Re: [PATCH 12/38] target/riscv: SIMD 16-bit Miscellaneous Instructions

[Alistair Francis <alistair23@gmail.com>]

On Fri, Feb 12, 2021 at 10:28 AM LIU Zhiwei <zhiwei_liu@c-sky.com> wrote:
>
> Signed-off-by: LIU Zhiwei <zhiwei_liu@c-sky.com>

Acked-by: Alistair Francis <alistair.francis@wdc.com>

Alistair

> ---
>  target/riscv/helper.h                   |  12 ++
>  target/riscv/insn32.decode              |  13 ++
>  target/riscv/insn_trans/trans_rvp.c.inc |  42 ++++++
>  target/riscv/packed_helper.c            | 167 ++++++++++++++++++++++++
>  4 files changed, 234 insertions(+)
>
> diff --git a/target/riscv/helper.h b/target/riscv/helper.h
> index 6bb601b436..866484e37d 100644
> --- a/target/riscv/helper.h
> +++ b/target/riscv/helper.h
> @@ -1228,3 +1228,15 @@ DEF_HELPER_3(umul8, i64, env, tl, tl)
>  DEF_HELPER_3(umulx8, i64, env, tl, tl)
>  DEF_HELPER_3(khm8, tl, env, tl, tl)
>  DEF_HELPER_3(khmx8, tl, env, tl, tl)
> +
> +DEF_HELPER_3(smin16, tl, env, tl, tl)
> +DEF_HELPER_3(umin16, tl, env, tl, tl)
> +DEF_HELPER_3(smax16, tl, env, tl, tl)
> +DEF_HELPER_3(umax16, tl, env, tl, tl)
> +DEF_HELPER_3(sclip16, tl, env, tl, tl)
> +DEF_HELPER_3(uclip16, tl, env, tl, tl)
> +DEF_HELPER_2(kabs16, tl, env, tl)
> +DEF_HELPER_2(clrs16, tl, env, tl)
> +DEF_HELPER_2(clz16, tl, env, tl)
> +DEF_HELPER_2(clo16, tl, env, tl)
> +DEF_HELPER_2(swap16, tl, env, tl)
> diff --git a/target/riscv/insn32.decode b/target/riscv/insn32.decode
> index 9d165efba9..bc9d5fc967 100644
> --- a/target/riscv/insn32.decode
> +++ b/target/riscv/insn32.decode
> @@ -42,6 +42,7 @@
>  &i    imm rs1 rd
>  &j    imm rd
>  &r    rd rs1 rs2
> +&r2   rd rs1
>  &s    imm rs1 rs2
>  &u    imm rd
>  &shift     shamt rs1 rd
> @@ -695,3 +696,15 @@ umul8      1011100  ..... ..... 000 ..... 1111111 @r
>  umulx8     1011101  ..... ..... 000 ..... 1111111 @r
>  khm8       1000111  ..... ..... 000 ..... 1111111 @r
>  khmx8      1001111  ..... ..... 000 ..... 1111111 @r
> +
> +smin16     1000000  ..... ..... 000 ..... 1111111 @r
> +umin16     1001000  ..... ..... 000 ..... 1111111 @r
> +smax16     1000001  ..... ..... 000 ..... 1111111 @r
> +umax16     1001001  ..... ..... 000 ..... 1111111 @r
> +sclip16    1000010  0.... ..... 000 ..... 1111111 @sh4
> +uclip16    1000010  1.... ..... 000 ..... 1111111 @sh4
> +kabs16     1010110  10001 ..... 000 ..... 1111111 @r2
> +clrs16     1010111  01000 ..... 000 ..... 1111111 @r2
> +clz16      1010111  01001 ..... 000 ..... 1111111 @r2
> +clo16      1010111  01011 ..... 000 ..... 1111111 @r2
> +swap16     1010110  11001 ..... 000 ..... 1111111 @r2
> diff --git a/target/riscv/insn_trans/trans_rvp.c.inc b/target/riscv/insn_trans/trans_rvp.c.inc
> index 336f3418b1..56fb8b2523 100644
> --- a/target/riscv/insn_trans/trans_rvp.c.inc
> +++ b/target/riscv/insn_trans/trans_rvp.c.inc
> @@ -444,3 +444,45 @@ GEN_RVP_R_D64_OOL(umul8);
>  GEN_RVP_R_D64_OOL(umulx8);
>  GEN_RVP_R_OOL(khm8);
>  GEN_RVP_R_OOL(khmx8);
> +
> +/* SIMD 16-bit Miscellaneous Instructions */
> +GEN_RVP_R_OOL(smin16);
> +GEN_RVP_R_OOL(umin16);
> +GEN_RVP_R_OOL(smax16);
> +GEN_RVP_R_OOL(umax16);
> +GEN_RVP_SHIFTI(sclip16, sclip16, NULL);
> +GEN_RVP_SHIFTI(uclip16, uclip16, NULL);
> +
> +/* Out of line helpers for R2 format */
> +static bool
> +r2_ool(DisasContext *ctx, arg_r2 *a,
> +       void (* fn)(TCGv, TCGv_ptr, TCGv))
> +{
> +    TCGv src1, dst;
> +    if (!has_ext(ctx, RVP)) {
> +        return false;
> +    }
> +
> +    src1 = tcg_temp_new();
> +    dst = tcg_temp_new();
> +
> +    gen_get_gpr(src1, a->rs1);
> +    fn(dst, cpu_env, src1);
> +    gen_set_gpr(a->rd, dst);
> +
> +    tcg_temp_free(src1);
> +    tcg_temp_free(dst);
> +    return true;
> +}
> +
> +#define GEN_RVP_R2_OOL(NAME)                           \
> +static bool trans_##NAME(DisasContext *s, arg_r2 *a)   \
> +{                                                      \
> +    return r2_ool(s, a, gen_helper_##NAME);            \
> +}
> +
> +GEN_RVP_R2_OOL(kabs16);
> +GEN_RVP_R2_OOL(clrs16);
> +GEN_RVP_R2_OOL(clz16);
> +GEN_RVP_R2_OOL(clo16);
> +GEN_RVP_R2_OOL(swap16);
> diff --git a/target/riscv/packed_helper.c b/target/riscv/packed_helper.c
> index 56baefeb8e..a6ab011ace 100644
> --- a/target/riscv/packed_helper.c
> +++ b/target/riscv/packed_helper.c
> @@ -920,3 +920,170 @@ static inline void do_khmx8(CPURISCVState *env, void *vd, void *va,
>  }
>
>  RVPR(khmx8, 2, 1);
> +
> +/* SIMD 16-bit Miscellaneous Instructions */
> +static inline void do_smin16(CPURISCVState *env, void *vd, void *va,
> +                             void *vb, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va, *b = vb;
> +
> +    d[i] = (a[i] < b[i]) ? a[i] : b[i];
> +}
> +
> +RVPR(smin16, 1, 2);
> +
> +static inline void do_umin16(CPURISCVState *env, void *vd, void *va,
> +                             void *vb, uint8_t i)
> +{
> +    uint16_t *d = vd, *a = va, *b = vb;
> +
> +    d[i] = (a[i] < b[i]) ? a[i] : b[i];
> +}
> +
> +RVPR(umin16, 1, 2);
> +
> +static inline void do_smax16(CPURISCVState *env, void *vd, void *va,
> +                             void *vb, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va, *b = vb;
> +
> +    d[i] = (a[i] > b[i]) ? a[i] : b[i];
> +}
> +
> +RVPR(smax16, 1, 2);
> +
> +static inline void do_umax16(CPURISCVState *env, void *vd, void *va,
> +                             void *vb, uint8_t i)
> +{
> +    uint16_t *d = vd, *a = va, *b = vb;
> +
> +    d[i] = (a[i] > b[i]) ? a[i] : b[i];
> +}
> +
> +RVPR(umax16, 1, 2);
> +
> +static int64_t sat64(CPURISCVState *env, int64_t a, uint8_t shift)
> +{
> +    int64_t max = shift >= 64 ? INT64_MAX : (1ull << shift) - 1;
> +    int64_t min = shift >= 64 ? INT64_MIN : -(1ull << shift);
> +    int64_t result;
> +
> +    if (a > max) {
> +        result = max;
> +        env->vxsat = 0x1;
> +    } else if (a < min) {
> +        result = min;
> +        env->vxsat = 0x1;
> +    } else {
> +        result = a;
> +    }
> +    return result;
> +}
> +
> +static inline void do_sclip16(CPURISCVState *env, void *vd, void *va,
> +                              void *vb, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +    uint8_t shift = *(uint8_t *)vb & 0xf;
> +
> +    d[i] = sat64(env, a[i], shift);
> +}
> +
> +RVPR(sclip16, 1, 2);
> +
> +static uint64_t satu64(CPURISCVState *env, uint64_t a, uint8_t shift)
> +{
> +    uint64_t max = shift >= 64 ? UINT64_MAX : (1ull << shift) - 1;
> +    uint64_t result;
> +
> +    if (a > max) {
> +        result = max;
> +        env->vxsat = 0x1;
> +    } else {
> +        result = a;
> +    }
> +    return result;
> +}
> +
> +static inline void do_uclip16(CPURISCVState *env, void *vd, void *va,
> +                              void *vb, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +    uint8_t shift = *(uint8_t *)vb & 0xf;
> +
> +    if (a[i] < 0) {
> +        d[i] = 0;
> +        env->vxsat = 0x1;
> +    } else {
> +        d[i] = satu64(env, a[i], shift);
> +    }
> +}
> +
> +RVPR(uclip16, 1, 2);
> +
> +typedef void PackedFn2i(CPURISCVState *, void *, void *, uint8_t);
> +
> +static inline target_ulong rvpr2(CPURISCVState *env, target_ulong a,
> +                                 uint8_t step, uint8_t size, PackedFn2i *fn)
> +{
> +    int i, passes = sizeof(target_ulong) / size;
> +    target_ulong result;
> +
> +    for (i = 0; i < passes; i += step) {
> +        fn(env, &result, &a, i);
> +    }
> +    return result;
> +}
> +
> +#define RVPR2(NAME, STEP, SIZE)                                  \
> +target_ulong HELPER(NAME)(CPURISCVState *env, target_ulong a)    \
> +{                                                                \
> +    return rvpr2(env, a, STEP, SIZE, (PackedFn2i *)do_##NAME);   \
> +}
> +
> +static inline void do_kabs16(CPURISCVState *env, void *vd, void *va, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +
> +    if (a[i] == INT16_MIN) {
> +        d[i] = INT16_MAX;
> +        env->vxsat = 0x1;
> +    } else {
> +        d[i] = abs(a[i]);
> +    }
> +}
> +
> +RVPR2(kabs16, 1, 2);
> +
> +static inline void do_clrs16(CPURISCVState *env, void *vd, void *va, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +    d[i] = clrsb32(a[i]) - 16;
> +}
> +
> +RVPR2(clrs16, 1, 2);
> +
> +static inline void do_clz16(CPURISCVState *env, void *vd, void *va, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +    d[i] = (a[i] < 0) ? 0 : (clz32(a[i]) - 16);
> +}
> +
> +RVPR2(clz16, 1, 2);
> +
> +static inline void do_clo16(CPURISCVState *env, void *vd, void *va, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +    d[i] = (a[i] >= 0) ? 0 : (clo32(a[i]) - 16);
> +}
> +
> +RVPR2(clo16, 1, 2);
> +
> +static inline void do_swap16(CPURISCVState *env, void *vd, void *va, uint8_t i)
> +{
> +    int16_t *d = vd, *a = va;
> +    d[H2(i)] = a[H2(i + 1)];
> +    d[H2(i + 1)] = a[H2(i)];
> +}
> +
> +RVPR2(swap16, 2, 2);
> --
> 2.17.1
>