tcg: Add muluh and mulsh opcodes

Use them in places where mulu2 and muls2 are used.
Optimize mulx2 with dead low part to mulxh.

Reviewed-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Richard Henderson <rth@twiddle.net>

tcg/aarch64/tcg-target.h

@@ -61,6 +61,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i32         0
 #define TCG_TARGET_HAS_mulu2_i32        0
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #define TCG_TARGET_HAS_div_i64          0
 #define TCG_TARGET_HAS_rem_i64          0
@@ -87,6 +89,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i64         0
 #define TCG_TARGET_HAS_mulu2_i64        0
 #define TCG_TARGET_HAS_muls2_i64        0
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 
 enum {
     TCG_AREG0 = TCG_REG_X19,

tcg/arm/tcg-target.h

@@ -80,6 +80,8 @@ extern bool use_idiv_instructions;
 #define TCG_TARGET_HAS_deposit_i32      1
 #define TCG_TARGET_HAS_movcond_i32      1
 #define TCG_TARGET_HAS_muls2_i32        1
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 #define TCG_TARGET_HAS_div_i32          use_idiv_instructions
 #define TCG_TARGET_HAS_rem_i32          0
 

tcg/hppa/tcg-target.h

@@ -100,6 +100,8 @@ typedef enum {
 #define TCG_TARGET_HAS_deposit_i32      1
 #define TCG_TARGET_HAS_movcond_i32      1
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 /* optional instructions automatically implemented */
 #define TCG_TARGET_HAS_neg_i32          0 /* sub rd, 0, rs */

tcg/i386/tcg-target.h

@@ -96,6 +96,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i32         1
 #define TCG_TARGET_HAS_mulu2_i32        1
 #define TCG_TARGET_HAS_muls2_i32        1
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #if TCG_TARGET_REG_BITS == 64
 #define TCG_TARGET_HAS_div2_i64         1
@@ -122,6 +124,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i64         1
 #define TCG_TARGET_HAS_mulu2_i64        1
 #define TCG_TARGET_HAS_muls2_i64        1
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 #endif
 
 #define TCG_TARGET_deposit_i32_valid(ofs, len) \

tcg/ia64/tcg-target.h

@@ -146,6 +146,10 @@ typedef enum {
 #define TCG_TARGET_HAS_mulu2_i64        0
 #define TCG_TARGET_HAS_muls2_i32        0
 #define TCG_TARGET_HAS_muls2_i64        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i32        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 
 #define TCG_TARGET_deposit_i32_valid(ofs, len) ((len) <= 16)
 #define TCG_TARGET_deposit_i64_valid(ofs, len) ((len) <= 16)

tcg/mips/tcg-target.h

@@ -89,6 +89,8 @@ typedef enum {
 #define TCG_TARGET_HAS_eqv_i32          0
 #define TCG_TARGET_HAS_nand_i32         0
 #define TCG_TARGET_HAS_muls2_i32        1
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 /* optional instructions only implemented on MIPS4, MIPS32 and Loongson 2 */
 #if (defined(__mips_isa_rev) && (__mips_isa_rev >= 1)) || \

tcg/optimize.c

@@ -198,6 +198,8 @@ static TCGOpcode op_to_mov(TCGOpcode op)
 
 static TCGArg do_constant_folding_2(TCGOpcode op, TCGArg x, TCGArg y)
 {
+    uint64_t l64, h64;
+
     switch (op) {
     CASE_OP_32_64(add):
         return x + y;
@@ -290,6 +292,18 @@ static TCGArg do_constant_folding_2(TCGOpcode op, TCGArg x, TCGArg y)
     case INDEX_op_ext32u_i64:
         return (uint32_t)x;
 
+    case INDEX_op_muluh_i32:
+        return ((uint64_t)(uint32_t)x * (uint32_t)y) >> 32;
+    case INDEX_op_mulsh_i32:
+        return ((int64_t)(int32_t)x * (int32_t)y) >> 32;
+
+    case INDEX_op_muluh_i64:
+        mulu64(&l64, &h64, x, y);
+        return h64;
+    case INDEX_op_mulsh_i64:
+        muls64(&l64, &h64, x, y);
+        return h64;
+
     default:
         fprintf(stderr,
                 "Unrecognized operation %d in do_constant_folding.\n", op);
@@ -531,6 +545,8 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr,
         CASE_OP_32_64(eqv):
         CASE_OP_32_64(nand):
         CASE_OP_32_64(nor):
+        CASE_OP_32_64(muluh):
+        CASE_OP_32_64(mulsh):
             swap_commutative(args[0], &args[1], &args[2]);
             break;
         CASE_OP_32_64(brcond):
@@ -771,6 +787,8 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr,
         switch (op) {
         CASE_OP_32_64(and):
         CASE_OP_32_64(mul):
+        CASE_OP_32_64(muluh):
+        CASE_OP_32_64(mulsh):
             if ((temps[args[2]].state == TCG_TEMP_CONST
                 && temps[args[2]].val == 0)) {
                 s->gen_opc_buf[op_index] = op_to_movi(op);
@@ -882,6 +900,8 @@ static TCGArg *tcg_constant_folding(TCGContext *s, uint16_t *tcg_opc_ptr,
         CASE_OP_32_64(eqv):
         CASE_OP_32_64(nand):
         CASE_OP_32_64(nor):
+        CASE_OP_32_64(muluh):
+        CASE_OP_32_64(mulsh):
             if (temps[args[1]].state == TCG_TEMP_CONST
                 && temps[args[2]].state == TCG_TEMP_CONST) {
                 s->gen_opc_buf[op_index] = op_to_movi(op);

tcg/ppc/tcg-target.h

@@ -96,6 +96,8 @@ typedef enum {
 #define TCG_TARGET_HAS_deposit_i32      1
 #define TCG_TARGET_HAS_movcond_i32      1
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #define TCG_AREG0 TCG_REG_R27
 

tcg/ppc64/tcg-target.h

@@ -95,6 +95,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i32         0
 #define TCG_TARGET_HAS_mulu2_i32        0
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #define TCG_TARGET_HAS_div_i64          1
 #define TCG_TARGET_HAS_rem_i64          0
@@ -118,6 +120,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i64         1
 #define TCG_TARGET_HAS_mulu2_i64        1
 #define TCG_TARGET_HAS_muls2_i64        1
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 
 #define TCG_AREG0 TCG_REG_R27
 

tcg/s390/tcg-target.h

@@ -69,6 +69,8 @@ typedef enum TCGReg {
 #define TCG_TARGET_HAS_sub2_i32         1
 #define TCG_TARGET_HAS_mulu2_i32        0
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #define TCG_TARGET_HAS_div2_i64         1
 #define TCG_TARGET_HAS_rot_i64          1
@@ -94,6 +96,8 @@ typedef enum TCGReg {
 #define TCG_TARGET_HAS_sub2_i64         1
 #define TCG_TARGET_HAS_mulu2_i64        1
 #define TCG_TARGET_HAS_muls2_i64        0
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 
 extern bool tcg_target_deposit_valid(int ofs, int len);
 #define TCG_TARGET_deposit_i32_valid  tcg_target_deposit_valid

tcg/sparc/tcg-target.h

@@ -107,6 +107,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i32         1
 #define TCG_TARGET_HAS_mulu2_i32        1
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #if TCG_TARGET_REG_BITS == 64
 #define TCG_TARGET_HAS_div_i64          1
@@ -134,6 +136,8 @@ typedef enum {
 #define TCG_TARGET_HAS_sub2_i64         0
 #define TCG_TARGET_HAS_mulu2_i64        0
 #define TCG_TARGET_HAS_muls2_i64        0
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 #endif
 
 #define TCG_AREG0 TCG_REG_I0

tcg/tcg-op.h

@@ -1039,10 +1039,18 @@ static inline void tcg_gen_mul_i64(TCGv_i64 ret, TCGv_i64 arg1, TCGv_i64 arg2)
     t0 = tcg_temp_new_i64();
     t1 = tcg_temp_new_i32();
 
-    tcg_gen_op4_i32(INDEX_op_mulu2_i32, TCGV_LOW(t0), TCGV_HIGH(t0),
+    if (TCG_TARGET_HAS_mulu2_i32) {
-                    TCGV_LOW(arg1), TCGV_LOW(arg2));
+        tcg_gen_op4_i32(INDEX_op_mulu2_i32, TCGV_LOW(t0), TCGV_HIGH(t0),
-    /* Allow the optimizer room to replace mulu2 with two moves.  */
+                        TCGV_LOW(arg1), TCGV_LOW(arg2));
-    tcg_gen_op0(INDEX_op_nop);
+        /* Allow the optimizer room to replace mulu2 with two moves.  */
+        tcg_gen_op0(INDEX_op_nop);
+    } else {
+        tcg_debug_assert(TCG_TARGET_HAS_muluh_i32);
+        tcg_gen_op3_i32(INDEX_op_mul_i32, TCGV_LOW(t0),
+                        TCGV_LOW(arg1), TCGV_LOW(arg2));
+        tcg_gen_op3_i32(INDEX_op_muluh_i32, TCGV_HIGH(t0),
+                        TCGV_LOW(arg1), TCGV_LOW(arg2));
+    }
 
     tcg_gen_mul_i32(t1, TCGV_LOW(arg1), TCGV_HIGH(arg2));
     tcg_gen_add_i32(TCGV_HIGH(t0), TCGV_HIGH(t0), t1);
@@ -2401,6 +2409,12 @@ static inline void tcg_gen_mulu2_i32(TCGv_i32 rl, TCGv_i32 rh,
         tcg_gen_op4_i32(INDEX_op_mulu2_i32, rl, rh, arg1, arg2);
         /* Allow the optimizer room to replace mulu2 with two moves.  */
         tcg_gen_op0(INDEX_op_nop);
+    } else if (TCG_TARGET_HAS_muluh_i32) {
+        TCGv_i32 t = tcg_temp_new_i32();
+        tcg_gen_op3_i32(INDEX_op_mul_i32, t, arg1, arg2);
+        tcg_gen_op3_i32(INDEX_op_muluh_i32, rh, arg1, arg2);
+        tcg_gen_mov_i32(rl, t);
+        tcg_temp_free_i32(t);
     } else {
         TCGv_i64 t0 = tcg_temp_new_i64();
         TCGv_i64 t1 = tcg_temp_new_i64();
@@ -2420,6 +2434,12 @@ static inline void tcg_gen_muls2_i32(TCGv_i32 rl, TCGv_i32 rh,
         tcg_gen_op4_i32(INDEX_op_muls2_i32, rl, rh, arg1, arg2);
         /* Allow the optimizer room to replace muls2 with two moves.  */
         tcg_gen_op0(INDEX_op_nop);
+    } else if (TCG_TARGET_HAS_mulsh_i32) {
+        TCGv_i32 t = tcg_temp_new_i32();
+        tcg_gen_op3_i32(INDEX_op_mul_i32, t, arg1, arg2);
+        tcg_gen_op3_i32(INDEX_op_mulsh_i32, rh, arg1, arg2);
+        tcg_gen_mov_i32(rl, t);
+        tcg_temp_free_i32(t);
     } else if (TCG_TARGET_REG_BITS == 32 && TCG_TARGET_HAS_mulu2_i32) {
         TCGv_i32 t0 = tcg_temp_new_i32();
         TCGv_i32 t1 = tcg_temp_new_i32();
@@ -2499,6 +2519,12 @@ static inline void tcg_gen_mulu2_i64(TCGv_i64 rl, TCGv_i64 rh,
         tcg_gen_op4_i64(INDEX_op_mulu2_i64, rl, rh, arg1, arg2);
         /* Allow the optimizer room to replace mulu2 with two moves.  */
         tcg_gen_op0(INDEX_op_nop);
+    } else if (TCG_TARGET_HAS_muluh_i64) {
+        TCGv_i64 t = tcg_temp_new_i64();
+        tcg_gen_op3_i64(INDEX_op_mul_i64, t, arg1, arg2);
+        tcg_gen_op3_i64(INDEX_op_muluh_i64, rh, arg1, arg2);
+        tcg_gen_mov_i64(rl, t);
+        tcg_temp_free_i64(t);
     } else if (TCG_TARGET_HAS_mulu2_i64) {
         TCGv_i64 t0 = tcg_temp_new_i64();
         TCGv_i64 t1 = tcg_temp_new_i64();
@@ -2540,6 +2566,12 @@ static inline void tcg_gen_muls2_i64(TCGv_i64 rl, TCGv_i64 rh,
         tcg_gen_op4_i64(INDEX_op_muls2_i64, rl, rh, arg1, arg2);
         /* Allow the optimizer room to replace muls2 with two moves.  */
         tcg_gen_op0(INDEX_op_nop);
+    } else if (TCG_TARGET_HAS_mulsh_i64) {
+        TCGv_i64 t = tcg_temp_new_i64();
+        tcg_gen_op3_i64(INDEX_op_mul_i64, t, arg1, arg2);
+        tcg_gen_op3_i64(INDEX_op_mulsh_i64, rh, arg1, arg2);
+        tcg_gen_mov_i64(rl, t);
+        tcg_temp_free_i64(t);
     } else {
         TCGv_i64 t0 = tcg_temp_new_i64();
         int sizemask = 0;

tcg/tcg-opc.h

@@ -91,6 +91,8 @@ DEF(add2_i32, 2, 4, 0, IMPL(TCG_TARGET_HAS_add2_i32))
 DEF(sub2_i32, 2, 4, 0, IMPL(TCG_TARGET_HAS_sub2_i32))
 DEF(mulu2_i32, 2, 2, 0, IMPL(TCG_TARGET_HAS_mulu2_i32))
 DEF(muls2_i32, 2, 2, 0, IMPL(TCG_TARGET_HAS_muls2_i32))
+DEF(muluh_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_muluh_i32))
+DEF(mulsh_i32, 1, 2, 0, IMPL(TCG_TARGET_HAS_mulsh_i32))
 DEF(brcond2_i32, 0, 4, 2, TCG_OPF_BB_END | IMPL(TCG_TARGET_REG_BITS == 32))
 DEF(setcond2_i32, 1, 4, 1, IMPL(TCG_TARGET_REG_BITS == 32))
 
@@ -167,6 +169,8 @@ DEF(add2_i64, 2, 4, 0, IMPL64 | IMPL(TCG_TARGET_HAS_add2_i64))
 DEF(sub2_i64, 2, 4, 0, IMPL64 | IMPL(TCG_TARGET_HAS_sub2_i64))
 DEF(mulu2_i64, 2, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_mulu2_i64))
 DEF(muls2_i64, 2, 2, 0, IMPL64 | IMPL(TCG_TARGET_HAS_muls2_i64))
+DEF(muluh_i64, 1, 2, 0, IMPL(TCG_TARGET_HAS_muluh_i64))
+DEF(mulsh_i64, 1, 2, 0, IMPL(TCG_TARGET_HAS_mulsh_i64))
 
 /* QEMU specific */
 #if TARGET_LONG_BITS > TCG_TARGET_REG_BITS

tcg/tcg.c

@@ -1252,12 +1252,13 @@ static inline void tcg_la_bb_end(TCGContext *s, uint8_t *dead_temps,
 static void tcg_liveness_analysis(TCGContext *s)
 {
     int i, op_index, nb_args, nb_iargs, nb_oargs, arg, nb_ops;
-    TCGOpcode op, op_new;
+    TCGOpcode op, op_new, op_new2;
     TCGArg *args;
     const TCGOpDef *def;
     uint8_t *dead_temps, *mem_temps;
     uint16_t dead_args;
     uint8_t sync_args;
+    bool have_op_new2;
     
     s->gen_opc_ptr++; /* skip end */
 
@@ -1394,29 +1395,52 @@ static void tcg_liveness_analysis(TCGContext *s)
             goto do_not_remove;
 
         case INDEX_op_mulu2_i32:
+            op_new = INDEX_op_mul_i32;
+            op_new2 = INDEX_op_muluh_i32;
+            have_op_new2 = TCG_TARGET_HAS_muluh_i32;
+            goto do_mul2;
         case INDEX_op_muls2_i32:
             op_new = INDEX_op_mul_i32;
+            op_new2 = INDEX_op_mulsh_i32;
+            have_op_new2 = TCG_TARGET_HAS_mulsh_i32;
             goto do_mul2;
         case INDEX_op_mulu2_i64:
+            op_new = INDEX_op_mul_i64;
+            op_new2 = INDEX_op_muluh_i64;
+            have_op_new2 = TCG_TARGET_HAS_muluh_i64;
+            goto do_mul2;
         case INDEX_op_muls2_i64:
             op_new = INDEX_op_mul_i64;
+            op_new2 = INDEX_op_mulsh_i64;
+            have_op_new2 = TCG_TARGET_HAS_mulsh_i64;
+            goto do_mul2;
         do_mul2:
             args -= 4;
             nb_iargs = 2;
             nb_oargs = 2;
-            /* Likewise, test for the high part of the operation dead.  */
             if (dead_temps[args[1]] && !mem_temps[args[1]]) {
                 if (dead_temps[args[0]] && !mem_temps[args[0]]) {
+                    /* Both parts of the operation are dead.  */
                     goto do_remove;
                 }
+                /* The high part of the operation is dead; generate the low. */
                 s->gen_opc_buf[op_index] = op = op_new;
                 args[1] = args[2];
                 args[2] = args[3];
-                assert(s->gen_opc_buf[op_index + 1] == INDEX_op_nop);
+            } else if (have_op_new2 && dead_temps[args[0]]
-                tcg_set_nop(s, s->gen_opc_buf + op_index + 1, args + 3, 1);
+                       && !mem_temps[args[0]]) {
-                /* Fall through and mark the single-word operation live.  */
+                /* The low part of the operation is dead; generate the high.  */
-                nb_oargs = 1;
+                s->gen_opc_buf[op_index] = op = op_new2;
+                args[0] = args[1];
+                args[1] = args[2];
+                args[2] = args[3];
+            } else {
+                goto do_not_remove;
             }
+            assert(s->gen_opc_buf[op_index + 1] == INDEX_op_nop);
+            tcg_set_nop(s, s->gen_opc_buf + op_index + 1, args + 3, 1);
+            /* Mark the single-word operation live.  */
+            nb_oargs = 1;
             goto do_not_remove;
 
         default:

tcg/tcg.h

@@ -85,6 +85,8 @@ typedef uint64_t TCGRegSet;
 #define TCG_TARGET_HAS_sub2_i64         0
 #define TCG_TARGET_HAS_mulu2_i64        0
 #define TCG_TARGET_HAS_muls2_i64        0
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 /* Turn some undef macros into true macros.  */
 #define TCG_TARGET_HAS_add2_i32         1
 #define TCG_TARGET_HAS_sub2_i32         1

tcg/tci/tcg-target.h

@@ -76,6 +76,8 @@
 #define TCG_TARGET_HAS_rot_i32          1
 #define TCG_TARGET_HAS_movcond_i32      0
 #define TCG_TARGET_HAS_muls2_i32        0
+#define TCG_TARGET_HAS_muluh_i32        0
+#define TCG_TARGET_HAS_mulsh_i32        0
 
 #if TCG_TARGET_REG_BITS == 64
 #define TCG_TARGET_HAS_bswap16_i64      1
@@ -100,13 +102,14 @@
 #define TCG_TARGET_HAS_rot_i64          1
 #define TCG_TARGET_HAS_movcond_i64      0
 #define TCG_TARGET_HAS_muls2_i64        0
-
 #define TCG_TARGET_HAS_add2_i32         0
 #define TCG_TARGET_HAS_sub2_i32         0
 #define TCG_TARGET_HAS_mulu2_i32        0
 #define TCG_TARGET_HAS_add2_i64         0
 #define TCG_TARGET_HAS_sub2_i64         0
 #define TCG_TARGET_HAS_mulu2_i64        0
+#define TCG_TARGET_HAS_muluh_i64        0
+#define TCG_TARGET_HAS_mulsh_i64        0
 #endif /* TCG_TARGET_REG_BITS == 64 */
 
 /* Number of registers available.