diff mbox series

[12/18] tcg/loongarch64: Support LASX in tcg_out_addsub_vec

Message ID 20240527211912.14060-13-richard.henderson@linaro.org
State New
Headers show
Series tcg/loongarch64: Support v64 and v256 | expand

Commit Message

Richard Henderson May 27, 2024, 9:19 p.m. UTC
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/loongarch64/tcg-target.c.inc | 36 ++++++++++++++++++--------------
 1 file changed, 20 insertions(+), 16 deletions(-)

Comments

gaosong June 18, 2024, 1:34 a.m. UTC | #1
在 2024/5/28 上午5:19, Richard Henderson 写道:
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>   tcg/loongarch64/tcg-target.c.inc | 36 ++++++++++++++++++--------------
>   1 file changed, 20 insertions(+), 16 deletions(-)
Reviewed-by: Song Gao <gaosong@loongson.cn>

Thanks.
Song Gao
> diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc
> index 47011488dd..652aa261a3 100644
> --- a/tcg/loongarch64/tcg-target.c.inc
> +++ b/tcg/loongarch64/tcg-target.c.inc
> @@ -1758,21 +1758,25 @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece,
>       tcg_out_dup_vec(s, type, vece, rd, TCG_REG_TMP0);
>   }
>   
> -static void tcg_out_addsub_vec(TCGContext *s, unsigned vece, const TCGArg a0,
> -                               const TCGArg a1, const TCGArg a2,
> +static void tcg_out_addsub_vec(TCGContext *s, bool lasx, unsigned vece,
> +                               TCGArg a0, TCGArg a1, TCGArg a2,
>                                  bool a2_is_const, bool is_add)
>   {
> -    static const LoongArchInsn add_vec_insn[4] = {
> -        OPC_VADD_B, OPC_VADD_H, OPC_VADD_W, OPC_VADD_D
> +    static const LoongArchInsn add_vec_insn[2][4] = {
> +        { OPC_VADD_B, OPC_VADD_H, OPC_VADD_W, OPC_VADD_D },
> +        { OPC_XVADD_B, OPC_XVADD_H, OPC_XVADD_W, OPC_XVADD_D },
>       };
> -    static const LoongArchInsn add_vec_imm_insn[4] = {
> -        OPC_VADDI_BU, OPC_VADDI_HU, OPC_VADDI_WU, OPC_VADDI_DU
> +    static const LoongArchInsn add_vec_imm_insn[2][4] = {
> +        { OPC_VADDI_BU, OPC_VADDI_HU, OPC_VADDI_WU, OPC_VADDI_DU },
> +        { OPC_XVADDI_BU, OPC_XVADDI_HU, OPC_XVADDI_WU, OPC_XVADDI_DU },
>       };
> -    static const LoongArchInsn sub_vec_insn[4] = {
> -        OPC_VSUB_B, OPC_VSUB_H, OPC_VSUB_W, OPC_VSUB_D
> +    static const LoongArchInsn sub_vec_insn[2][4] = {
> +        { OPC_VSUB_B, OPC_VSUB_H, OPC_VSUB_W, OPC_VSUB_D },
> +        { OPC_XVSUB_B, OPC_XVSUB_H, OPC_XVSUB_W, OPC_XVSUB_D },
>       };
> -    static const LoongArchInsn sub_vec_imm_insn[4] = {
> -        OPC_VSUBI_BU, OPC_VSUBI_HU, OPC_VSUBI_WU, OPC_VSUBI_DU
> +    static const LoongArchInsn sub_vec_imm_insn[2][4] = {
> +        { OPC_VSUBI_BU, OPC_VSUBI_HU, OPC_VSUBI_WU, OPC_VSUBI_DU },
> +        { OPC_XVSUBI_BU, OPC_XVSUBI_HU, OPC_XVSUBI_WU, OPC_XVSUBI_DU },
>       };
>       LoongArchInsn insn;
>   
> @@ -1783,10 +1787,10 @@ static void tcg_out_addsub_vec(TCGContext *s, unsigned vece, const TCGArg a0,
>               value = -value;
>           }
>           if (value < 0) {
> -            insn = sub_vec_imm_insn[vece];
> +            insn = sub_vec_imm_insn[lasx][vece];
>               value = -value;
>           } else {
> -            insn = add_vec_imm_insn[vece];
> +            insn = add_vec_imm_insn[lasx][vece];
>           }
>   
>           /* Constraint TCG_CT_CONST_VADD ensures validity. */
> @@ -1797,9 +1801,9 @@ static void tcg_out_addsub_vec(TCGContext *s, unsigned vece, const TCGArg a0,
>       }
>   
>       if (is_add) {
> -        insn = add_vec_insn[vece];
> +        insn = add_vec_insn[lasx][vece];
>       } else {
> -        insn = sub_vec_insn[vece];
> +        insn = sub_vec_insn[lasx][vece];
>       }
>       tcg_out32(s, encode_vdvjvk_insn(insn, a0, a1, a2));
>   }
> @@ -1963,10 +1967,10 @@ static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc,
>           }
>           break;
>       case INDEX_op_add_vec:
> -        tcg_out_addsub_vec(s, vece, a0, a1, a2, const_args[2], true);
> +        tcg_out_addsub_vec(s, false, vece, a0, a1, a2, const_args[2], true);
>           break;
>       case INDEX_op_sub_vec:
> -        tcg_out_addsub_vec(s, vece, a0, a1, a2, const_args[2], false);
> +        tcg_out_addsub_vec(s, false, vece, a0, a1, a2, const_args[2], false);
>           break;
>       case INDEX_op_neg_vec:
>           tcg_out32(s, encode_vdvj_insn(neg_vec_insn[vece], a0, a1));
diff mbox series

Patch

diff --git a/tcg/loongarch64/tcg-target.c.inc b/tcg/loongarch64/tcg-target.c.inc
index 47011488dd..652aa261a3 100644
--- a/tcg/loongarch64/tcg-target.c.inc
+++ b/tcg/loongarch64/tcg-target.c.inc
@@ -1758,21 +1758,25 @@  static void tcg_out_dupi_vec(TCGContext *s, TCGType type, unsigned vece,
     tcg_out_dup_vec(s, type, vece, rd, TCG_REG_TMP0);
 }
 
-static void tcg_out_addsub_vec(TCGContext *s, unsigned vece, const TCGArg a0,
-                               const TCGArg a1, const TCGArg a2,
+static void tcg_out_addsub_vec(TCGContext *s, bool lasx, unsigned vece,
+                               TCGArg a0, TCGArg a1, TCGArg a2,
                                bool a2_is_const, bool is_add)
 {
-    static const LoongArchInsn add_vec_insn[4] = {
-        OPC_VADD_B, OPC_VADD_H, OPC_VADD_W, OPC_VADD_D
+    static const LoongArchInsn add_vec_insn[2][4] = {
+        { OPC_VADD_B, OPC_VADD_H, OPC_VADD_W, OPC_VADD_D },
+        { OPC_XVADD_B, OPC_XVADD_H, OPC_XVADD_W, OPC_XVADD_D },
     };
-    static const LoongArchInsn add_vec_imm_insn[4] = {
-        OPC_VADDI_BU, OPC_VADDI_HU, OPC_VADDI_WU, OPC_VADDI_DU
+    static const LoongArchInsn add_vec_imm_insn[2][4] = {
+        { OPC_VADDI_BU, OPC_VADDI_HU, OPC_VADDI_WU, OPC_VADDI_DU },
+        { OPC_XVADDI_BU, OPC_XVADDI_HU, OPC_XVADDI_WU, OPC_XVADDI_DU },
     };
-    static const LoongArchInsn sub_vec_insn[4] = {
-        OPC_VSUB_B, OPC_VSUB_H, OPC_VSUB_W, OPC_VSUB_D
+    static const LoongArchInsn sub_vec_insn[2][4] = {
+        { OPC_VSUB_B, OPC_VSUB_H, OPC_VSUB_W, OPC_VSUB_D },
+        { OPC_XVSUB_B, OPC_XVSUB_H, OPC_XVSUB_W, OPC_XVSUB_D },
     };
-    static const LoongArchInsn sub_vec_imm_insn[4] = {
-        OPC_VSUBI_BU, OPC_VSUBI_HU, OPC_VSUBI_WU, OPC_VSUBI_DU
+    static const LoongArchInsn sub_vec_imm_insn[2][4] = {
+        { OPC_VSUBI_BU, OPC_VSUBI_HU, OPC_VSUBI_WU, OPC_VSUBI_DU },
+        { OPC_XVSUBI_BU, OPC_XVSUBI_HU, OPC_XVSUBI_WU, OPC_XVSUBI_DU },
     };
     LoongArchInsn insn;
 
@@ -1783,10 +1787,10 @@  static void tcg_out_addsub_vec(TCGContext *s, unsigned vece, const TCGArg a0,
             value = -value;
         }
         if (value < 0) {
-            insn = sub_vec_imm_insn[vece];
+            insn = sub_vec_imm_insn[lasx][vece];
             value = -value;
         } else {
-            insn = add_vec_imm_insn[vece];
+            insn = add_vec_imm_insn[lasx][vece];
         }
 
         /* Constraint TCG_CT_CONST_VADD ensures validity. */
@@ -1797,9 +1801,9 @@  static void tcg_out_addsub_vec(TCGContext *s, unsigned vece, const TCGArg a0,
     }
 
     if (is_add) {
-        insn = add_vec_insn[vece];
+        insn = add_vec_insn[lasx][vece];
     } else {
-        insn = sub_vec_insn[vece];
+        insn = sub_vec_insn[lasx][vece];
     }
     tcg_out32(s, encode_vdvjvk_insn(insn, a0, a1, a2));
 }
@@ -1963,10 +1967,10 @@  static void tcg_out_vec_op(TCGContext *s, TCGOpcode opc,
         }
         break;
     case INDEX_op_add_vec:
-        tcg_out_addsub_vec(s, vece, a0, a1, a2, const_args[2], true);
+        tcg_out_addsub_vec(s, false, vece, a0, a1, a2, const_args[2], true);
         break;
     case INDEX_op_sub_vec:
-        tcg_out_addsub_vec(s, vece, a0, a1, a2, const_args[2], false);
+        tcg_out_addsub_vec(s, false, vece, a0, a1, a2, const_args[2], false);
         break;
     case INDEX_op_neg_vec:
         tcg_out32(s, encode_vdvj_insn(neg_vec_insn[vece], a0, a1));