diff mbox series

[RESEND,v5,51/57] target/loongarch: Implement xvinsgr2vr xvpickve2gr

Message ID 20230907083158.3975132-52-gaosong@loongson.cn
State New
Headers show
Series Add LoongArch LASX instructions | expand

Commit Message

gaosong Sept. 7, 2023, 8:31 a.m. UTC
This patch includes:
- XVINSGR2VR.{W/D};
- XVPICKVE2GR.{W/D}[U].

Signed-off-by: Song Gao <gaosong@loongson.cn>
---
 target/loongarch/insns.decode               |  7 +++
 target/loongarch/disas.c                    | 17 ++++++++
 target/loongarch/insn_trans/trans_vec.c.inc | 48 +++++++++++++++++++++
 3 files changed, 72 insertions(+)

Comments

Richard Henderson Sept. 11, 2023, 10:27 p.m. UTC | #1
On 9/7/23 01:31, Song Gao wrote:
> +static bool trans_xvinsgr2vr_w(DisasContext *ctx, arg_vr_i *a)
> +{
> +    if (!avail_LASX(ctx)) {
> +        return false;
> +    }
> +    return trans_vinsgr2vr_w(ctx, a);
> +}

Using the other translator doesn't help.

> static bool trans_vinsgr2vr_w(DisasContext *ctx, arg_vr_i *a)
> {
>     TCGv src = gpr_src(ctx, a->rj, EXT_NONE);
> 
>     if (!avail_LSX(ctx)) {
>         return false;
>     }
> 
>     CHECK_SXE;

This portion doesn't apply, and you miss the check_vec for the larger LASX.

>     tcg_gen_st32_i64(src, cpu_env,
>                      offsetof(CPULoongArchState, fpr[a->vd].vreg.W(a->imm)));
>     return true;
> }

The only thing that is left is this one line, so I'm not sure it's worth splitting out a 
common helper function.


r~
gaosong Sept. 12, 2023, 9:09 a.m. UTC | #2
在 2023/9/12 上午6:27, Richard Henderson 写道:
> On 9/7/23 01:31, Song Gao wrote:
>> +static bool trans_xvinsgr2vr_w(DisasContext *ctx, arg_vr_i *a)
>> +{
>> +    if (!avail_LASX(ctx)) {
>> +        return false;
>> +    }
>> +    return trans_vinsgr2vr_w(ctx, a);
>> +}
> 
> Using the other translator doesn't help.
> 
>> static bool trans_vinsgr2vr_w(DisasContext *ctx, arg_vr_i *a)
>> {
>>     TCGv src = gpr_src(ctx, a->rj, EXT_NONE);
>>
>>     if (!avail_LSX(ctx)) {
>>         return false;
>>     }
>>
>>     CHECK_SXE;
> 
> This portion doesn't apply, and you miss the check_vec for the larger LASX.
> 
>>     tcg_gen_st32_i64(src, cpu_env,
>>                      offsetof(CPULoongArchState, 
>> fpr[a->vd].vreg.W(a->imm)));
>>     return true;
>> }
> 
> The only thing that is left is this one line, so I'm not sure it's worth 
> splitting out a common helper function.
> 
> I think we need, like this:

static bool gen_g2v_vl(DisasContext *ctx, arg_vr_i *a, uint32_t oprsz, 
MemOp mop,
                        void (*func)(TCGv, TCGv_ptr, tcg_target_long))
{
     TCGv src = gpr_src(ctx, a->rj, EXT_NONE);

     if (!check_vec(ctx, oprsz)) {
         return true;
     }

     func(src, cpu_env, vec_reg_offset(a->vd, a->imm, mop));

     return true;
}

static bool gen_g2v(DisasContext *ctx, arg_vr_i *a, MemOp mop,
                     void (*func)(TCGv, TCGv_ptr, tcg_target_long))
{
     return gen_g2v_vl(ctx, a, 16, mop, func);
}

static bool gen_g2x(DisasContext *ctx, arg_vr_i *a, MemOp mop,
                     void (*func)(TCGv, TCGv_ptr, tcg_target_long))
{
     return gen_g2v_vl(ctx, a, 32, mop, func);
}

TRANS(vinsgr2vr_b, LSX, gen_g2v, MO_8, tcg_gen_st8_i64)
TRANS(vinsgr2vr_h, LSX, gen_g2v, MO_16, tcg_gen_st16_i64)
TRANS(vinsgr2vr_w, LSX, gen_g2v, MO_32, tcg_gen_st32_i64)
TRANS(vinsgr2vr_d, LSX, gen_g2v, MO_64, tcg_gen_st_i64)
TRANS(xvinsgr2vr_w, LASX, gen_g2x, MO_32, tcg_gen_st32_i64)
TRANS(xvinsgr2vr_d, LASX, gen_g2x, MO_64, tcg_gen_st_i64)

Thanks.
Song Gao
Richard Henderson Sept. 12, 2023, 4:20 p.m. UTC | #3
On 9/12/23 02:09, gaosong wrote:
> static bool gen_g2v_vl(DisasContext *ctx, arg_vr_i *a, uint32_t oprsz, MemOp mop,
>                         void (*func)(TCGv, TCGv_ptr, tcg_target_long))
> {
>      TCGv src = gpr_src(ctx, a->rj, EXT_NONE);
> 
>      if (!check_vec(ctx, oprsz)) {
>          return true;
>      }
> 
>      func(src, cpu_env, vec_reg_offset(a->vd, a->imm, mop));
> 
>      return true;
> }
> 
> static bool gen_g2v(DisasContext *ctx, arg_vr_i *a, MemOp mop,
>                      void (*func)(TCGv, TCGv_ptr, tcg_target_long))
> {
>      return gen_g2v_vl(ctx, a, 16, mop, func);
> }
> 
> static bool gen_g2x(DisasContext *ctx, arg_vr_i *a, MemOp mop,
>                      void (*func)(TCGv, TCGv_ptr, tcg_target_long))
> {
>      return gen_g2v_vl(ctx, a, 32, mop, func);
> }
> 
> TRANS(vinsgr2vr_b, LSX, gen_g2v, MO_8, tcg_gen_st8_i64)
> TRANS(vinsgr2vr_h, LSX, gen_g2v, MO_16, tcg_gen_st16_i64)
> TRANS(vinsgr2vr_w, LSX, gen_g2v, MO_32, tcg_gen_st32_i64)
> TRANS(vinsgr2vr_d, LSX, gen_g2v, MO_64, tcg_gen_st_i64)
> TRANS(xvinsgr2vr_w, LASX, gen_g2x, MO_32, tcg_gen_st32_i64)
> TRANS(xvinsgr2vr_d, LASX, gen_g2x, MO_64, tcg_gen_st_i64)

Looks perfect, thanks.


r~
diff mbox series

Patch

diff --git a/target/loongarch/insns.decode b/target/loongarch/insns.decode
index ad6751fdfb..bb3bb447ae 100644
--- a/target/loongarch/insns.decode
+++ b/target/loongarch/insns.decode
@@ -1976,6 +1976,13 @@  xvsetallnez_h    0111 01101001 11001 01101 ..... 00 ...   @cv
 xvsetallnez_w    0111 01101001 11001 01110 ..... 00 ...   @cv
 xvsetallnez_d    0111 01101001 11001 01111 ..... 00 ...   @cv
 
+xvinsgr2vr_w     0111 01101110 10111 10 ... ..... .....   @vr_ui3
+xvinsgr2vr_d     0111 01101110 10111 110 .. ..... .....   @vr_ui2
+xvpickve2gr_w    0111 01101110 11111 10 ... ..... .....   @rv_ui3
+xvpickve2gr_d    0111 01101110 11111 110 .. ..... .....   @rv_ui2
+xvpickve2gr_wu   0111 01101111 00111 10 ... ..... .....   @rv_ui3
+xvpickve2gr_du   0111 01101111 00111 110 .. ..... .....   @rv_ui2
+
 xvreplgr2vr_b    0111 01101001 11110 00000 ..... .....    @vr
 xvreplgr2vr_h    0111 01101001 11110 00001 ..... .....    @vr
 xvreplgr2vr_w    0111 01101001 11110 00010 ..... .....    @vr
diff --git a/target/loongarch/disas.c b/target/loongarch/disas.c
index abe113b150..04f9f9fa4b 100644
--- a/target/loongarch/disas.c
+++ b/target/loongarch/disas.c
@@ -1738,6 +1738,16 @@  static void output_vv_x(DisasContext *ctx, arg_vv *a, const char *mnemonic)
     output(ctx, mnemonic, "x%d, x%d", a->vd, a->vj);
 }
 
+static void output_vr_i_x(DisasContext *ctx, arg_vr_i *a, const char *mnemonic)
+{
+    output(ctx, mnemonic, "x%d, r%d, 0x%x", a->vd, a->rj, a->imm);
+}
+
+static void output_rv_i_x(DisasContext *ctx, arg_rv_i *a, const char *mnemonic)
+{
+    output(ctx, mnemonic, "r%d, x%d, 0x%x", a->rd, a->vj, a->imm);
+}
+
 INSN_LASX(xvadd_b,           vvv)
 INSN_LASX(xvadd_h,           vvv)
 INSN_LASX(xvadd_w,           vvv)
@@ -2497,6 +2507,13 @@  INSN_LASX(xvsetallnez_h,     cv)
 INSN_LASX(xvsetallnez_w,     cv)
 INSN_LASX(xvsetallnez_d,     cv)
 
+INSN_LASX(xvinsgr2vr_w,      vr_i)
+INSN_LASX(xvinsgr2vr_d,      vr_i)
+INSN_LASX(xvpickve2gr_w,     rv_i)
+INSN_LASX(xvpickve2gr_d,     rv_i)
+INSN_LASX(xvpickve2gr_wu,    rv_i)
+INSN_LASX(xvpickve2gr_du,    rv_i)
+
 INSN_LASX(xvreplgr2vr_b,     vr)
 INSN_LASX(xvreplgr2vr_h,     vr)
 INSN_LASX(xvreplgr2vr_w,     vr)
diff --git a/target/loongarch/insn_trans/trans_vec.c.inc b/target/loongarch/insn_trans/trans_vec.c.inc
index b68daa53ae..bf44a4d1fc 100644
--- a/target/loongarch/insn_trans/trans_vec.c.inc
+++ b/target/loongarch/insn_trans/trans_vec.c.inc
@@ -5268,6 +5268,54 @@  static bool trans_vpickve2gr_du(DisasContext *ctx, arg_rv_i *a)
     return true;
 }
 
+static bool trans_xvinsgr2vr_w(DisasContext *ctx, arg_vr_i *a)
+{
+    if (!avail_LASX(ctx)) {
+        return false;
+    }
+    return trans_vinsgr2vr_w(ctx, a);
+}
+
+static bool trans_xvinsgr2vr_d(DisasContext *ctx, arg_vr_i *a)
+{
+    if (!avail_LASX(ctx)) {
+        return false;
+    }
+    return trans_vinsgr2vr_d(ctx, a);
+}
+
+static bool trans_xvpickve2gr_w(DisasContext *ctx, arg_rv_i *a)
+{
+    if (!avail_LASX(ctx)) {
+        return false;
+    }
+    return trans_vpickve2gr_w(ctx, a);
+}
+
+static bool trans_xvpickve2gr_d(DisasContext *ctx, arg_rv_i *a)
+{
+    if (!avail_LASX(ctx)) {
+        return false;
+    }
+    return trans_vpickve2gr_d(ctx, a);
+}
+
+static bool trans_xvpickve2gr_wu(DisasContext *ctx, arg_rv_i *a)
+{
+    if (!avail_LASX(ctx)) {
+        return false;
+    }
+    return trans_vpickve2gr_wu(ctx, a);
+}
+
+static bool trans_xvpickve2gr_du(DisasContext *ctx, arg_rv_i *a)
+{
+    if (!avail_LASX(ctx)) {
+        return false;
+    }
+    return trans_vpickve2gr_du(ctx, a);
+}
+
 static bool gvec_dup_vl(DisasContext *ctx, arg_vr *a,
                         uint32_t oprsz, MemOp mop)
 {