[PATCH 27/46] tcg/optimize: Use fold_masks_zs in fold_qemu_ld

[Richard Henderson]

Be careful not to call fold_masks_zs when the memory operation
is wide enough to require multiple outputs, so split into two
functions: fold_qemu_ld_1reg and fold_qemu_ld_2reg.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/optimize.c | 28 ++++++++++++++++++++++------
 1 file changed, 22 insertions(+), 6 deletions(-)

diff --git a/tcg/optimize.c b/tcg/optimize.c
index 76ad02d73b..6f41ef5adb 100644
--- a/tcg/optimize.c
+++ b/tcg/optimize.c
@@ -2092,24 +2092,33 @@ static bool fold_orc(OptContext *ctx, TCGOp *op)
     return fold_masks_zs(ctx, op, -1, s_mask);
 }
 
-static bool fold_qemu_ld(OptContext *ctx, TCGOp *op)
+static bool fold_qemu_ld_1reg(OptContext *ctx, TCGOp *op)
 {
     const TCGOpDef *def = &tcg_op_defs[op->opc];
     MemOpIdx oi = op->args[def->nb_oargs + def->nb_iargs];
     MemOp mop = get_memop(oi);
     int width = 8 * memop_size(mop);
+    uint64_t z_mask = -1, s_mask = 0;
 
     if (width < 64) {
-        ctx->s_mask = MAKE_64BIT_MASK(width, 64 - width);
+        s_mask = MAKE_64BIT_MASK(width, 64 - width);
         if (!(mop & MO_SIGN)) {
-            ctx->z_mask = MAKE_64BIT_MASK(0, width);
-            ctx->s_mask <<= 1;
+            z_mask = MAKE_64BIT_MASK(0, width);
+            s_mask <<= 1;
         }
     }
 
     /* Opcodes that touch guest memory stop the mb optimization.  */
     ctx->prev_mb = NULL;
-    return false;
+
+    return fold_masks_zs(ctx, op, z_mask, s_mask);
+}
+
+static bool fold_qemu_ld_2reg(OptContext *ctx, TCGOp *op)
+{
+    /* Opcodes that touch guest memory stop the mb optimization.  */
+    ctx->prev_mb = NULL;
+    return finish_folding(ctx, op);
 }
 
 static bool fold_qemu_st(OptContext *ctx, TCGOp *op)
@@ -3001,11 +3010,18 @@ void tcg_optimize(TCGContext *s)
             break;
         case INDEX_op_qemu_ld_a32_i32:
         case INDEX_op_qemu_ld_a64_i32:
+            done = fold_qemu_ld_1reg(&ctx, op);
+            break;
         case INDEX_op_qemu_ld_a32_i64:
         case INDEX_op_qemu_ld_a64_i64:
+            if (TCG_TARGET_REG_BITS == 64) {
+                done = fold_qemu_ld_1reg(&ctx, op);
+                break;
+            }
+            QEMU_FALLTHROUGH;
         case INDEX_op_qemu_ld_a32_i128:
         case INDEX_op_qemu_ld_a64_i128:
-            done = fold_qemu_ld(&ctx, op);
+            done = fold_qemu_ld_2reg(&ctx, op);
             break;
         case INDEX_op_qemu_st8_a32_i32:
         case INDEX_op_qemu_st8_a64_i32:
-- 
2.43.0