cpu(env), addr, msize) & BP_MEM_READ)) { goto fault; } if (mtedesc && arm_tlb_mte_tagged(&info.attrs) && !mte_probe(env, mtedesc, addr)) { goto fault; } host_fn(vd, reg_off, info.host); } reg_off += esize; } while (reg_off & 63); } return; fault: record_fault(env, reg_off, reg_max); } static inline QEMU_ALWAYS_INLINE void sve_ldff1_z_mte(CPUARMState *env, void *vd, uint64_t *vg, void *vm, target_ulong base, uint32_t desc, uintptr_t retaddr, const int esz, const int msz, zreg_off_fn *off_fn, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* Remove mtedesc from the normal sve descriptor. */ desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* * ??? TODO: For the 32-bit offset extractions, base + ofs cannot * offset base entirely over the address space hole to change the * pointer tag, or change the bit55 selector. So we could here * examine TBI + TCMA like we do for sve_ldN_r_mte(). */ sve_ldff1_z(env, vd, vg, vm, base, desc, retaddr, mtedesc, esz, msz, off_fn, host_fn, tlb_fn); } #define DO_LDFF1_ZPZ_S(MEM, OFS, MSZ) \ void HELPER(sve_ldff##MEM##_##OFS) \ (CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ldff1_z(env, vd, vg, vm, base, desc, GETPC(), 0, MO_32, MSZ, \ off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } \ void HELPER(sve_ldff##MEM##_##OFS##_mte) \ (CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ldff1_z_mte(env, vd, vg, vm, base, desc, GETPC(), MO_32, MSZ, \ off_##OFS##_s, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } #define DO_LDFF1_ZPZ_D(MEM, OFS, MSZ) \ void HELPER(sve_ldff##MEM##_##OFS) \ (CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ldff1_z(env, vd, vg, vm, base, desc, GETPC(), 0, MO_64, MSZ, \ off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } \ void HELPER(sve_ldff##MEM##_##OFS##_mte) \ (CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_ldff1_z_mte(env, vd, vg, vm, base, desc, GETPC(), MO_64, MSZ, \ off_##OFS##_d, sve_ld1##MEM##_host, sve_ld1##MEM##_tlb); \ } DO_LDFF1_ZPZ_S(bsu, zsu, MO_8) DO_LDFF1_ZPZ_S(bsu, zss, MO_8) DO_LDFF1_ZPZ_D(bdu, zsu, MO_8) DO_LDFF1_ZPZ_D(bdu, zss, MO_8) DO_LDFF1_ZPZ_D(bdu, zd, MO_8) DO_LDFF1_ZPZ_S(bss, zsu, MO_8) DO_LDFF1_ZPZ_S(bss, zss, MO_8) DO_LDFF1_ZPZ_D(bds, zsu, MO_8) DO_LDFF1_ZPZ_D(bds, zss, MO_8) DO_LDFF1_ZPZ_D(bds, zd, MO_8) DO_LDFF1_ZPZ_S(hsu_le, zsu, MO_16) DO_LDFF1_ZPZ_S(hsu_le, zss, MO_16) DO_LDFF1_ZPZ_D(hdu_le, zsu, MO_16) DO_LDFF1_ZPZ_D(hdu_le, zss, MO_16) DO_LDFF1_ZPZ_D(hdu_le, zd, MO_16) DO_LDFF1_ZPZ_S(hsu_be, zsu, MO_16) DO_LDFF1_ZPZ_S(hsu_be, zss, MO_16) DO_LDFF1_ZPZ_D(hdu_be, zsu, MO_16) DO_LDFF1_ZPZ_D(hdu_be, zss, MO_16) DO_LDFF1_ZPZ_D(hdu_be, zd, MO_16) DO_LDFF1_ZPZ_S(hss_le, zsu, MO_16) DO_LDFF1_ZPZ_S(hss_le, zss, MO_16) DO_LDFF1_ZPZ_D(hds_le, zsu, MO_16) DO_LDFF1_ZPZ_D(hds_le, zss, MO_16) DO_LDFF1_ZPZ_D(hds_le, zd, MO_16) DO_LDFF1_ZPZ_S(hss_be, zsu, MO_16) DO_LDFF1_ZPZ_S(hss_be, zss, MO_16) DO_LDFF1_ZPZ_D(hds_be, zsu, MO_16) DO_LDFF1_ZPZ_D(hds_be, zss, MO_16) DO_LDFF1_ZPZ_D(hds_be, zd, MO_16) DO_LDFF1_ZPZ_S(ss_le, zsu, MO_32) DO_LDFF1_ZPZ_S(ss_le, zss, MO_32) DO_LDFF1_ZPZ_D(sdu_le, zsu, MO_32) DO_LDFF1_ZPZ_D(sdu_le, zss, MO_32) DO_LDFF1_ZPZ_D(sdu_le, zd, MO_32) DO_LDFF1_ZPZ_S(ss_be, zsu, MO_32) DO_LDFF1_ZPZ_S(ss_be, zss, MO_32) DO_LDFF1_ZPZ_D(sdu_be, zsu, MO_32) DO_LDFF1_ZPZ_D(sdu_be, zss, MO_32) DO_LDFF1_ZPZ_D(sdu_be, zd, MO_32) DO_LDFF1_ZPZ_D(sds_le, zsu, MO_32) DO_LDFF1_ZPZ_D(sds_le, zss, MO_32) DO_LDFF1_ZPZ_D(sds_le, zd, MO_32) DO_LDFF1_ZPZ_D(sds_be, zsu, MO_32) DO_LDFF1_ZPZ_D(sds_be, zss, MO_32) DO_LDFF1_ZPZ_D(sds_be, zd, MO_32) DO_LDFF1_ZPZ_D(dd_le, zsu, MO_64) DO_LDFF1_ZPZ_D(dd_le, zss, MO_64) DO_LDFF1_ZPZ_D(dd_le, zd, MO_64) DO_LDFF1_ZPZ_D(dd_be, zsu, MO_64) DO_LDFF1_ZPZ_D(dd_be, zss, MO_64) DO_LDFF1_ZPZ_D(dd_be, zd, MO_64) /* Stores with a vector index. */ static inline QEMU_ALWAYS_INLINE void sve_st1_z(CPUARMState *env, void *vd, uint64_t *vg, void *vm, target_ulong base, uint32_t desc, uintptr_t retaddr, uint32_t mtedesc, int esize, int msize, zreg_off_fn *off_fn, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { const int mmu_idx = cpu_mmu_index(env, false); const intptr_t reg_max = simd_oprsz(desc); const int scale = simd_data(desc); void *host[ARM_MAX_VQ * 4]; intptr_t reg_off, i; SVEHostPage info, info2; /* * Probe all of the elements for host addresses and flags. */ i = reg_off = 0; do { uint64_t pg = vg[reg_off >> 6]; do { target_ulong addr = base + (off_fn(vm, reg_off) << scale); target_ulong in_page = -(addr | TARGET_PAGE_MASK); host[i] = NULL; if (likely((pg >> (reg_off & 63)) & 1)) { if (likely(in_page >= msize)) { sve_probe_page(&info, false, env, addr, 0, MMU_DATA_STORE, mmu_idx, retaddr); if (!(info.flags & TLB_MMIO)) { host[i] = info.host; } } else { /* * Element crosses the page boundary. * Probe both pages, but do not record the host address, * so that we use the slow path. */ sve_probe_page(&info, false, env, addr, 0, MMU_DATA_STORE, mmu_idx, retaddr); sve_probe_page(&info2, false, env, addr + in_page, 0, MMU_DATA_STORE, mmu_idx, retaddr); info.flags |= info2.flags; } if (unlikely(info.flags & TLB_WATCHPOINT)) { cpu_check_watchpoint(env_cpu(env), addr, msize, info.attrs, BP_MEM_WRITE, retaddr); } if (mtedesc && arm_tlb_mte_tagged(&info.attrs)) { mte_check(env, mtedesc, addr, retaddr); } } i += 1; reg_off += esize; } while (reg_off & 63); } while (reg_off < reg_max); /* * Now that we have recognized all exceptions except SyncExternal * (from TLB_MMIO), which we cannot avoid, perform all of the stores. * * Note for the common case of an element in RAM, not crossing a page * boundary, we have stored the host address in host[]. This doubles * as a first-level check against the predicate, since only enabled * elements have non-null host addresses. */ i = reg_off = 0; do { void *h = host[i]; if (likely(h != NULL)) { host_fn(vd, reg_off, h); } else if ((vg[reg_off >> 6] >> (reg_off & 63)) & 1) { target_ulong addr = base + (off_fn(vm, reg_off) << scale); tlb_fn(env, vd, reg_off, addr, retaddr); } i += 1; reg_off += esize; } while (reg_off < reg_max); } static inline QEMU_ALWAYS_INLINE void sve_st1_z_mte(CPUARMState *env, void *vd, uint64_t *vg, void *vm, target_ulong base, uint32_t desc, uintptr_t retaddr, int esize, int msize, zreg_off_fn *off_fn, sve_ldst1_host_fn *host_fn, sve_ldst1_tlb_fn *tlb_fn) { uint32_t mtedesc = desc >> (SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* Remove mtedesc from the normal sve descriptor. */ desc = extract32(desc, 0, SIMD_DATA_SHIFT + SVE_MTEDESC_SHIFT); /* * ??? TODO: For the 32-bit offset extractions, base + ofs cannot * offset base entirely over the address space hole to change the * pointer tag, or change the bit55 selector. So we could here * examine TBI + TCMA like we do for sve_ldN_r_mte(). */ sve_st1_z(env, vd, vg, vm, base, desc, retaddr, mtedesc, esize, msize, off_fn, host_fn, tlb_fn); } #define DO_ST1_ZPZ_S(MEM, OFS, MSZ) \ void HELPER(sve_st##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_st1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 4, 1 << MSZ, \ off_##OFS##_s, sve_st1##MEM##_host, sve_st1##MEM##_tlb); \ } \ void HELPER(sve_st##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_st1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 4, 1 << MSZ, \ off_##OFS##_s, sve_st1##MEM##_host, sve_st1##MEM##_tlb); \ } #define DO_ST1_ZPZ_D(MEM, OFS, MSZ) \ void HELPER(sve_st##MEM##_##OFS)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_st1_z(env, vd, vg, vm, base, desc, GETPC(), 0, 8, 1 << MSZ, \ off_##OFS##_d, sve_st1##MEM##_host, sve_st1##MEM##_tlb); \ } \ void HELPER(sve_st##MEM##_##OFS##_mte)(CPUARMState *env, void *vd, void *vg, \ void *vm, target_ulong base, uint32_t desc) \ { \ sve_st1_z_mte(env, vd, vg, vm, base, desc, GETPC(), 8, 1 << MSZ, \ off_##OFS##_d, sve_st1##MEM##_host, sve_st1##MEM##_tlb); \ } DO_ST1_ZPZ_S(bs, zsu, MO_8) DO_ST1_ZPZ_S(hs_le, zsu, MO_16) DO_ST1_ZPZ_S(hs_be, zsu, MO_16) DO_ST1_ZPZ_S(ss_le, zsu, MO_32) DO_ST1_ZPZ_S(ss_be, zsu, MO_32) DO_ST1_ZPZ_S(bs, zss, MO_8) DO_ST1_ZPZ_S(hs_le, zss, MO_16) DO_ST1_ZPZ_S(hs_be, zss, MO_16) DO_ST1_ZPZ_S(ss_le, zss, MO_32) DO_ST1_ZPZ_S(ss_be, zss, MO_32) DO_ST1_ZPZ_D(bd, zsu, MO_8) DO_ST1_ZPZ_D(hd_le, zsu, MO_16) DO_ST1_ZPZ_D(hd_be, zsu, MO_16) DO_ST1_ZPZ_D(sd_le, zsu, MO_32) DO_ST1_ZPZ_D(sd_be, zsu, MO_32) DO_ST1_ZPZ_D(dd_le, zsu, MO_64) DO_ST1_ZPZ_D(dd_be, zsu, MO_64) DO_ST1_ZPZ_D(bd, zss, MO_8) DO_ST1_ZPZ_D(hd_le, zss, MO_16) DO_ST1_ZPZ_D(hd_be, zss, MO_16) DO_ST1_ZPZ_D(sd_le, zss, MO_32) DO_ST1_ZPZ_D(sd_be, zss, MO_32) DO_ST1_ZPZ_D(dd_le, zss, MO_64) DO_ST1_ZPZ_D(dd_be, zss, MO_64) DO_ST1_ZPZ_D(bd, zd, MO_8) DO_ST1_ZPZ_D(hd_le, zd, MO_16) DO_ST1_ZPZ_D(hd_be, zd, MO_16) DO_ST1_ZPZ_D(sd_le, zd, MO_32) DO_ST1_ZPZ_D(sd_be, zd, MO_32) DO_ST1_ZPZ_D(dd_le, zd, MO_64) DO_ST1_ZPZ_D(dd_be, zd, MO_64) #undef DO_ST1_ZPZ_S #undef DO_ST1_ZPZ_D void HELPER(sve2_eor3)(void *vd, void *vn, void *vm, void *vk, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm, *k = vk; for (i = 0; i < opr_sz; ++i) { d[i] = n[i] ^ m[i] ^ k[i]; } } void HELPER(sve2_bcax)(void *vd, void *vn, void *vm, void *vk, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm, *k = vk; for (i = 0; i < opr_sz; ++i) { d[i] = n[i] ^ (m[i] & ~k[i]); } } void HELPER(sve2_bsl1n)(void *vd, void *vn, void *vm, void *vk, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm, *k = vk; for (i = 0; i < opr_sz; ++i) { d[i] = (~n[i] & k[i]) | (m[i] & ~k[i]); } } void HELPER(sve2_bsl2n)(void *vd, void *vn, void *vm, void *vk, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm, *k = vk; for (i = 0; i < opr_sz; ++i) { d[i] = (n[i] & k[i]) | (~m[i] & ~k[i]); } } void HELPER(sve2_nbsl)(void *vd, void *vn, void *vm, void *vk, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; uint64_t *d = vd, *n = vn, *m = vm, *k = vk; for (i = 0; i < opr_sz; ++i) { d[i] = ~((n[i] & k[i]) | (m[i] & ~k[i])); } } /* * Returns true if m0 or m1 contains the low uint8_t/uint16_t in n. * See hasless(v,1) from * https://graphics.stanford.edu/~seander/bithacks.html#ZeroInWord */ static inline bool do_match2(uint64_t n, uint64_t m0, uint64_t m1, int esz) { int bits = 8 << esz; uint64_t ones = dup_const(esz, 1); uint64_t signs = ones << (bits - 1); uint64_t cmp0, cmp1; cmp1 = dup_const(esz, n); cmp0 = cmp1 ^ m0; cmp1 = cmp1 ^ m1; cmp0 = (cmp0 - ones) & ~cmp0; cmp1 = (cmp1 - ones) & ~cmp1; return (cmp0 | cmp1) & signs; } static inline uint32_t do_match(void *vd, void *vn, void *vm, void *vg, uint32_t desc, int esz, bool nmatch) { uint16_t esz_mask = pred_esz_masks[esz]; intptr_t opr_sz = simd_oprsz(desc); uint32_t flags = PREDTEST_INIT; intptr_t i, j, k; for (i = 0; i < opr_sz; i += 16) { uint64_t m0 = *(uint64_t *)(vm + i); uint64_t m1 = *(uint64_t *)(vm + i + 8); uint16_t pg = *(uint16_t *)(vg + H1_2(i >> 3)) & esz_mask; uint16_t out = 0; for (j = 0; j < 16; j += 8) { uint64_t n = *(uint64_t *)(vn + i + j); for (k = 0; k < 8; k += 1 << esz) { if (pg & (1 << (j + k))) { bool o = do_match2(n >> (k * 8), m0, m1, esz); out |= (o ^ nmatch) << (j + k); } } } *(uint16_t *)(vd + H1_2(i >> 3)) = out; flags = iter_predtest_fwd(out, pg, flags); } return flags; } #define DO_PPZZ_MATCH(NAME, ESZ, INV) \ uint32_t HELPER(NAME)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) \ { \ return do_match(vd, vn, vm, vg, desc, ESZ, INV); \ } DO_PPZZ_MATCH(sve2_match_ppzz_b, MO_8, false) DO_PPZZ_MATCH(sve2_match_ppzz_h, MO_16, false) DO_PPZZ_MATCH(sve2_nmatch_ppzz_b, MO_8, true) DO_PPZZ_MATCH(sve2_nmatch_ppzz_h, MO_16, true) #undef DO_PPZZ_MATCH void HELPER(sve2_histcnt_s)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { ARMVectorReg scratch; intptr_t i, j; intptr_t opr_sz = simd_oprsz(desc); uint32_t *d = vd, *n = vn, *m = vm; uint8_t *pg = vg; if (d == n) { n = memcpy(&scratch, n, opr_sz); if (d == m) { m = n; } } else if (d == m) { m = memcpy(&scratch, m, opr_sz); } for (i = 0; i < opr_sz; i += 4) { uint64_t count = 0; uint8_t pred; pred = pg[H1(i >> 3)] >> (i & 7); if (pred & 1) { uint32_t nn = n[H4(i >> 2)]; for (j = 0; j <= i; j += 4) { pred = pg[H1(j >> 3)] >> (j & 7); if ((pred & 1) && nn == m[H4(j >> 2)]) { ++count; } } } d[H4(i >> 2)] = count; } } void HELPER(sve2_histcnt_d)(void *vd, void *vn, void *vm, void *vg, uint32_t desc) { ARMVectorReg scratch; intptr_t i, j; intptr_t opr_sz = simd_oprsz(desc); uint64_t *d = vd, *n = vn, *m = vm; uint8_t *pg = vg; if (d == n) { n = memcpy(&scratch, n, opr_sz); if (d == m) { m = n; } } else if (d == m) { m = memcpy(&scratch, m, opr_sz); } for (i = 0; i < opr_sz / 8; ++i) { uint64_t count = 0; if (pg[H1(i)] & 1) { uint64_t nn = n[i]; for (j = 0; j <= i; ++j) { if ((pg[H1(j)] & 1) && nn == m[j]) { ++count; } } } d[i] = count; } } /* * Returns the number of bytes in m0 and m1 that match n. * Unlike do_match2 we don't just need true/false, we need an exact count. * This requires two extra logical operations. */ static inline uint64_t do_histseg_cnt(uint8_t n, uint64_t m0, uint64_t m1) { const uint64_t mask = dup_const(MO_8, 0x7f); uint64_t cmp0, cmp1; cmp1 = dup_const(MO_8, n); cmp0 = cmp1 ^ m0; cmp1 = cmp1 ^ m1; /* * 1: clear msb of each byte to avoid carry to next byte (& mask) * 2: carry in to msb if byte != 0 (+ mask) * 3: set msb if cmp has msb set (| cmp) * 4: set ~msb to ignore them (| mask) * We now have 0xff for byte != 0 or 0x7f for byte == 0. * 5: invert, resulting in 0x80 if and only if byte == 0. */ cmp0 = ~(((cmp0 & mask) + mask) | cmp0 | mask); cmp1 = ~(((cmp1 & mask) + mask) | cmp1 | mask); /* * Combine the two compares in a way that the bits do * not overlap, and so preserves the count of set bits. * If the host has an efficient instruction for ctpop, * then ctpop(x) + ctpop(y) has the same number of * operations as ctpop(x | (y >> 1)). If the host does * not have an efficient ctpop, then we only want to * use it once. */ return ctpop64(cmp0 | (cmp1 >> 1)); } void HELPER(sve2_histseg)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, j; intptr_t opr_sz = simd_oprsz(desc); for (i = 0; i < opr_sz; i += 16) { uint64_t n0 = *(uint64_t *)(vn + i); uint64_t m0 = *(uint64_t *)(vm + i); uint64_t n1 = *(uint64_t *)(vn + i + 8); uint64_t m1 = *(uint64_t *)(vm + i + 8); uint64_t out0 = 0; uint64_t out1 = 0; for (j = 0; j < 64; j += 8) { uint64_t cnt0 = do_histseg_cnt(n0 >> j, m0, m1); uint64_t cnt1 = do_histseg_cnt(n1 >> j, m0, m1); out0 |= cnt0 << j; out1 |= cnt1 << j; } *(uint64_t *)(vd + i) = out0; *(uint64_t *)(vd + i + 8) = out1; } } void HELPER(sve2_xar_b)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; int shr = simd_data(desc); int shl = 8 - shr; uint64_t mask = dup_const(MO_8, 0xff >> shr); uint64_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; ++i) { uint64_t t = n[i] ^ m[i]; d[i] = ((t >> shr) & mask) | ((t << shl) & ~mask); } } void HELPER(sve2_xar_h)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 8; int shr = simd_data(desc); int shl = 16 - shr; uint64_t mask = dup_const(MO_16, 0xffff >> shr); uint64_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; ++i) { uint64_t t = n[i] ^ m[i]; d[i] = ((t >> shr) & mask) | ((t << shl) & ~mask); } } void HELPER(sve2_xar_s)(void *vd, void *vn, void *vm, uint32_t desc) { intptr_t i, opr_sz = simd_oprsz(desc) / 4; int shr = simd_data(desc); uint32_t *d = vd, *n = vn, *m = vm; for (i = 0; i < opr_sz; ++i) { d[i] = ror32(n[i] ^ m[i], shr); } } void HELPER(fmmla_s)(void *vd, void *vn, void *vm, void *va, void *status, uint32_t desc) { intptr_t s, opr_sz = simd_oprsz(desc) / (sizeof(float32) * 4); for (s = 0; s < opr_sz; ++s) { float32 *n = vn + s * sizeof(float32) * 4; float32 *m = vm + s * sizeof(float32) * 4; float32 *a = va + s * sizeof(float32) * 4; float32 *d = vd + s * sizeof(float32) * 4; float32 n00 = n[H4(0)], n01 = n[H4(1)]; float32 n10 = n[H4(2)], n11 = n[H4(3)]; float32 m00 = m[H4(0)], m01 = m[H4(1)]; float32 m10 = m[H4(2)], m11 = m[H4(3)]; float32 p0, p1; /* i = 0, j = 0 */ p0 = float32_mul(n00, m00, status); p1 = float32_mul(n01, m01, status); d[H4(0)] = float32_add(a[H4(0)], float32_add(p0, p1, status), status); /* i = 0, j = 1 */ p0 = float32_mul(n00, m10, status); p1 = float32_mul(n01, m11, status); d[H4(1)] = float32_add(a[H4(1)], float32_add(p0, p1, status), status); /* i = 1, j = 0 */ p0 = float32_mul(n10, m00, status); p1 = float32_mul(n11, m01, status); d[H4(2)] = float32_add(a[H4(2)], float32_add(p0, p1, status), status); /* i = 1, j = 1 */ p0 = float32_mul(n10, m10, status); p1 = float32_mul(n11, m11, status); d[H4(3)] = float32_add(a[H4(3)], float32_add(p0, p1, status), status); } } void HELPER(fmmla_d)(void *vd, void *vn, void *vm, void *va, void *status, uint32_t desc) { intptr_t s, opr_sz = simd_oprsz(desc) / (sizeof(float64) * 4); for (s = 0; s < opr_sz; ++s) { float64 *n = vn + s * sizeof(float64) * 4; float64 *m = vm + s * sizeof(float64) * 4; float64 *a = va + s * sizeof(float64) * 4; float64 *d = vd + s * sizeof(float64) * 4; float64 n00 = n[0], n01 = n[1], n10 = n[2], n11 = n[3]; float64 m00 = m[0], m01 = m[1], m10 = m[2], m11 = m[3]; float64 p0, p1; /* i = 0, j = 0 */ p0 = float64_mul(n00, m00, status); p1 = float64_mul(n01, m01, status); d[0] = float64_add(a[0], float64_add(p0, p1, status), status); /* i = 0, j = 1 */ p0 = float64_mul(n00, m10, status); p1 = float64_mul(n01, m11, status); d[1] = float64_add(a[1], float64_add(p0, p1, status), status); /* i = 1, j = 0 */ p0 = float64_mul(n10, m00, status); p1 = float64_mul(n11, m01, status); d[2] = float64_add(a[2], float64_add(p0, p1, status), status); /* i = 1, j = 1 */ p0 = float64_mul(n10, m10, status); p1 = float64_mul(n11, m11, status); d[3] = float64_add(a[3], float64_add(p0, p1, status), status); } } #define DO_FCVTNT(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc); \ uint64_t *g = vg; \ do { \ uint64_t pg = g[(i - 1) >> 6]; \ do { \ i -= sizeof(TYPEW); \ if (likely((pg >> (i & 63)) & 1)) { \ TYPEW nn = *(TYPEW *)(vn + HW(i)); \ *(TYPEN *)(vd + HN(i + sizeof(TYPEN))) = OP(nn, status); \ } \ } while (i & 63); \ } while (i != 0); \ } DO_FCVTNT(sve_bfcvtnt, uint32_t, uint16_t, H1_4, H1_2, float32_to_bfloat16) DO_FCVTNT(sve2_fcvtnt_sh, uint32_t, uint16_t, H1_4, H1_2, sve_f32_to_f16) DO_FCVTNT(sve2_fcvtnt_ds, uint64_t, uint32_t, H1_8, H1_4, float64_to_float32) #define DO_FCVTLT(NAME, TYPEW, TYPEN, HW, HN, OP) \ void HELPER(NAME)(void *vd, void *vn, void *vg, void *status, uint32_t desc) \ { \ intptr_t i = simd_oprsz(desc); \ uint64_t *g = vg; \ do { \ uint64_t pg = g[(i - 1) >> 6]; \ do { \ i -= sizeof(TYPEW); \ if (likely((pg >> (i & 63)) & 1)) { \ TYPEN nn = *(TYPEN *)(vn + HN(i + sizeof(TYPEN))); \ *(TYPEW *)(vd + HW(i)) = OP(nn, status); \ } \ } while (i & 63); \ } while (i != 0); \ } DO_FCVTLT(sve2_fcvtlt_hs, uint32_t, uint16_t, H1_4, H1_2, sve_f16_to_f32) DO_FCVTLT(sve2_fcvtlt_sd, uint64_t, uint32_t, H1_8, H1_4, float32_to_float64) #undef DO_FCVTLT #undef DO_FCVTNT