[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PATCH v2 21/37] target/riscv: 64-bit Add/Subtract Instructions
From: |
LIU Zhiwei |
Subject: |
[PATCH v2 21/37] target/riscv: 64-bit Add/Subtract Instructions |
Date: |
Thu, 10 Jun 2021 15:58:52 +0800 |
64-bit add/subtract with saturation or halving operation.
Signed-off-by: LIU Zhiwei <zhiwei_liu@c-sky.com>
---
target/riscv/helper.h | 11 ++
target/riscv/insn32.decode | 11 ++
target/riscv/insn_trans/trans_rvp.c.inc | 74 +++++++++++++
target/riscv/packed_helper.c | 132 ++++++++++++++++++++++++
4 files changed, 228 insertions(+)
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index a0e3131512..192ef42d2a 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -1330,3 +1330,14 @@ DEF_HELPER_4(pbsada, tl, env, tl, tl, tl)
DEF_HELPER_4(smaqa, tl, env, tl, tl, tl)
DEF_HELPER_4(umaqa, tl, env, tl, tl, tl)
DEF_HELPER_4(smaqa_su, tl, env, tl, tl, tl)
+
+DEF_HELPER_3(add64, i64, env, i64, i64)
+DEF_HELPER_3(radd64, i64, env, i64, i64)
+DEF_HELPER_3(uradd64, i64, env, i64, i64)
+DEF_HELPER_3(kadd64, i64, env, i64, i64)
+DEF_HELPER_3(ukadd64, i64, env, i64, i64)
+DEF_HELPER_3(sub64, i64, env, i64, i64)
+DEF_HELPER_3(rsub64, i64, env, i64, i64)
+DEF_HELPER_3(ursub64, i64, env, i64, i64)
+DEF_HELPER_3(ksub64, i64, env, i64, i64)
+DEF_HELPER_3(uksub64, i64, env, i64, i64)
diff --git a/target/riscv/insn32.decode b/target/riscv/insn32.decode
index 96288370a6..5156fa060e 100644
--- a/target/riscv/insn32.decode
+++ b/target/riscv/insn32.decode
@@ -933,3 +933,14 @@ pbsada 1111111 ..... ..... 000 ..... 1110111 @r
smaqa 1100100 ..... ..... 000 ..... 1110111 @r
umaqa 1100110 ..... ..... 000 ..... 1110111 @r
smaqa_su 1100101 ..... ..... 000 ..... 1110111 @r
+
+add64 1100000 ..... ..... 001 ..... 1110111 @r
+radd64 1000000 ..... ..... 001 ..... 1110111 @r
+uradd64 1010000 ..... ..... 001 ..... 1110111 @r
+kadd64 1001000 ..... ..... 001 ..... 1110111 @r
+ukadd64 1011000 ..... ..... 001 ..... 1110111 @r
+sub64 1100001 ..... ..... 001 ..... 1110111 @r
+rsub64 1000001 ..... ..... 001 ..... 1110111 @r
+ursub64 1010001 ..... ..... 001 ..... 1110111 @r
+ksub64 1001001 ..... ..... 001 ..... 1110111 @r
+uksub64 1011001 ..... ..... 001 ..... 1110111 @r
diff --git a/target/riscv/insn_trans/trans_rvp.c.inc
b/target/riscv/insn_trans/trans_rvp.c.inc
index 1a10f13318..e04c79931d 100644
--- a/target/riscv/insn_trans/trans_rvp.c.inc
+++ b/target/riscv/insn_trans/trans_rvp.c.inc
@@ -516,3 +516,77 @@ GEN_RVP_R_ACC_OOL(pbsada);
GEN_RVP_R_ACC_OOL(smaqa);
GEN_RVP_R_ACC_OOL(umaqa);
GEN_RVP_R_ACC_OOL(smaqa_su);
+
+/*
+ *** 64-bit Profile Instructions
+ */
+/* 64-bit Addition & Subtraction Instructions */
+static bool
+r_d64_s64_s64_ool(DisasContext *ctx, arg_r *a,
+ void (* fn)(TCGv_i64, TCGv_ptr, TCGv_i64, TCGv_i64))
+{
+ TCGv t1, t2;
+ TCGv_i64 src1, src2, dst;
+
+ if (!has_ext(ctx, RVP) || !ctx->ext_psfoperand) {
+ return false;
+ }
+
+ src1 = tcg_temp_new_i64();
+ src2 = tcg_temp_new_i64();
+ dst = tcg_temp_new_i64();
+
+ if (is_32bit(ctx)) {
+ TCGv a0, a1, b0, b1;
+ a0 = tcg_temp_new();
+ a1 = tcg_temp_new();
+ b0 = tcg_temp_new();
+ b1 = tcg_temp_new();
+
+ gen_get_gpr(a0, a->rs1);
+ gen_get_gpr(a1, a->rs1 + 1);
+ tcg_gen_concat_tl_i64(src1, a0, a1);
+ gen_get_gpr(b0, a->rs2);
+ gen_get_gpr(b1, a->rs2 + 1);
+ tcg_gen_concat_tl_i64(src2, b0, b1);
+
+ tcg_temp_free(a0);
+ tcg_temp_free(a1);
+ tcg_temp_free(b0);
+ tcg_temp_free(b1);
+ } else {
+ t1 = tcg_temp_new();
+ t2 = tcg_temp_new();
+ gen_get_gpr(t1, a->rs1);
+ tcg_gen_ext_tl_i64(src1, t1);
+ gen_get_gpr(t2, a->rs2);
+ tcg_gen_ext_tl_i64(src2, t2);
+ tcg_temp_free(t1);
+ tcg_temp_free(t2);
+ }
+
+ fn(dst, cpu_env, src1, src2);
+ set_pair_regs(ctx, dst, a->rd);
+
+ tcg_temp_free_i64(src1);
+ tcg_temp_free_i64(src2);
+ tcg_temp_free_i64(dst);
+ return true;
+}
+
+#define GEN_RVP_R_D64_S64_S64_OOL(NAME) \
+static bool trans_##NAME(DisasContext *s, arg_r *a) \
+{ \
+ return r_d64_s64_s64_ool(s, a, gen_helper_##NAME); \
+}
+
+GEN_RVP_R_D64_S64_S64_OOL(add64);
+GEN_RVP_R_D64_S64_S64_OOL(radd64);
+GEN_RVP_R_D64_S64_S64_OOL(uradd64);
+GEN_RVP_R_D64_S64_S64_OOL(kadd64);
+GEN_RVP_R_D64_S64_S64_OOL(ukadd64);
+GEN_RVP_R_D64_S64_S64_OOL(sub64);
+GEN_RVP_R_D64_S64_S64_OOL(rsub64);
+GEN_RVP_R_D64_S64_S64_OOL(ursub64);
+GEN_RVP_R_D64_S64_S64_OOL(ksub64);
+GEN_RVP_R_D64_S64_S64_OOL(uksub64);
diff --git a/target/riscv/packed_helper.c b/target/riscv/packed_helper.c
index 02178d6e61..b8be234d97 100644
--- a/target/riscv/packed_helper.c
+++ b/target/riscv/packed_helper.c
@@ -2088,3 +2088,135 @@ static inline void do_smaqa_su(CPURISCVState *env, void
*vd, void *va,
}
RVPR_ACC(smaqa_su, 1, 4);
+
+/*
+ *** 64-bit Profile Instructions
+ */
+/* 64-bit Addition & Subtraction Instructions */
+
+/* Define a common function to loop elements in packed register */
+static inline uint64_t
+rvpr64_64_64(CPURISCVState *env, uint64_t a, uint64_t b,
+ uint8_t step, uint8_t size, PackedFn3i *fn)
+{
+ int i, passes = sizeof(uint64_t) / size;
+ uint64_t result = 0;
+
+ for (i = 0; i < passes; i += step) {
+ fn(env, &result, &a, &b, i);
+ }
+ return result;
+}
+
+#define RVPR64_64_64(NAME, STEP, SIZE) \
+uint64_t HELPER(NAME)(CPURISCVState *env, uint64_t a, uint64_t b) \
+{ \
+ return rvpr64_64_64(env, a, b, STEP, SIZE, (PackedFn3i *)do_##NAME); \
+}
+
+static inline void do_add64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int64_t *d = vd, *a = va, *b = vb;
+ *d = *a + *b;
+}
+
+RVPR64_64_64(add64, 1, 8);
+
+static inline int64_t hadd64(int64_t a, int64_t b)
+{
+ int64_t res = a + b;
+ int64_t over = (res ^ a) & (res ^ b) & INT64_MIN;
+
+ /* With signed overflow, bit 64 is inverse of bit 63. */
+ return (res >> 1) ^ over;
+}
+
+static inline void do_radd64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int64_t *d = vd, *a = va, *b = vb;
+ *d = hadd64(*a, *b);
+}
+
+RVPR64_64_64(radd64, 1, 8);
+
+static inline uint64_t haddu64(uint64_t a, uint64_t b)
+{
+ uint64_t res = a + b;
+ bool over = res < a;
+
+ return over ? ((res >> 1) | INT64_MIN) : (res >> 1);
+}
+
+static inline void do_uradd64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ uint64_t *d = vd, *a = va, *b = vb;
+ *d = haddu64(*a, *b);
+}
+
+RVPR64_64_64(uradd64, 1, 8);
+
+static inline void do_kadd64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int64_t *d = vd, *a = va, *b = vb;
+ *d = sadd64(env, 0, *a, *b);
+}
+
+RVPR64_64_64(kadd64, 1, 8);
+
+static inline void do_ukadd64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ uint64_t *d = vd, *a = va, *b = vb;
+ *d = saddu64(env, 0, *a, *b);
+}
+
+RVPR64_64_64(ukadd64, 1, 8);
+
+static inline void do_sub64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int64_t *d = vd, *a = va, *b = vb;
+ *d = *a - *b;
+}
+
+RVPR64_64_64(sub64, 1, 8);
+
+static inline void do_rsub64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int64_t *d = vd, *a = va, *b = vb;
+ *d = hsub64(*a, *b);
+}
+
+RVPR64_64_64(rsub64, 1, 8);
+
+static inline void do_ursub64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ uint64_t *d = vd, *a = va, *b = vb;
+ *d = hsubu64(*a, *b);
+}
+
+RVPR64_64_64(ursub64, 1, 8);
+
+static inline void do_ksub64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int64_t *d = vd, *a = va, *b = vb;
+ *d = ssub64(env, 0, *a, *b);
+}
+
+RVPR64_64_64(ksub64, 1, 8);
+
+static inline void do_uksub64(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ uint64_t *d = vd, *a = va, *b = vb;
+ *d = ssubu64(env, 0, *a, *b);
+}
+
+RVPR64_64_64(uksub64, 1, 8);
--
2.25.1
- [PATCH v2 11/37] target/riscv: SIMD 16-bit Miscellaneous Instructions, (continued)
- [PATCH v2 11/37] target/riscv: SIMD 16-bit Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 12/37] target/riscv: SIMD 8-bit Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 13/37] target/riscv: 8-bit Unpacking Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 14/37] target/riscv: 16-bit Packing Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 15/37] target/riscv: Signed MSW 32x32 Multiply and Add Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 16/37] target/riscv: Signed MSW 32x16 Multiply and Add Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 17/37] target/riscv: Signed 16-bit Multiply 32-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 18/37] target/riscv: Signed 16-bit Multiply 64-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 19/37] target/riscv: Partial-SIMD Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 20/37] target/riscv: 8-bit Multiply with 32-bit Add Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 21/37] target/riscv: 64-bit Add/Subtract Instructions,
LIU Zhiwei <=
- [PATCH v2 22/37] target/riscv: 32-bit Multiply 64-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 23/37] target/riscv: Signed 16-bit Multiply with 64-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 24/37] target/riscv: Non-SIMD Q15 saturation ALU Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 25/37] target/riscv: Non-SIMD Q31 saturation ALU Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 26/37] target/riscv: 32-bit Computation Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 27/37] target/riscv: Non-SIMD Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 28/37] target/riscv: RV64 Only SIMD 32-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 29/37] target/riscv: RV64 Only SIMD 32-bit Shift Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 30/37] target/riscv: RV64 Only SIMD 32-bit Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 31/37] target/riscv: RV64 Only SIMD Q15 saturating Multiply Instructions, LIU Zhiwei, 2021/06/10