tcg: Add opcodes for vector minmax arithmetic

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

10 files changed, 164 insertions, 0 deletions

diff --git a/tcg/README b/tcg/README
index dd9fd9c86c..603f4df659 100644
--- a/tcg/README
+++ b/tcg/README
@@ -554,6 +554,16 @@ E.g. VECL=1 -> 64 << 1 -> v128, and VECE=2 -> 1 << 2 -> i32.
 
   Similarly, v0 = -v1.
 
+* smin_vec:
+* umin_vec:
+
+  Similarly, v0 = MIN(v1, v2), for signed and unsigned element types.
+
+* smax_vec:
+* umax_vec:
+
+  Similarly, v0 = MAX(v1, v2), for signed and unsigned element types.
+
 * ssadd_vec:
 * sssub_vec:
 * usadd_vec:
diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index 98556bcf22..545a6eec75 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -136,6 +136,7 @@ typedef enum {
 #define TCG_TARGET_HAS_cmp_vec          1
 #define TCG_TARGET_HAS_mul_vec          1
 #define TCG_TARGET_HAS_sat_vec          0
+#define TCG_TARGET_HAS_minmax_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 44381062e6..7bd7eae672 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -186,6 +186,7 @@ extern bool have_avx2;
 #define TCG_TARGET_HAS_cmp_vec          1
 #define TCG_TARGET_HAS_mul_vec          1
 #define TCG_TARGET_HAS_sat_vec          0
+#define TCG_TARGET_HAS_minmax_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 0a33f51065..3ee44fcb75 100644
--- a/tcg/tcg-op-gvec.c
+++ b/tcg/tcg-op-gvec.c
@@ -1810,6 +1810,114 @@ void tcg_gen_gvec_ussub(unsigned vece, uint32_t dofs, uint32_t aofs,
     tcg_gen_gvec_3(dofs, aofs, bofs, oprsz, maxsz, &g[vece]);
 }
 
+void tcg_gen_gvec_smin(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz)
+{
+    static const GVecGen3 g[4] = {
+        { .fniv = tcg_gen_smin_vec,
+          .fno = gen_helper_gvec_smin8,
+          .opc = INDEX_op_smin_vec,
+          .vece = MO_8 },
+        { .fniv = tcg_gen_smin_vec,
+          .fno = gen_helper_gvec_smin16,
+          .opc = INDEX_op_smin_vec,
+          .vece = MO_16 },
+        { .fni4 = tcg_gen_smin_i32,
+          .fniv = tcg_gen_smin_vec,
+          .fno = gen_helper_gvec_smin32,
+          .opc = INDEX_op_smin_vec,
+          .vece = MO_32 },
+        { .fni8 = tcg_gen_smin_i64,
+          .fniv = tcg_gen_smin_vec,
+          .fno = gen_helper_gvec_smin64,
+          .opc = INDEX_op_smin_vec,
+          .vece = MO_64 }
+    };
+    tcg_debug_assert(vece <= MO_64);
+    tcg_gen_gvec_3(dofs, aofs, bofs, oprsz, maxsz, &g[vece]);
+}
+
+void tcg_gen_gvec_umin(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz)
+{
+    static const GVecGen3 g[4] = {
+        { .fniv = tcg_gen_umin_vec,
+          .fno = gen_helper_gvec_umin8,
+          .opc = INDEX_op_umin_vec,
+          .vece = MO_8 },
+        { .fniv = tcg_gen_umin_vec,
+          .fno = gen_helper_gvec_umin16,
+          .opc = INDEX_op_umin_vec,
+          .vece = MO_16 },
+        { .fni4 = tcg_gen_umin_i32,
+          .fniv = tcg_gen_umin_vec,
+          .fno = gen_helper_gvec_umin32,
+          .opc = INDEX_op_umin_vec,
+          .vece = MO_32 },
+        { .fni8 = tcg_gen_umin_i64,
+          .fniv = tcg_gen_umin_vec,
+          .fno = gen_helper_gvec_umin64,
+          .opc = INDEX_op_umin_vec,
+          .vece = MO_64 }
+    };
+    tcg_debug_assert(vece <= MO_64);
+    tcg_gen_gvec_3(dofs, aofs, bofs, oprsz, maxsz, &g[vece]);
+}
+
+void tcg_gen_gvec_smax(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz)
+{
+    static const GVecGen3 g[4] = {
+        { .fniv = tcg_gen_smax_vec,
+          .fno = gen_helper_gvec_smax8,
+          .opc = INDEX_op_smax_vec,
+          .vece = MO_8 },
+        { .fniv = tcg_gen_smax_vec,
+          .fno = gen_helper_gvec_smax16,
+          .opc = INDEX_op_smax_vec,
+          .vece = MO_16 },
+        { .fni4 = tcg_gen_smax_i32,
+          .fniv = tcg_gen_smax_vec,
+          .fno = gen_helper_gvec_smax32,
+          .opc = INDEX_op_smax_vec,
+          .vece = MO_32 },
+        { .fni8 = tcg_gen_smax_i64,
+          .fniv = tcg_gen_smax_vec,
+          .fno = gen_helper_gvec_smax64,
+          .opc = INDEX_op_smax_vec,
+          .vece = MO_64 }
+    };
+    tcg_debug_assert(vece <= MO_64);
+    tcg_gen_gvec_3(dofs, aofs, bofs, oprsz, maxsz, &g[vece]);
+}
+
+void tcg_gen_gvec_umax(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz)
+{
+    static const GVecGen3 g[4] = {
+        { .fniv = tcg_gen_umax_vec,
+          .fno = gen_helper_gvec_umax8,
+          .opc = INDEX_op_umax_vec,
+          .vece = MO_8 },
+        { .fniv = tcg_gen_umax_vec,
+          .fno = gen_helper_gvec_umax16,
+          .opc = INDEX_op_umax_vec,
+          .vece = MO_16 },
+        { .fni4 = tcg_gen_umax_i32,
+          .fniv = tcg_gen_umax_vec,
+          .fno = gen_helper_gvec_umax32,
+          .opc = INDEX_op_umax_vec,
+          .vece = MO_32 },
+        { .fni8 = tcg_gen_umax_i64,
+          .fniv = tcg_gen_umax_vec,
+          .fno = gen_helper_gvec_umax64,
+          .opc = INDEX_op_umax_vec,
+          .vece = MO_64 }
+    };
+    tcg_debug_assert(vece <= MO_64);
+    tcg_gen_gvec_3(dofs, aofs, bofs, oprsz, maxsz, &g[vece]);
+}
+
 /* Perform a vector negation using normal negation and a mask.
    Compare gen_subv_mask above.  */
 static void gen_negv_mask(TCGv_i64 d, TCGv_i64 b, TCGv_i64 m)
diff --git a/tcg/tcg-op-gvec.h b/tcg/tcg-op-gvec.h
index 2cb447112e..4734eef7de 100644
--- a/tcg/tcg-op-gvec.h
+++ b/tcg/tcg-op-gvec.h
@@ -234,6 +234,16 @@ void tcg_gen_gvec_usadd(unsigned vece, uint32_t dofs, uint32_t aofs,
 void tcg_gen_gvec_ussub(unsigned vece, uint32_t dofs, uint32_t aofs,
                         uint32_t bofs, uint32_t oprsz, uint32_t maxsz);
 
+/* Min/max.  */
+void tcg_gen_gvec_smin(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz);
+void tcg_gen_gvec_umin(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz);
+void tcg_gen_gvec_smax(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz);
+void tcg_gen_gvec_umax(unsigned vece, uint32_t dofs, uint32_t aofs,
+                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz);
+
 void tcg_gen_gvec_and(unsigned vece, uint32_t dofs, uint32_t aofs,
                       uint32_t bofs, uint32_t oprsz, uint32_t maxsz);
 void tcg_gen_gvec_or(unsigned vece, uint32_t dofs, uint32_t aofs,
diff --git a/tcg/tcg-op-vec.c b/tcg/tcg-op-vec.c
index 675aa09258..36f35022ac 100644
--- a/tcg/tcg-op-vec.c
+++ b/tcg/tcg-op-vec.c
@@ -433,3 +433,23 @@ void tcg_gen_ussub_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b)
 {
     do_op3(vece, r, a, b, INDEX_op_ussub_vec);
 }
+
+void tcg_gen_smin_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b)
+{
+    do_op3(vece, r, a, b, INDEX_op_smin_vec);
+}
+
+void tcg_gen_umin_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b)
+{
+    do_op3(vece, r, a, b, INDEX_op_umin_vec);
+}
+
+void tcg_gen_smax_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b)
+{
+    do_op3(vece, r, a, b, INDEX_op_smax_vec);
+}
+
+void tcg_gen_umax_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b)
+{
+    do_op3(vece, r, a, b, INDEX_op_umax_vec);
+}
diff --git a/tcg/tcg-op.h b/tcg/tcg-op.h
index 4a93d730e8..2d98868d8f 100644
--- a/tcg/tcg-op.h
+++ b/tcg/tcg-op.h
@@ -971,6 +971,10 @@ void tcg_gen_ssadd_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
 void tcg_gen_usadd_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
 void tcg_gen_sssub_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
 void tcg_gen_ussub_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
+void tcg_gen_smin_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
+void tcg_gen_umin_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
+void tcg_gen_smax_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
+void tcg_gen_umax_vec(unsigned vece, TCGv_vec r, TCGv_vec a, TCGv_vec b);
 
 void tcg_gen_shli_vec(unsigned vece, TCGv_vec r, TCGv_vec a, int64_t i);
 void tcg_gen_shri_vec(unsigned vece, TCGv_vec r, TCGv_vec a, int64_t i);
diff --git a/tcg/tcg-opc.h b/tcg/tcg-opc.h
index 94b2ed80af..4e0238ad1a 100644
--- a/tcg/tcg-opc.h
+++ b/tcg/tcg-opc.h
@@ -226,6 +226,10 @@ DEF(ssadd_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_sat_vec))
 DEF(usadd_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_sat_vec))
 DEF(sssub_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_sat_vec))
 DEF(ussub_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_sat_vec))
+DEF(smin_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_minmax_vec))
+DEF(umin_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_minmax_vec))
+DEF(smax_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_minmax_vec))
+DEF(umax_vec, 1, 2, 0, IMPLVEC | IMPL(TCG_TARGET_HAS_minmax_vec))
 
 DEF(and_vec, 1, 2, 0, IMPLVEC)
 DEF(or_vec, 1, 2, 0, IMPLVEC)
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 6363f0cb29..20a5d8f315 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1612,6 +1612,11 @@ bool tcg_op_supported(TCGOpcode op)
     case INDEX_op_sssub_vec:
     case INDEX_op_ussub_vec:
         return have_vec && TCG_TARGET_HAS_sat_vec;
+    case INDEX_op_smin_vec:
+    case INDEX_op_umin_vec:
+    case INDEX_op_smax_vec:
+    case INDEX_op_umax_vec:
+        return have_vec && TCG_TARGET_HAS_minmax_vec;
 
     default:
         tcg_debug_assert(op > INDEX_op_last_generic && op < NB_OPS);
diff --git a/tcg/tcg.h b/tcg/tcg.h
index 5590dc512a..045c24a357 100644
--- a/tcg/tcg.h
+++ b/tcg/tcg.h
@@ -184,6 +184,7 @@ typedef uint64_t TCGRegSet;
 #define TCG_TARGET_HAS_shv_vec          0
 #define TCG_TARGET_HAS_mul_vec          0
 #define TCG_TARGET_HAS_sat_vec          0
+#define TCG_TARGET_HAS_minmax_vec       0
 #else
 #define TCG_TARGET_MAYBE_vec            1
 #endif