1. Patchew
  2. QEMU
  3. [PATCH v2 00/36] tcg 5.1 omnibus patch set
  4. View patch

[PATCH v2 24/36] tcg/i386: Use tcg_constant_vec with tcg vec expanders

Richard Henderson posted 36 patches 5 years, 9 months ago
Maintainers: Paolo Bonzini <pbonzini@redhat.com>, David Gibson <david@gibson.dropbear.id.au>, Richard Henderson <richard.henderson@linaro.org>, Richard Henderson <rth@twiddle.net>, David Hildenbrand <david@redhat.com>, Aurelien Jarno <aurelien@aurel32.net>, Andrzej Zaborowski <balrogg@gmail.com>, Alistair Francis <Alistair.Francis@wdc.com>, Palmer Dabbelt <palmer@dabbelt.com>, Stefan Weil <sw@weilnetz.de>, Aleksandar Rikalo <aleksandar.rikalo@rt-rk.com>, Cornelia Huck <cohuck@redhat.com>, Aleksandar Markovic <aleksandar.qemu.devel@gmail.com>
[PATCH v2 24/36] tcg/i386: Use tcg_constant_vec with tcg vec expanders
Posted by Richard Henderson 5 years, 9 months ago 
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/i386/tcg-target.inc.c | 26 +++++++++++++-------------
 1 file changed, 13 insertions(+), 13 deletions(-)

diff --git a/tcg/i386/tcg-target.inc.c b/tcg/i386/tcg-target.inc.c
index 9cb627d6eb..deace219d2 100644
--- a/tcg/i386/tcg-target.inc.c
+++ b/tcg/i386/tcg-target.inc.c
@@ -3452,7 +3452,7 @@ static void expand_vec_sari(TCGType type, unsigned vece,
 static void expand_vec_mul(TCGType type, unsigned vece,
                            TCGv_vec v0, TCGv_vec v1, TCGv_vec v2)
 {
-    TCGv_vec t1, t2, t3, t4;
+    TCGv_vec t1, t2, t3, t4, zero;
 
     tcg_debug_assert(vece == MO_8);
 
@@ -3470,11 +3470,11 @@ static void expand_vec_mul(TCGType type, unsigned vece,
     case TCG_TYPE_V64:
         t1 = tcg_temp_new_vec(TCG_TYPE_V128);
         t2 = tcg_temp_new_vec(TCG_TYPE_V128);
-        tcg_gen_dup16i_vec(t2, 0);
+        zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0);
         vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8,
-                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t2));
+                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero));
         vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8,
-                  tcgv_vec_arg(t2), tcgv_vec_arg(t2), tcgv_vec_arg(v2));
+                  tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2));
         tcg_gen_mul_vec(MO_16, t1, t1, t2);
         tcg_gen_shri_vec(MO_16, t1, t1, 8);
         vec_gen_3(INDEX_op_x86_packus_vec, TCG_TYPE_V128, MO_8,
@@ -3489,15 +3489,15 @@ static void expand_vec_mul(TCGType type, unsigned vece,
         t2 = tcg_temp_new_vec(type);
         t3 = tcg_temp_new_vec(type);
         t4 = tcg_temp_new_vec(type);
-        tcg_gen_dup16i_vec(t4, 0);
+        zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0);
         vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8,
-                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t4));
+                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero));
         vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8,
-                  tcgv_vec_arg(t2), tcgv_vec_arg(t4), tcgv_vec_arg(v2));
+                  tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2));
         vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8,
-                  tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(t4));
+                  tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(zero));
         vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8,
-                  tcgv_vec_arg(t4), tcgv_vec_arg(t4), tcgv_vec_arg(v2));
+                  tcgv_vec_arg(t4), tcgv_vec_arg(zero), tcgv_vec_arg(v2));
         tcg_gen_mul_vec(MO_16, t1, t1, t2);
         tcg_gen_mul_vec(MO_16, t3, t3, t4);
         tcg_gen_shri_vec(MO_16, t1, t1, 8);
@@ -3525,7 +3525,7 @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0,
         NEED_UMIN = 8,
         NEED_UMAX = 16,
     };
-    TCGv_vec t1, t2;
+    TCGv_vec t1, t2, t3;
     uint8_t fixup;
 
     switch (cond) {
@@ -3596,9 +3596,9 @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0,
     } else if (fixup & NEED_BIAS) {
         t1 = tcg_temp_new_vec(type);
         t2 = tcg_temp_new_vec(type);
-        tcg_gen_dupi_vec(vece, t2, 1ull << ((8 << vece) - 1));
-        tcg_gen_sub_vec(vece, t1, v1, t2);
-        tcg_gen_sub_vec(vece, t2, v2, t2);
+        t3 = tcg_constant_vec(type, vece, 1ull << ((8 << vece) - 1));
+        tcg_gen_sub_vec(vece, t1, v1, t3);
+        tcg_gen_sub_vec(vece, t2, v2, t3);
         v1 = t1;
         v2 = t2;
         cond = tcg_signed_cond(cond);
-- 
2.20.1
Re: [PATCH v2 24/36] tcg/i386: Use tcg_constant_vec with tcg vec expanders
Posted by Alex Bennée 5 years, 9 months ago 
Richard Henderson <richard.henderson@linaro.org> writes:

> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

> ---
>  tcg/i386/tcg-target.inc.c | 26 +++++++++++++-------------
>  1 file changed, 13 insertions(+), 13 deletions(-)
>
> diff --git a/tcg/i386/tcg-target.inc.c b/tcg/i386/tcg-target.inc.c
> index 9cb627d6eb..deace219d2 100644
> --- a/tcg/i386/tcg-target.inc.c
> +++ b/tcg/i386/tcg-target.inc.c
> @@ -3452,7 +3452,7 @@ static void expand_vec_sari(TCGType type, unsigned vece,
>  static void expand_vec_mul(TCGType type, unsigned vece,
>                             TCGv_vec v0, TCGv_vec v1, TCGv_vec v2)
>  {
> -    TCGv_vec t1, t2, t3, t4;
> +    TCGv_vec t1, t2, t3, t4, zero;
>  
>      tcg_debug_assert(vece == MO_8);
>  
> @@ -3470,11 +3470,11 @@ static void expand_vec_mul(TCGType type, unsigned vece,
>      case TCG_TYPE_V64:
>          t1 = tcg_temp_new_vec(TCG_TYPE_V128);
>          t2 = tcg_temp_new_vec(TCG_TYPE_V128);
> -        tcg_gen_dup16i_vec(t2, 0);
> +        zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0);
>          vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8,
> -                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t2));
> +                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero));
>          vec_gen_3(INDEX_op_x86_punpckl_vec, TCG_TYPE_V128, MO_8,
> -                  tcgv_vec_arg(t2), tcgv_vec_arg(t2), tcgv_vec_arg(v2));
> +                  tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2));
>          tcg_gen_mul_vec(MO_16, t1, t1, t2);
>          tcg_gen_shri_vec(MO_16, t1, t1, 8);
>          vec_gen_3(INDEX_op_x86_packus_vec, TCG_TYPE_V128, MO_8,
> @@ -3489,15 +3489,15 @@ static void expand_vec_mul(TCGType type, unsigned vece,
>          t2 = tcg_temp_new_vec(type);
>          t3 = tcg_temp_new_vec(type);
>          t4 = tcg_temp_new_vec(type);
> -        tcg_gen_dup16i_vec(t4, 0);
> +        zero = tcg_constant_vec(TCG_TYPE_V128, MO_8, 0);
>          vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8,
> -                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(t4));
> +                  tcgv_vec_arg(t1), tcgv_vec_arg(v1), tcgv_vec_arg(zero));
>          vec_gen_3(INDEX_op_x86_punpckl_vec, type, MO_8,
> -                  tcgv_vec_arg(t2), tcgv_vec_arg(t4), tcgv_vec_arg(v2));
> +                  tcgv_vec_arg(t2), tcgv_vec_arg(zero), tcgv_vec_arg(v2));
>          vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8,
> -                  tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(t4));
> +                  tcgv_vec_arg(t3), tcgv_vec_arg(v1), tcgv_vec_arg(zero));
>          vec_gen_3(INDEX_op_x86_punpckh_vec, type, MO_8,
> -                  tcgv_vec_arg(t4), tcgv_vec_arg(t4), tcgv_vec_arg(v2));
> +                  tcgv_vec_arg(t4), tcgv_vec_arg(zero), tcgv_vec_arg(v2));
>          tcg_gen_mul_vec(MO_16, t1, t1, t2);
>          tcg_gen_mul_vec(MO_16, t3, t3, t4);
>          tcg_gen_shri_vec(MO_16, t1, t1, 8);
> @@ -3525,7 +3525,7 @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0,
>          NEED_UMIN = 8,
>          NEED_UMAX = 16,
>      };
> -    TCGv_vec t1, t2;
> +    TCGv_vec t1, t2, t3;
>      uint8_t fixup;
>  
>      switch (cond) {
> @@ -3596,9 +3596,9 @@ static bool expand_vec_cmp_noinv(TCGType type, unsigned vece, TCGv_vec v0,
>      } else if (fixup & NEED_BIAS) {
>          t1 = tcg_temp_new_vec(type);
>          t2 = tcg_temp_new_vec(type);
> -        tcg_gen_dupi_vec(vece, t2, 1ull << ((8 << vece) - 1));
> -        tcg_gen_sub_vec(vece, t1, v1, t2);
> -        tcg_gen_sub_vec(vece, t2, v2, t2);
> +        t3 = tcg_constant_vec(type, vece, 1ull << ((8 << vece) - 1));
> +        tcg_gen_sub_vec(vece, t1, v1, t3);
> +        tcg_gen_sub_vec(vece, t2, v2, t3);
>          v1 = t1;
>          v2 = t2;
>          cond = tcg_signed_cond(cond);


-- 
Alex Bennée

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.