tcg: Use tcg_constant_{i32,i64,vec} with gvec expanders

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

1 files changed, 50 insertions, 77 deletions

diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c
index 1a41dfa908..498a959839 100644
--- a/tcg/tcg-op-gvec.c
+++ b/tcg/tcg-op-gvec.c
@@ -115,7 +115,7 @@ void tcg_gen_gvec_2_ool(uint32_t dofs, uint32_t aofs,
                         gen_helper_gvec_2 *fn)
 {
     TCGv_ptr a0, a1;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -127,7 +127,6 @@ void tcg_gen_gvec_2_ool(uint32_t dofs, uint32_t aofs,
 
     tcg_temp_free_ptr(a0);
     tcg_temp_free_ptr(a1);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with two vector operands
@@ -137,7 +136,7 @@ void tcg_gen_gvec_2i_ool(uint32_t dofs, uint32_t aofs, TCGv_i64 c,
                          gen_helper_gvec_2i *fn)
 {
     TCGv_ptr a0, a1;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -149,7 +148,6 @@ void tcg_gen_gvec_2i_ool(uint32_t dofs, uint32_t aofs, TCGv_i64 c,
 
     tcg_temp_free_ptr(a0);
     tcg_temp_free_ptr(a1);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with three vector operands.  */
@@ -158,7 +156,7 @@ void tcg_gen_gvec_3_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs,
                         gen_helper_gvec_3 *fn)
 {
     TCGv_ptr a0, a1, a2;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -173,7 +171,6 @@ void tcg_gen_gvec_3_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs,
     tcg_temp_free_ptr(a0);
     tcg_temp_free_ptr(a1);
     tcg_temp_free_ptr(a2);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with four vector operands.  */
@@ -182,7 +179,7 @@ void tcg_gen_gvec_4_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs,
                         int32_t data, gen_helper_gvec_4 *fn)
 {
     TCGv_ptr a0, a1, a2, a3;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -200,7 +197,6 @@ void tcg_gen_gvec_4_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs,
     tcg_temp_free_ptr(a1);
     tcg_temp_free_ptr(a2);
     tcg_temp_free_ptr(a3);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with five vector operands.  */
@@ -209,7 +205,7 @@ void tcg_gen_gvec_5_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs,
                         uint32_t maxsz, int32_t data, gen_helper_gvec_5 *fn)
 {
     TCGv_ptr a0, a1, a2, a3, a4;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -230,7 +226,6 @@ void tcg_gen_gvec_5_ool(uint32_t dofs, uint32_t aofs, uint32_t bofs,
     tcg_temp_free_ptr(a2);
     tcg_temp_free_ptr(a3);
     tcg_temp_free_ptr(a4);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with three vector operands
@@ -240,7 +235,7 @@ void tcg_gen_gvec_2_ptr(uint32_t dofs, uint32_t aofs,
                         int32_t data, gen_helper_gvec_2_ptr *fn)
 {
     TCGv_ptr a0, a1;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -252,7 +247,6 @@ void tcg_gen_gvec_2_ptr(uint32_t dofs, uint32_t aofs,
 
     tcg_temp_free_ptr(a0);
     tcg_temp_free_ptr(a1);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with three vector operands
@@ -262,7 +256,7 @@ void tcg_gen_gvec_3_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs,
                         int32_t data, gen_helper_gvec_3_ptr *fn)
 {
     TCGv_ptr a0, a1, a2;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -277,7 +271,6 @@ void tcg_gen_gvec_3_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs,
     tcg_temp_free_ptr(a0);
     tcg_temp_free_ptr(a1);
     tcg_temp_free_ptr(a2);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with four vector operands
@@ -288,7 +281,7 @@ void tcg_gen_gvec_4_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs,
                         gen_helper_gvec_4_ptr *fn)
 {
     TCGv_ptr a0, a1, a2, a3;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -306,7 +299,6 @@ void tcg_gen_gvec_4_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs,
     tcg_temp_free_ptr(a1);
     tcg_temp_free_ptr(a2);
     tcg_temp_free_ptr(a3);
-    tcg_temp_free_i32(desc);
 }
 
 /* Generate a call to a gvec-style helper with five vector operands
@@ -317,7 +309,7 @@ void tcg_gen_gvec_5_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs,
                         gen_helper_gvec_5_ptr *fn)
 {
     TCGv_ptr a0, a1, a2, a3, a4;
-    TCGv_i32 desc = tcg_const_i32(simd_desc(oprsz, maxsz, data));
+    TCGv_i32 desc = tcg_constant_i32(simd_desc(oprsz, maxsz, data));
 
     a0 = tcg_temp_new_ptr();
     a1 = tcg_temp_new_ptr();
@@ -338,7 +330,6 @@ void tcg_gen_gvec_5_ptr(uint32_t dofs, uint32_t aofs, uint32_t bofs,
     tcg_temp_free_ptr(a2);
     tcg_temp_free_ptr(a3);
     tcg_temp_free_ptr(a4);
-    tcg_temp_free_i32(desc);
 }
 
 /* Return true if we want to implement something of OPRSZ bytes
@@ -605,9 +596,9 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz,
                 || (TCG_TARGET_REG_BITS == 64
                     && (in_c == 0 || in_c == -1
                         || !check_size_impl(oprsz, 4)))) {
-                t_64 = tcg_const_i64(in_c);
+                t_64 = tcg_constant_i64(in_c);
             } else {
-                t_32 = tcg_const_i32(in_c);
+                t_32 = tcg_constant_i32(in_c);
             }
         }
 
@@ -648,11 +639,11 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz,
             t_val = tcg_temp_new_i32();
             tcg_gen_extrl_i64_i32(t_val, in_64);
         } else {
-            t_val = tcg_const_i32(in_c);
+            t_val = tcg_constant_i32(in_c);
         }
         gen_helper_memset(t_ptr, t_ptr, t_val, t_size);
 
-        if (!in_32) {
+        if (in_64) {
             tcg_temp_free_i32(t_val);
         }
         tcg_temp_free_ptr(t_size);
@@ -660,15 +651,14 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz,
         return;
     }
 
-    t_desc = tcg_const_i32(simd_desc(oprsz, maxsz, 0));
+    t_desc = tcg_constant_i32(simd_desc(oprsz, maxsz, 0));
 
     if (vece == MO_64) {
         if (in_64) {
             gen_helper_gvec_dup64(t_ptr, t_desc, in_64);
         } else {
-            t_64 = tcg_const_i64(in_c);
+            t_64 = tcg_constant_i64(in_c);
             gen_helper_gvec_dup64(t_ptr, t_desc, t_64);
-            tcg_temp_free_i64(t_64);
         }
     } else {
         typedef void dup_fn(TCGv_ptr, TCGv_i32, TCGv_i32);
@@ -680,24 +670,23 @@ static void do_dup(unsigned vece, uint32_t dofs, uint32_t oprsz,
 
         if (in_32) {
             fns[vece](t_ptr, t_desc, in_32);
-        } else {
+        } else if (in_64) {
             t_32 = tcg_temp_new_i32();
-            if (in_64) {
-                tcg_gen_extrl_i64_i32(t_32, in_64);
-            } else if (vece == MO_8) {
-                tcg_gen_movi_i32(t_32, in_c & 0xff);
+            tcg_gen_extrl_i64_i32(t_32, in_64);
+            fns[vece](t_ptr, t_desc, t_32);
+            tcg_temp_free_i32(t_32);
+        } else {
+            if (vece == MO_8) {
+                in_c &= 0xff;
             } else if (vece == MO_16) {
-                tcg_gen_movi_i32(t_32, in_c & 0xffff);
-            } else {
-                tcg_gen_movi_i32(t_32, in_c);
+                in_c &= 0xffff;
             }
+            t_32 = tcg_constant_i32(in_c);
             fns[vece](t_ptr, t_desc, t_32);
-            tcg_temp_free_i32(t_32);
         }
     }
 
     tcg_temp_free_ptr(t_ptr);
-    tcg_temp_free_i32(t_desc);
     return;
 
  done:
@@ -1247,10 +1236,9 @@ void tcg_gen_gvec_2i(uint32_t dofs, uint32_t aofs, uint32_t oprsz,
             if (g->fno) {
                 tcg_gen_gvec_2_ool(dofs, aofs, oprsz, maxsz, c, g->fno);
             } else {
-                TCGv_i64 tcg_c = tcg_const_i64(c);
+                TCGv_i64 tcg_c = tcg_constant_i64(c);
                 tcg_gen_gvec_2i_ool(dofs, aofs, tcg_c, oprsz,
                                     maxsz, c, g->fnoi);
-                tcg_temp_free_i64(tcg_c);
             }
             oprsz = maxsz;
         }
@@ -1744,16 +1732,14 @@ static void gen_addv_mask(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 m)
 
 void tcg_gen_vec_add8_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
 {
-    TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80));
+    TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80));
     gen_addv_mask(d, a, b, m);
-    tcg_temp_free_i64(m);
 }
 
 void tcg_gen_vec_add16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
 {
-    TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000));
+    TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000));
     gen_addv_mask(d, a, b, m);
-    tcg_temp_free_i64(m);
 }
 
 void tcg_gen_vec_add32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
@@ -1837,9 +1823,8 @@ void tcg_gen_gvec_adds(unsigned vece, uint32_t dofs, uint32_t aofs,
 void tcg_gen_gvec_addi(unsigned vece, uint32_t dofs, uint32_t aofs,
                        int64_t c, uint32_t oprsz, uint32_t maxsz)
 {
-    TCGv_i64 tmp = tcg_const_i64(c);
+    TCGv_i64 tmp = tcg_constant_i64(c);
     tcg_gen_gvec_adds(vece, dofs, aofs, tmp, oprsz, maxsz);
-    tcg_temp_free_i64(tmp);
 }
 
 static const TCGOpcode vecop_list_sub[] = { INDEX_op_sub_vec, 0 };
@@ -1897,16 +1882,14 @@ static void gen_subv_mask(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 m)
 
 void tcg_gen_vec_sub8_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
 {
-    TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80));
+    TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80));
     gen_subv_mask(d, a, b, m);
-    tcg_temp_free_i64(m);
 }
 
 void tcg_gen_vec_sub16_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
 {
-    TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000));
+    TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000));
     gen_subv_mask(d, a, b, m);
-    tcg_temp_free_i64(m);
 }
 
 void tcg_gen_vec_sub32_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
@@ -2017,9 +2000,8 @@ void tcg_gen_gvec_muls(unsigned vece, uint32_t dofs, uint32_t aofs,
 void tcg_gen_gvec_muli(unsigned vece, uint32_t dofs, uint32_t aofs,
                        int64_t c, uint32_t oprsz, uint32_t maxsz)
 {
-    TCGv_i64 tmp = tcg_const_i64(c);
+    TCGv_i64 tmp = tcg_constant_i64(c);
     tcg_gen_gvec_muls(vece, dofs, aofs, tmp, oprsz, maxsz);
-    tcg_temp_free_i64(tmp);
 }
 
 void tcg_gen_gvec_ssadd(unsigned vece, uint32_t dofs, uint32_t aofs,
@@ -2076,18 +2058,16 @@ void tcg_gen_gvec_sssub(unsigned vece, uint32_t dofs, uint32_t aofs,
 
 static void tcg_gen_usadd_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b)
 {
-    TCGv_i32 max = tcg_const_i32(-1);
+    TCGv_i32 max = tcg_constant_i32(-1);
     tcg_gen_add_i32(d, a, b);
     tcg_gen_movcond_i32(TCG_COND_LTU, d, d, a, max, d);
-    tcg_temp_free_i32(max);
 }
 
 static void tcg_gen_usadd_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
 {
-    TCGv_i64 max = tcg_const_i64(-1);
+    TCGv_i64 max = tcg_constant_i64(-1);
     tcg_gen_add_i64(d, a, b);
     tcg_gen_movcond_i64(TCG_COND_LTU, d, d, a, max, d);
-    tcg_temp_free_i64(max);
 }
 
 void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs,
@@ -2120,18 +2100,16 @@ void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs,
 
 static void tcg_gen_ussub_i32(TCGv_i32 d, TCGv_i32 a, TCGv_i32 b)
 {
-    TCGv_i32 min = tcg_const_i32(0);
+    TCGv_i32 min = tcg_constant_i32(0);
     tcg_gen_sub_i32(d, a, b);
     tcg_gen_movcond_i32(TCG_COND_LTU, d, a, b, min, d);
-    tcg_temp_free_i32(min);
 }
 
 static void tcg_gen_ussub_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b)
 {
-    TCGv_i64 min = tcg_const_i64(0);
+    TCGv_i64 min = tcg_constant_i64(0);
     tcg_gen_sub_i64(d, a, b);
     tcg_gen_movcond_i64(TCG_COND_LTU, d, a, b, min, d);
-    tcg_temp_free_i64(min);
 }
 
 void tcg_gen_gvec_ussub(unsigned vece, uint32_t dofs, uint32_t aofs,
@@ -2292,16 +2270,14 @@ static void gen_negv_mask(TCGv_i64 d, TCGv_i64 b, TCGv_i64 m)
 
 void tcg_gen_vec_neg8_i64(TCGv_i64 d, TCGv_i64 b)
 {
-    TCGv_i64 m = tcg_const_i64(dup_const(MO_8, 0x80));
+    TCGv_i64 m = tcg_constant_i64(dup_const(MO_8, 0x80));
     gen_negv_mask(d, b, m);
-    tcg_temp_free_i64(m);
 }
 
 void tcg_gen_vec_neg16_i64(TCGv_i64 d, TCGv_i64 b)
 {
-    TCGv_i64 m = tcg_const_i64(dup_const(MO_16, 0x8000));
+    TCGv_i64 m = tcg_constant_i64(dup_const(MO_16, 0x8000));
     gen_negv_mask(d, b, m);
-    tcg_temp_free_i64(m);
 }
 
 void tcg_gen_vec_neg32_i64(TCGv_i64 d, TCGv_i64 b)
@@ -2570,9 +2546,8 @@ void tcg_gen_gvec_ands(unsigned vece, uint32_t dofs, uint32_t aofs,
 void tcg_gen_gvec_andi(unsigned vece, uint32_t dofs, uint32_t aofs,
                        int64_t c, uint32_t oprsz, uint32_t maxsz)
 {
-    TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c));
+    TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c));
     tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_ands);
-    tcg_temp_free_i64(tmp);
 }
 
 static const GVecGen2s gop_xors = {
@@ -2595,9 +2570,8 @@ void tcg_gen_gvec_xors(unsigned vece, uint32_t dofs, uint32_t aofs,
 void tcg_gen_gvec_xori(unsigned vece, uint32_t dofs, uint32_t aofs,
                        int64_t c, uint32_t oprsz, uint32_t maxsz)
 {
-    TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c));
+    TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c));
     tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_xors);
-    tcg_temp_free_i64(tmp);
 }
 
 static const GVecGen2s gop_ors = {
@@ -2620,9 +2594,8 @@ void tcg_gen_gvec_ors(unsigned vece, uint32_t dofs, uint32_t aofs,
 void tcg_gen_gvec_ori(unsigned vece, uint32_t dofs, uint32_t aofs,
                       int64_t c, uint32_t oprsz, uint32_t maxsz)
 {
-    TCGv_i64 tmp = tcg_const_i64(dup_const(vece, c));
+    TCGv_i64 tmp = tcg_constant_i64(dup_const(vece, c));
     tcg_gen_gvec_2s(dofs, aofs, oprsz, maxsz, tmp, &gop_ors);
-    tcg_temp_free_i64(tmp);
 }
 
 void tcg_gen_vec_shl8i_i64(TCGv_i64 d, TCGv_i64 a, int64_t c)
@@ -3110,9 +3083,9 @@ static void tcg_gen_shlv_mod_vec(unsigned vece, TCGv_vec d,
                                  TCGv_vec a, TCGv_vec b)
 {
     TCGv_vec t = tcg_temp_new_vec_matching(d);
+    TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1);
 
-    tcg_gen_dupi_vec(vece, t, (8 << vece) - 1);
-    tcg_gen_and_vec(vece, t, t, b);
+    tcg_gen_and_vec(vece, t, b, m);
     tcg_gen_shlv_vec(vece, d, a, t);
     tcg_temp_free_vec(t);
 }
@@ -3173,9 +3146,9 @@ static void tcg_gen_shrv_mod_vec(unsigned vece, TCGv_vec d,
                                  TCGv_vec a, TCGv_vec b)
 {
     TCGv_vec t = tcg_temp_new_vec_matching(d);
+    TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1);
 
-    tcg_gen_dupi_vec(vece, t, (8 << vece) - 1);
-    tcg_gen_and_vec(vece, t, t, b);
+    tcg_gen_and_vec(vece, t, b, m);
     tcg_gen_shrv_vec(vece, d, a, t);
     tcg_temp_free_vec(t);
 }
@@ -3236,9 +3209,9 @@ static void tcg_gen_sarv_mod_vec(unsigned vece, TCGv_vec d,
                                  TCGv_vec a, TCGv_vec b)
 {
     TCGv_vec t = tcg_temp_new_vec_matching(d);
+    TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1);
 
-    tcg_gen_dupi_vec(vece, t, (8 << vece) - 1);
-    tcg_gen_and_vec(vece, t, t, b);
+    tcg_gen_and_vec(vece, t, b, m);
     tcg_gen_sarv_vec(vece, d, a, t);
     tcg_temp_free_vec(t);
 }
@@ -3299,9 +3272,9 @@ static void tcg_gen_rotlv_mod_vec(unsigned vece, TCGv_vec d,
                                   TCGv_vec a, TCGv_vec b)
 {
     TCGv_vec t = tcg_temp_new_vec_matching(d);
+    TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1);
 
-    tcg_gen_dupi_vec(vece, t, (8 << vece) - 1);
-    tcg_gen_and_vec(vece, t, t, b);
+    tcg_gen_and_vec(vece, t, b, m);
     tcg_gen_rotlv_vec(vece, d, a, t);
     tcg_temp_free_vec(t);
 }
@@ -3358,9 +3331,9 @@ static void tcg_gen_rotrv_mod_vec(unsigned vece, TCGv_vec d,
                                   TCGv_vec a, TCGv_vec b)
 {
     TCGv_vec t = tcg_temp_new_vec_matching(d);
+    TCGv_vec m = tcg_constant_vec_matching(d, vece, (8 << vece) - 1);
 
-    tcg_gen_dupi_vec(vece, t, (8 << vece) - 1);
-    tcg_gen_and_vec(vece, t, t, b);
+    tcg_gen_and_vec(vece, t, b, m);
     tcg_gen_rotrv_vec(vece, d, a, t);
     tcg_temp_free_vec(t);
 }