tcg: Add support for vector bitwise select

This operation performs d = (b & a) | (c & ~a), and is present
on a majority of host vector units.  Include gvec expanders.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
diff --git a/tcg/README b/tcg/README
index cbdfd3b..76057ab 100644
--- a/tcg/README
+++ b/tcg/README
@@ -627,6 +627,10 @@
 
   Compare vectors by element, storing -1 for true and 0 for false.
 
+* bitsel_vec v0, v1, v2, v3
+
+  Bitwise select, v0 = (v2 & v1) | (v3 & ~v1), across the entire vector.
+
 *********
 
 Note 1: Some shortcuts are defined when the last operand is known to be
diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index e43554c..52ee664 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -140,6 +140,7 @@
 #define TCG_TARGET_HAS_mul_vec          1
 #define TCG_TARGET_HAS_sat_vec          1
 #define TCG_TARGET_HAS_minmax_vec       1
+#define TCG_TARGET_HAS_bitsel_vec       0
 
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index 66f16fb..08a0386 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -190,6 +190,7 @@
 #define TCG_TARGET_HAS_mul_vec          1
 #define TCG_TARGET_HAS_sat_vec          1
 #define TCG_TARGET_HAS_minmax_vec       1
+#define TCG_TARGET_HAS_bitsel_vec       0
 
 #define TCG_TARGET_deposit_i32_valid(ofs, len) \
     (((ofs) == 0 && (len) == 8) || ((ofs) == 8 && (len) == 8) || \
diff --git a/tcg/tcg-op-gvec.c b/tcg/tcg-op-gvec.c
index bbf70e3..f18464c 100644
--- a/tcg/tcg-op-gvec.c
+++ b/tcg/tcg-op-gvec.c
@@ -3195,3 +3195,26 @@
         expand_clr(dofs + oprsz, maxsz - oprsz);
     }
 }
+
+static void tcg_gen_bitsel_i64(TCGv_i64 d, TCGv_i64 a, TCGv_i64 b, TCGv_i64 c)
+{
+    TCGv_i64 t = tcg_temp_new_i64();
+
+    tcg_gen_and_i64(t, b, a);
+    tcg_gen_andc_i64(d, c, a);
+    tcg_gen_or_i64(d, d, t);
+    tcg_temp_free_i64(t);
+}
+
+void tcg_gen_gvec_bitsel(unsigned vece, uint32_t dofs, uint32_t aofs,
+                         uint32_t bofs, uint32_t cofs,
+                         uint32_t oprsz, uint32_t maxsz)
+{
+    static const GVecGen4 g = {
+        .fni8 = tcg_gen_bitsel_i64,
+        .fniv = tcg_gen_bitsel_vec,
+        .fno = gen_helper_gvec_bitsel,
+    };
+
+    tcg_gen_gvec_4(dofs, aofs, bofs, cofs, oprsz, maxsz, &g);
+}
diff --git a/tcg/tcg-op-gvec.h b/tcg/tcg-op-gvec.h
index 52a398c..2a9e0c7 100644
--- a/tcg/tcg-op-gvec.h
+++ b/tcg/tcg-op-gvec.h
@@ -343,6 +343,13 @@
                       uint32_t oprsz, uint32_t maxsz);
 
 /*
+ * Perform vector bit select: d = (b & a) | (c & ~a).
+ */
+void tcg_gen_gvec_bitsel(unsigned vece, uint32_t dofs, uint32_t aofs,
+                         uint32_t bofs, uint32_t cofs,
+                         uint32_t oprsz, uint32_t maxsz);
+
+/*
  * 64-bit vector operations.  Use these when the register has been allocated
  * with tcg_global_mem_new_i64, and so we cannot also address it via pointer.
  * OPRSZ = MAXSZ = 8.
diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c
index 543508d..99cbf29 100644
--- a/tcg/tcg-op-vec.c
+++ b/tcg/tcg-op-vec.c
@@ -88,6 +88,7 @@
         case INDEX_op_dup2_vec:
         case INDEX_op_ld_vec:
         case INDEX_op_st_vec:
+        case INDEX_op_bitsel_vec:
             /* These opcodes are mandatory and should not be listed.  */
             g_assert_not_reached();
         default:
@@ -691,3 +692,28 @@
 {
     do_shifts(vece, r, a, b, INDEX_op_sars_vec, INDEX_op_sarv_vec);
 }
+
+void tcg_gen_bitsel_vec(unsigned vece, TCGv_vec r, TCGv_vec a,
+                        TCGv_vec b, TCGv_vec c)
+{
+    TCGTemp *rt = tcgv_vec_temp(r);
+    TCGTemp *at = tcgv_vec_temp(a);
+    TCGTemp *bt = tcgv_vec_temp(b);
+    TCGTemp *ct = tcgv_vec_temp(c);
+    TCGType type = rt->base_type;
+
+    tcg_debug_assert(at->base_type >= type);
+    tcg_debug_assert(bt->base_type >= type);
+    tcg_debug_assert(ct->base_type >= type);
+
+    if (TCG_TARGET_HAS_bitsel_vec) {
+        vec_gen_4(INDEX_op_bitsel_vec, type, MO_8,
+                  temp_arg(rt), temp_arg(at), temp_arg(bt), temp_arg(ct));
+    } else {
+        TCGv_vec t = tcg_temp_new_vec(type);
+        tcg_gen_and_vec(MO_8, t, a, b);
+        tcg_gen_andc_vec(MO_8, r, c, a);
+        tcg_gen_or_vec(MO_8, r, r, t);
+        tcg_temp_free_vec(t);
+    }
+}
diff --git a/tcg/tcg-op.h b/tcg/tcg-op.h
index 660fe20..268860e 100644
--- a/tcg/tcg-op.h
+++ b/tcg/tcg-op.h
@@ -1000,6 +1000,9 @@
 void tcg_gen_cmp_vec(TCGCond cond, unsigned vece, TCGv_vec r,
                      TCGv_vec a, TCGv_vec b);
 
+void tcg_gen_bitsel_vec(unsigned vece, TCGv_vec r, TCGv_vec a,
+                        TCGv_vec b, TCGv_vec c);
+
 void tcg_gen_ld_vec(TCGv_vec r, TCGv_ptr base, TCGArg offset);
 void tcg_gen_st_vec(TCGv_vec r, TCGv_ptr base, TCGArg offset);
 void tcg_gen_stl_vec(TCGv_vec r, TCGv_ptr base, TCGArg offset, TCGType t);
diff --git a/tcg/tcg-opc.h b/tcg/tcg-opc.h
index 4a2dd11..c05b714 100644
--- a/tcg/tcg-opc.h
+++ b/tcg/tcg-opc.h
@@ -256,6 +256,8 @@
 
 DEF(cmp_vec, 1, 2, 1, IMPLVEC)
 
+DEF(bitsel_vec, 1, 3, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_bitsel_vec))
+
 DEF(last_generic, 0, 0, 0, TCG_OPF_NOT_PRESENT)
 
 #if TCG_TARGET_MAYBE_vec
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 24083b8..5d947db 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1646,6 +1646,8 @@
     case INDEX_op_smax_vec:
     case INDEX_op_umax_vec:
         return have_vec && TCG_TARGET_HAS_minmax_vec;
+    case INDEX_op_bitsel_vec:
+        return have_vec && TCG_TARGET_HAS_bitsel_vec;
 
     default:
         tcg_debug_assert(op > INDEX_op_last_generic && op < NB_OPS);
diff --git a/tcg/tcg.h b/tcg/tcg.h
index 0e01a70..72f9f6c 100644
--- a/tcg/tcg.h
+++ b/tcg/tcg.h
@@ -187,6 +187,7 @@
 #define TCG_TARGET_HAS_mul_vec          0
 #define TCG_TARGET_HAS_sat_vec          0
 #define TCG_TARGET_HAS_minmax_vec       0
+#define TCG_TARGET_HAS_bitsel_vec       0
 #else
 #define TCG_TARGET_MAYBE_vec            1
 #endif