diff mbox series

[v3,27/37] target/riscv: Non-SIMD Miscellaneous Instructions

Message ID 20210624105521.3964-28-zhiwei_liu@c-sky.com (mailing list archive)
State New, archived
Headers show
Series target/riscv: support packed extension v0.9.4 | expand

Commit Message

LIU Zhiwei June 24, 2021, 10:55 a.m. UTC
Bit reverse, average, rounding shift, extract and insert byte
instructions.

Signed-off-by: LIU Zhiwei <zhiwei_liu@c-sky.com>
---
 target/riscv/helper.h                   |   6 +
 target/riscv/insn32.decode              |  16 ++
 target/riscv/insn_trans/trans_rvp.c.inc | 241 ++++++++++++++++++++++++
 target/riscv/packed_helper.c            |  77 ++++++++
 4 files changed, 340 insertions(+)
diff mbox series

Patch

diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index 3063b583f3..bdd5ca1251 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -1393,3 +1393,9 @@  DEF_HELPER_3(maxw, tl, env, tl, tl)
 DEF_HELPER_3(minw, tl, env, tl, tl)
 DEF_HELPER_3(mulr64, i64, env, tl, tl)
 DEF_HELPER_3(mulsr64, i64, env, tl, tl)
+
+DEF_HELPER_3(ave, tl, env, tl, tl)
+DEF_HELPER_3(sra_u, tl, env, tl, tl)
+DEF_HELPER_3(bitrev, tl, env, tl, tl)
+DEF_HELPER_3(wext, tl, env, i64, tl)
+DEF_HELPER_4(bpick, tl, env, tl, tl, tl)
diff --git a/target/riscv/insn32.decode b/target/riscv/insn32.decode
index 9cfe5570b0..b70f6f0dc2 100644
--- a/target/riscv/insn32.decode
+++ b/target/riscv/insn32.decode
@@ -26,6 +26,7 @@ 
 %sh7    20:7
 %sh4    20:4
 %sh3    20:3
+%sh6    20:6
 %csr    20:12
 %rm     12:3
 %nf     29:3                     !function=ex_plus_1
@@ -44,6 +45,7 @@ 
 &j    imm rd
 &r    rd rs1 rs2
 &r2   rd rs1
+&r4   rd rs1 rs2 rs3
 &s    imm rs1 rs2
 &u    imm rd
 &shift     shamt rs1 rd
@@ -65,6 +67,7 @@ 
 @sh      ......  ...... .....  ... ..... ....... &shift  shamt=%sh7     %rs1 %rd
 @sh4     ......  ...... .....  ... ..... ....... &shift  shamt=%sh4      %rs1 %rd
 @sh3     ......  ...... .....  ... ..... ....... &shift  shamt=%sh3      %rs1 %rd
+@sh6     ......  ...... .....  ... ..... ....... &shift  shamt=%sh6      %rs1 %rd
 @csr     ............   .....  ... ..... .......               %csr     %rs1 %rd
 
 @atom_ld ..... aq:1 rl:1 ..... ........ ..... ....... &atomic rs2=0     %rs1 %rd
@@ -74,6 +77,7 @@ 
 @r_rm    .......   ..... ..... ... ..... ....... %rs2 %rs1 %rm %rd
 @r2_rm   .......   ..... ..... ... ..... ....... %rs1 %rm %rd
 @r2      .......   ..... ..... ... ..... ....... &r2 %rs1 %rd
+@r4      ..... ..  ..... ..... ... ..... ....... %rs3 %rs2 %rs1 %rd
 @r2_nfvm ... ... vm:1 ..... ..... ... ..... ....... &r2nfvm %nf %rs1 %rd
 @r2_vm   ...... vm:1 ..... ..... ... ..... ....... &rmr %rs2 %rd
 @r1_vm   ...... vm:1 ..... ..... ... ..... ....... %rd
@@ -997,3 +1001,15 @@  maxw       1111001  ..... ..... 000 ..... 1110111 @r
 minw       1111000  ..... ..... 000 ..... 1110111 @r
 mulr64     1111000  ..... ..... 001 ..... 1110111 @r
 mulsr64    1110000  ..... ..... 001 ..... 1110111 @r
+
+ave        1110000  ..... ..... 000 ..... 1110111 @r
+sra_u      0010010  ..... ..... 001 ..... 1110111 @r
+srai_u     110101  ...... ..... 001 ..... 1110111 @sh6
+bitrev     1110011  ..... ..... 000 ..... 1110111 @r
+bitrevi    111010  ...... ..... 000 ..... 1110111 @sh6
+wext       1100111  ..... ..... 000 ..... 1110111 @r
+wexti      1101111  ..... ..... 000 ..... 1110111 @sh5
+bpick      .....00  ..... ..... 011 ..... 1110111 @r4
+insb       1010110  00 ... ..... 000 ..... 1110111 @sh3
+maddr32    1100010  ..... ..... 001 ..... 1110111 @r
+msubr32    1100011  ..... ..... 001 ..... 1110111 @r
diff --git a/target/riscv/insn_trans/trans_rvp.c.inc b/target/riscv/insn_trans/trans_rvp.c.inc
index b720c6e037..51e140d157 100644
--- a/target/riscv/insn_trans/trans_rvp.c.inc
+++ b/target/riscv/insn_trans/trans_rvp.c.inc
@@ -708,3 +708,244 @@  GEN_RVP_R_OOL(minw);
 GEN_RVP_R_OOL(maxw);
 GEN_RVP_R_D64_OOL(mulr64);
 GEN_RVP_R_D64_OOL(mulsr64);
+
+/* Non-SIMD Miscellaneous Instructions */
+GEN_RVP_R_OOL(ave);
+GEN_RVP_R_OOL(sra_u);
+GEN_RVP_SHIFTI(srai_u, NULL, gen_helper_sra_u);
+GEN_RVP_R_OOL(bitrev);
+GEN_RVP_SHIFTI(bitrevi, NULL, gen_helper_bitrev);
+
+static bool
+r_s64_ool(DisasContext *ctx, arg_r *a,
+          void (* fn)(TCGv, TCGv_ptr, TCGv_i64, TCGv))
+{
+    TCGv_i64 src1;
+    TCGv src2, dst;
+
+    if (!has_ext(ctx, RVP) || !ctx->ext_psfoperand) {
+        return false;
+    }
+
+    src1 = tcg_temp_new_i64();
+    src2 = tcg_temp_new();
+    dst = tcg_temp_new();
+
+    if (is_32bit(ctx)) {
+        TCGv t0, t1;
+        t0 = tcg_temp_new();
+        t1 = tcg_temp_new();
+        gen_get_gpr(t0, a->rs1);
+        gen_get_gpr(t1, a->rs1 + 1);
+        tcg_gen_concat_tl_i64(src1, t0, t1);
+        tcg_temp_free(t0);
+        tcg_temp_free(t1);
+    } else {
+        TCGv t0;
+        t0 = tcg_temp_new();
+        gen_get_gpr(t0, a->rs1);
+        tcg_gen_ext_tl_i64(src1, t0);
+        tcg_temp_free(t0);
+    }
+    gen_get_gpr(src2, a->rs2);
+    fn(dst, cpu_env, src1, src2);
+    gen_set_gpr(a->rd, dst);
+
+    tcg_temp_free_i64(src1);
+    tcg_temp_free(src2);
+    tcg_temp_free(dst);
+    return true;
+}
+
+#define GEN_RVP_R_S64_OOL(NAME)                        \
+static bool trans_##NAME(DisasContext *s, arg_r *a)    \
+{                                                      \
+    return r_s64_ool(s, a, gen_helper_##NAME);         \
+}
+
+GEN_RVP_R_S64_OOL(wext);
+
+static bool rvp_shifti_s64_ool(DisasContext *ctx, arg_shift *a,
+                               void (* fn)(TCGv, TCGv_ptr, TCGv_i64, TCGv))
+{
+    TCGv_i64 src1;
+    TCGv shift, dst;
+
+    if (!has_ext(ctx, RVP) || !ctx->ext_psfoperand) {
+        return false;
+    }
+
+    src1 = tcg_temp_new_i64();
+    dst = tcg_temp_new();
+
+    if (is_32bit(ctx)) {
+        TCGv t0, t1;
+        t0 = tcg_temp_new();
+        t1 = tcg_temp_new();
+        gen_get_gpr(t0, a->rs1);
+        gen_get_gpr(t1, a->rs1 + 1);
+        tcg_gen_concat_tl_i64(src1, t0, t1);
+        tcg_temp_free(t0);
+        tcg_temp_free(t1);
+    } else {
+        TCGv t0;
+        t0 = tcg_temp_new();
+        gen_get_gpr(t0, a->rs1);
+        tcg_gen_ext_tl_i64(src1, t0);
+        tcg_temp_free(t0);
+    }
+    shift = tcg_const_tl(a->shamt);
+    fn(dst, cpu_env, src1, shift);
+    gen_set_gpr(a->rd, dst);
+
+    tcg_temp_free_i64(src1);
+    tcg_temp_free(shift);
+    tcg_temp_free(dst);
+    return true;
+}
+
+#define GEN_RVP_SHIFTI_S64_OOL(NAME, OP)                    \
+static bool trans_##NAME(DisasContext *s, arg_shift *a)     \
+{                                                           \
+    return rvp_shifti_s64_ool(s, a, gen_helper_##OP);       \
+}
+
+GEN_RVP_SHIFTI_S64_OOL(wexti, wext);
+
+typedef void gen_helper_rvp_r4(TCGv, TCGv_ptr, TCGv, TCGv, TCGv);
+
+static bool r4_ool(DisasContext *ctx, arg_r4 *a, gen_helper_rvp_r4 *fn)
+{
+    TCGv src1, src2, src3, dst;
+    if (!has_ext(ctx, RVP)) {
+        return false;
+    }
+
+    src1 = tcg_temp_new();
+    src2 = tcg_temp_new();
+    src3 = tcg_temp_new();
+    dst = tcg_temp_new();
+
+    gen_get_gpr(src1, a->rs1);
+    gen_get_gpr(src2, a->rs2);
+    gen_get_gpr(src3, a->rs3);
+    fn(dst, cpu_env, src1, src2, src3);
+    gen_set_gpr(a->rd, dst);
+
+    tcg_temp_free(src1);
+    tcg_temp_free(src2);
+    tcg_temp_free(src3);
+    tcg_temp_free(dst);
+    return true;
+}
+
+#define GEN_RVP_R4_OOL(NAME)                           \
+static bool trans_##NAME(DisasContext *s, arg_r4 *a)   \
+{                                                      \
+    return r4_ool(s, a, gen_helper_##NAME);            \
+}
+
+GEN_RVP_R4_OOL(bpick);
+
+static bool trans_insb(DisasContext *ctx, arg_shift *a)
+{
+    TCGv src1, dst, b0;
+    uint8_t shift;
+    if (!has_ext(ctx, RVP)) {
+        return false;
+    }
+    if (is_32bit(ctx)) {
+        shift = a->shamt & 0x3;
+    } else {
+        shift = a->shamt;
+    }
+    src1 = tcg_temp_new();
+    dst = tcg_temp_new();
+    b0 = tcg_temp_new();
+
+    gen_get_gpr(src1, a->rs1);
+    gen_get_gpr(dst, a->rd);
+
+    tcg_gen_andi_tl(b0, src1, 0xff);
+    tcg_gen_deposit_tl(dst, dst, b0, shift * 8, 8);
+    gen_set_gpr(a->rd, dst);
+
+    tcg_temp_free(src1);
+    tcg_temp_free(dst);
+    tcg_temp_free(b0);
+    return true;
+}
+
+static bool trans_maddr32(DisasContext *ctx, arg_r *a)
+{
+    TCGv src1, src2, dst;
+    TCGv_i32 w1, w2, w3;
+    if (!has_ext(ctx, RVP)) {
+        return false;
+    }
+
+    src1 = tcg_temp_new();
+    src2 = tcg_temp_new();
+    dst = tcg_temp_new();
+    w1 = tcg_temp_new_i32();
+    w2 = tcg_temp_new_i32();
+    w3 = tcg_temp_new_i32();
+
+    gen_get_gpr(src1, a->rs1);
+    gen_get_gpr(src2, a->rs2);
+    gen_get_gpr(dst, a->rd);
+
+    tcg_gen_trunc_tl_i32(w1, src1);
+    tcg_gen_trunc_tl_i32(w2, src2);
+    tcg_gen_trunc_tl_i32(w3, dst);
+
+    tcg_gen_mul_i32(w1, w1, w2);
+    tcg_gen_add_i32(w3, w3, w1);
+    tcg_gen_ext_i32_tl(dst, w3);
+    gen_set_gpr(a->rd, dst);
+
+    tcg_temp_free(src1);
+    tcg_temp_free(src2);
+    tcg_temp_free(dst);
+    tcg_temp_free_i32(w1);
+    tcg_temp_free_i32(w2);
+    tcg_temp_free_i32(w3);
+    return true;
+}
+
+static bool trans_msubr32(DisasContext *ctx, arg_r *a)
+{
+    TCGv src1, src2, dst;
+    TCGv_i32 w1, w2, w3;
+    if (!has_ext(ctx, RVP)) {
+        return false;
+    }
+
+    src1 = tcg_temp_new();
+    src2 = tcg_temp_new();
+    dst = tcg_temp_new();
+    w1 = tcg_temp_new_i32();
+    w2 = tcg_temp_new_i32();
+    w3 = tcg_temp_new_i32();
+
+    gen_get_gpr(src1, a->rs1);
+    gen_get_gpr(src2, a->rs2);
+    gen_get_gpr(dst, a->rd);
+
+    tcg_gen_trunc_tl_i32(w1, src1);
+    tcg_gen_trunc_tl_i32(w2, src2);
+    tcg_gen_trunc_tl_i32(w3, dst);
+
+    tcg_gen_mul_i32(w1, w1, w2);
+    tcg_gen_sub_i32(w3, w3, w1);
+    tcg_gen_ext_i32_tl(dst, w3);
+    gen_set_gpr(a->rd, dst);
+
+    tcg_temp_free(src1);
+    tcg_temp_free(src2);
+    tcg_temp_free(dst);
+    tcg_temp_free_i32(w1);
+    tcg_temp_free_i32(w2);
+    tcg_temp_free_i32(w3);
+    return true;
+}
diff --git a/target/riscv/packed_helper.c b/target/riscv/packed_helper.c
index c0e3b6bbdb..4e0c7a92eb 100644
--- a/target/riscv/packed_helper.c
+++ b/target/riscv/packed_helper.c
@@ -2910,3 +2910,80 @@  static inline void do_mulsr64(CPURISCVState *env, void *vd, void *va,
 }
 
 RVPR64(mulsr64);
+
+/* Miscellaneous Instructions */
+static inline void do_ave(CPURISCVState *env, void *vd, void *va,
+                          void *vb, uint8_t i)
+{
+    target_long *d = vd, *a = va, *b = vb, half;
+
+    half = hadd64(*a, *b);
+    if ((*a ^ *b) & 0x1) {
+        half++;
+    }
+    *d = half;
+}
+
+RVPR(ave, 1, sizeof(target_ulong));
+
+static inline void do_sra_u(CPURISCVState *env, void *vd, void *va,
+                            void *vb, uint8_t i)
+{
+    target_long *d = vd, *a = va;
+    uint8_t *b = vb;
+    uint8_t shift = riscv_has_ext(env, RV32) ? (*b & 0x1f) : (*b & 0x3f);
+
+    *d = vssra64(env, 0, *a, shift);
+}
+
+RVPR(sra_u, 1, sizeof(target_ulong));
+
+static inline void do_bitrev(CPURISCVState *env, void *vd, void *va,
+                             void *vb, uint8_t i)
+{
+    target_ulong *d = vd, *a = va;
+    uint8_t *b = vb;
+    uint8_t shift = riscv_has_ext(env, RV32) ? (*b & 0x1f) : (*b & 0x3f);
+
+    *d = revbit64(*a) >> (64 - shift - 1);
+}
+
+RVPR(bitrev, 1, sizeof(target_ulong));
+
+static inline target_ulong
+rvpr_64(CPURISCVState *env, uint64_t a, target_ulong b, PackedFn3 *fn)
+{
+    target_ulong result = 0;
+
+    fn(env, &result, &a, &b);
+    return result;
+}
+
+#define RVPR_64(NAME)                                       \
+target_ulong HELPER(NAME)(CPURISCVState *env, uint64_t a,   \
+                          target_ulong b)                   \
+{                                                           \
+    return rvpr_64(env, a, b, (PackedFn3 *)do_##NAME);      \
+}
+
+static inline void do_wext(CPURISCVState *env, void *vd, void *va,
+                           void *vb, uint8_t i)
+{
+    target_long *d = vd;
+    int64_t *a = va;
+    uint8_t b = *(uint8_t *)vb & 0x1f;
+
+    *d = sextract64(*a, b, 32);
+}
+
+RVPR_64(wext);
+
+static inline void do_bpick(CPURISCVState *env, void *vd, void *va,
+                            void *vb, void *vc, uint8_t i)
+{
+    target_long *d = vd, *a = va, *b = vb, *c = vc;
+
+    *d = (*c & *a) | (~*c & *b);
+}
+
+RVPR_ACC(bpick, 1, sizeof(target_ulong));