tcg/tci: Adjust passing of MemOpIdx

Since adding MO_ATOM_MASK, the maximum MemOpIdx requires 15 bits,
which overflows the 12 bit field allocated for TCI memory ops.
Expand the field to 16 bits for 2-operand memory ops, and place
the value in TCG_REG_TMP for 3-operand memory ops (same as we
already do for 4-operand memory ops).

Cures a debug assert for aarch64, with FEAT_LSE2 enabled.

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

2 files changed, 17 insertions, 34 deletions

diff --git a/tcg/tci.c b/tcg/tci.c
index 813572f..4640902 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -106,7 +106,7 @@ static void tci_args_rrm(uint32_t insn, TCGReg *r0,
 {
     *r0 = extract32(insn, 8, 4);
     *r1 = extract32(insn, 12, 4);
-    *m2 = extract32(insn, 20, 12);
+    *m2 = extract32(insn, 16, 16);
 }
 
 static void tci_args_rrr(uint32_t insn, TCGReg *r0, TCGReg *r1, TCGReg *r2)
@@ -141,15 +141,6 @@ static void tci_args_rrrc(uint32_t insn,
     *c3 = extract32(insn, 20, 4);
 }
 
-static void tci_args_rrrm(uint32_t insn,
-                          TCGReg *r0, TCGReg *r1, TCGReg *r2, MemOpIdx *m3)
-{
-    *r0 = extract32(insn, 8, 4);
-    *r1 = extract32(insn, 12, 4);
-    *r2 = extract32(insn, 16, 4);
-    *m3 = extract32(insn, 20, 12);
-}
-
 static void tci_args_rrrbb(uint32_t insn, TCGReg *r0, TCGReg *r1,
                            TCGReg *r2, uint8_t *i3, uint8_t *i4)
 {
@@ -929,8 +920,9 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
                 tci_args_rrm(insn, &r0, &r1, &oi);
                 taddr = regs[r1];
             } else {
-                tci_args_rrrm(insn, &r0, &r1, &r2, &oi);
+                tci_args_rrrr(insn, &r0, &r1, &r2, &r3);
                 taddr = tci_uint64(regs[r2], regs[r1]);
+                oi = regs[r3];
             }
         do_ld_i32:
             regs[r0] = tci_qemu_ld(env, taddr, oi, tb_ptr);
@@ -941,8 +933,9 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
                 tci_args_rrm(insn, &r0, &r1, &oi);
                 taddr = (uint32_t)regs[r1];
             } else {
-                tci_args_rrrm(insn, &r0, &r1, &r2, &oi);
+                tci_args_rrrr(insn, &r0, &r1, &r2, &r3);
                 taddr = (uint32_t)regs[r2];
+                oi = regs[r3];
             }
             goto do_ld_i64;
         case INDEX_op_qemu_ld_a64_i64:
@@ -972,8 +965,9 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
                 tci_args_rrm(insn, &r0, &r1, &oi);
                 taddr = regs[r1];
             } else {
-                tci_args_rrrm(insn, &r0, &r1, &r2, &oi);
+                tci_args_rrrr(insn, &r0, &r1, &r2, &r3);
                 taddr = tci_uint64(regs[r2], regs[r1]);
+                oi = regs[r3];
             }
         do_st_i32:
             tci_qemu_st(env, taddr, regs[r0], oi, tb_ptr);
@@ -985,9 +979,10 @@ uintptr_t QEMU_DISABLE_CFI tcg_qemu_tb_exec(CPUArchState *env,
                 tmp64 = regs[r0];
                 taddr = (uint32_t)regs[r1];
             } else {
-                tci_args_rrrm(insn, &r0, &r1, &r2, &oi);
+                tci_args_rrrr(insn, &r0, &r1, &r2, &r3);
                 tmp64 = tci_uint64(regs[r1], regs[r0]);
                 taddr = (uint32_t)regs[r2];
+                oi = regs[r3];
             }
             goto do_st_i64;
         case INDEX_op_qemu_st_a64_i64:
@@ -1293,9 +1288,10 @@ int print_insn_tci(bfd_vma addr, disassemble_info *info)
                                op_name, str_r(r0), str_r(r1), oi);
             break;
         case 3:
-            tci_args_rrrm(insn, &r0, &r1, &r2, &oi);
-            info->fprintf_func(info->stream, "%-12s  %s, %s, %s, %x",
-                               op_name, str_r(r0), str_r(r1), str_r(r2), oi);
+            tci_args_rrrr(insn, &r0, &r1, &r2, &r3);
+            info->fprintf_func(info->stream, "%-12s  %s, %s, %s, %s",
+                               op_name, str_r(r0), str_r(r1),
+                               str_r(r2), str_r(r3));
             break;
         case 4:
             tci_args_rrrrr(insn, &r0, &r1, &r2, &r3, &r4);
diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc
index c9516a5..5b456e1 100644
--- a/tcg/tci/tcg-target.c.inc
+++ b/tcg/tci/tcg-target.c.inc
@@ -331,11 +331,11 @@ static void tcg_out_op_rrm(TCGContext *s, TCGOpcode op,
 {
     tcg_insn_unit insn = 0;
 
-    tcg_debug_assert(m2 == extract32(m2, 0, 12));
+    tcg_debug_assert(m2 == extract32(m2, 0, 16));
     insn = deposit32(insn, 0, 8, op);
     insn = deposit32(insn, 8, 4, r0);
     insn = deposit32(insn, 12, 4, r1);
-    insn = deposit32(insn, 20, 12, m2);
+    insn = deposit32(insn, 16, 16, m2);
     tcg_out32(s, insn);
 }
 
@@ -392,20 +392,6 @@ static void tcg_out_op_rrrc(TCGContext *s, TCGOpcode op,
     tcg_out32(s, insn);
 }
 
-static void tcg_out_op_rrrm(TCGContext *s, TCGOpcode op,
-                            TCGReg r0, TCGReg r1, TCGReg r2, TCGArg m3)
-{
-    tcg_insn_unit insn = 0;
-
-    tcg_debug_assert(m3 == extract32(m3, 0, 12));
-    insn = deposit32(insn, 0, 8, op);
-    insn = deposit32(insn, 8, 4, r0);
-    insn = deposit32(insn, 12, 4, r1);
-    insn = deposit32(insn, 16, 4, r2);
-    insn = deposit32(insn, 20, 12, m3);
-    tcg_out32(s, insn);
-}
-
 static void tcg_out_op_rrrbb(TCGContext *s, TCGOpcode op, TCGReg r0,
                              TCGReg r1, TCGReg r2, uint8_t b3, uint8_t b4)
 {
@@ -860,7 +846,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
         if (TCG_TARGET_REG_BITS == 64) {
             tcg_out_op_rrm(s, opc, args[0], args[1], args[2]);
         } else {
-            tcg_out_op_rrrm(s, opc, args[0], args[1], args[2], args[3]);
+            tcg_out_movi(s, TCG_TYPE_I32, TCG_REG_TMP, args[4]);
+            tcg_out_op_rrrr(s, opc, args[0], args[1], args[2], TCG_REG_TMP);
         }
         break;
     case INDEX_op_qemu_ld_a64_i64: