Merge remote-tracking branch 'remotes/rth-gitlab/tags/pull-tcg-20210205' into staging

TCGCPUOps cleanups (claudio)
tcg/s390 compare fix (phil)
tcg/aarch64 rotli_vec fix
tcg/tci cleanups and fixes

# gpg: Signature made Fri 05 Feb 2021 22:55:10 GMT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* remotes/rth-gitlab/tags/pull-tcg-20210205: (46 commits)
  accel: introduce AccelCPUClass extending CPUClass
  accel: replace struct CpusAccel with AccelOpsClass
  accel: extend AccelState and AccelClass to user-mode
  cpu: tcg_ops: move to tcg-cpu-ops.h, keep a pointer in CPUClass
  cpu: move debug_check_watchpoint to tcg_ops
  cpu: move adjust_watchpoint_address to tcg_ops
  physmem: make watchpoint checking code TCG-only
  cpu: move do_unaligned_access to tcg_ops
  cpu: move cc->transaction_failed to tcg_ops
  cpu: move cc->do_interrupt to tcg_ops
  target/arm: do not use cc->do_interrupt for KVM directly
  cpu: Move debug_excp_handler to tcg_ops
  cpu: Move tlb_fill to tcg_ops
  cpu: Move cpu_exec_* to tcg_ops
  cpu: Move synchronize_from_tb() to tcg_ops
  accel/tcg: split TCG-only code from cpu_exec_realizefn
  target/riscv: remove CONFIG_TCG, as it is always TCG
  cpu: Introduce TCGCpuOperations struct
  tcg/tci: Remove TCG_CONST
  tcg/tci: Fix TCG_REG_R4 misusage
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
diff --git a/MAINTAINERS b/MAINTAINERS
index 0062694..8d8b0bf 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -129,6 +129,7 @@
 F: include/exec/tb-hash.h
 F: include/sysemu/cpus.h
 F: include/sysemu/tcg.h
+F: include/hw/core/tcg-cpu-ops.h
 
 FPU emulation
 M: Aurelien Jarno <aurelien@aurel32.net>
@@ -437,8 +438,10 @@
 M: Richard Henderson <richard.henderson@linaro.org>
 R: Paolo Bonzini <pbonzini@redhat.com>
 S: Maintained
-F: include/sysemu/accel.h
-F: accel/accel.c
+F: include/qemu/accel.h
+F: include/sysemu/accel-ops.h
+F: include/hw/core/accel-cpu.h
+F: accel/accel-*.c
 F: accel/Makefile.objs
 F: accel/stubs/Makefile.objs
 
diff --git a/accel/accel-common.c b/accel/accel-common.c
new file mode 100644
index 0000000..9901b05
--- /dev/null
+++ b/accel/accel-common.c
@@ -0,0 +1,105 @@
+/*
+ * QEMU accel class, components common to system emulation and user mode
+ *
+ * Copyright (c) 2003-2008 Fabrice Bellard
+ * Copyright (c) 2014 Red Hat Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu/accel.h"
+
+#include "cpu.h"
+#include "hw/core/accel-cpu.h"
+
+#ifndef CONFIG_USER_ONLY
+#include "accel-softmmu.h"
+#endif /* !CONFIG_USER_ONLY */
+
+static const TypeInfo accel_type = {
+    .name = TYPE_ACCEL,
+    .parent = TYPE_OBJECT,
+    .class_size = sizeof(AccelClass),
+    .instance_size = sizeof(AccelState),
+};
+
+/* Lookup AccelClass from opt_name. Returns NULL if not found */
+AccelClass *accel_find(const char *opt_name)
+{
+    char *class_name = g_strdup_printf(ACCEL_CLASS_NAME("%s"), opt_name);
+    AccelClass *ac = ACCEL_CLASS(object_class_by_name(class_name));
+    g_free(class_name);
+    return ac;
+}
+
+static void accel_init_cpu_int_aux(ObjectClass *klass, void *opaque)
+{
+    CPUClass *cc = CPU_CLASS(klass);
+    AccelCPUClass *accel_cpu = opaque;
+
+    cc->accel_cpu = accel_cpu;
+    if (accel_cpu->cpu_class_init) {
+        accel_cpu->cpu_class_init(cc);
+    }
+}
+
+/* initialize the arch-specific accel CpuClass interfaces */
+static void accel_init_cpu_interfaces(AccelClass *ac)
+{
+    const char *ac_name; /* AccelClass name */
+    char *acc_name;      /* AccelCPUClass name */
+    ObjectClass *acc;    /* AccelCPUClass */
+
+    ac_name = object_class_get_name(OBJECT_CLASS(ac));
+    g_assert(ac_name != NULL);
+
+    acc_name = g_strdup_printf("%s-%s", ac_name, CPU_RESOLVING_TYPE);
+    acc = object_class_by_name(acc_name);
+    g_free(acc_name);
+
+    if (acc) {
+        object_class_foreach(accel_init_cpu_int_aux,
+                             CPU_RESOLVING_TYPE, false, acc);
+    }
+}
+
+void accel_init_interfaces(AccelClass *ac)
+{
+#ifndef CONFIG_USER_ONLY
+    accel_init_ops_interfaces(ac);
+#endif /* !CONFIG_USER_ONLY */
+
+    accel_init_cpu_interfaces(ac);
+}
+
+static const TypeInfo accel_cpu_type = {
+    .name = TYPE_ACCEL_CPU,
+    .parent = TYPE_OBJECT,
+    .abstract = true,
+    .class_size = sizeof(AccelCPUClass),
+};
+
+static void register_accel_types(void)
+{
+    type_register_static(&accel_type);
+    type_register_static(&accel_cpu_type);
+}
+
+type_init(register_accel_types);
diff --git a/accel/accel.c b/accel/accel-softmmu.c
similarity index 64%
rename from accel/accel.c
rename to accel/accel-softmmu.c
index cb555e3..50fa5ac 100644
--- a/accel/accel.c
+++ b/accel/accel-softmmu.c
@@ -1,5 +1,5 @@
 /*
- * QEMU System Emulator, accelerator interfaces
+ * QEMU accel class, system emulation components
  *
  * Copyright (c) 2003-2008 Fabrice Bellard
  * Copyright (c) 2014 Red Hat Inc.
@@ -24,27 +24,11 @@
  */
 
 #include "qemu/osdep.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "hw/boards.h"
-#include "sysemu/arch_init.h"
-#include "sysemu/sysemu.h"
-#include "qom/object.h"
+#include "sysemu/cpus.h"
 
-static const TypeInfo accel_type = {
-    .name = TYPE_ACCEL,
-    .parent = TYPE_OBJECT,
-    .class_size = sizeof(AccelClass),
-    .instance_size = sizeof(AccelState),
-};
-
-/* Lookup AccelClass from opt_name. Returns NULL if not found */
-AccelClass *accel_find(const char *opt_name)
-{
-    char *class_name = g_strdup_printf(ACCEL_CLASS_NAME("%s"), opt_name);
-    AccelClass *ac = ACCEL_CLASS(object_class_by_name(class_name));
-    g_free(class_name);
-    return ac;
-}
+#include "accel-softmmu.h"
 
 int accel_init_machine(AccelState *accel, MachineState *ms)
 {
@@ -77,9 +61,40 @@
     }
 }
 
-static void register_accel_types(void)
+/* initialize the arch-independent accel operation interfaces */
+void accel_init_ops_interfaces(AccelClass *ac)
 {
-    type_register_static(&accel_type);
+    const char *ac_name;
+    char *ops_name;
+    AccelOpsClass *ops;
+
+    ac_name = object_class_get_name(OBJECT_CLASS(ac));
+    g_assert(ac_name != NULL);
+
+    ops_name = g_strdup_printf("%s" ACCEL_OPS_SUFFIX, ac_name);
+    ops = ACCEL_OPS_CLASS(object_class_by_name(ops_name));
+    g_free(ops_name);
+
+    /*
+     * all accelerators need to define ops, providing at least a mandatory
+     * non-NULL create_vcpu_thread operation.
+     */
+    g_assert(ops != NULL);
+    if (ops->ops_init) {
+        ops->ops_init(ops);
+    }
+    cpus_register_accel(ops);
 }
 
-type_init(register_accel_types);
+static const TypeInfo accel_ops_type_info = {
+    .name = TYPE_ACCEL_OPS,
+    .parent = TYPE_OBJECT,
+    .abstract = true,
+    .class_size = sizeof(AccelOpsClass),
+};
+
+static void accel_softmmu_register_types(void)
+{
+    type_register_static(&accel_ops_type_info);
+}
+type_init(accel_softmmu_register_types);
diff --git a/accel/accel-softmmu.h b/accel/accel-softmmu.h
new file mode 100644
index 0000000..5e192f1
--- /dev/null
+++ b/accel/accel-softmmu.h
@@ -0,0 +1,15 @@
+/*
+ * QEMU System Emulation accel internal functions
+ *
+ * Copyright 2021 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef ACCEL_SOFTMMU_H
+#define ACCEL_SOFTMMU_H
+
+void accel_init_ops_interfaces(AccelClass *ac);
+
+#endif /* ACCEL_SOFTMMU_H */
diff --git a/accel/accel-user.c b/accel/accel-user.c
new file mode 100644
index 0000000..22b6a1a
--- /dev/null
+++ b/accel/accel-user.c
@@ -0,0 +1,24 @@
+/*
+ * QEMU accel class, user-mode components
+ *
+ * Copyright 2021 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include "qemu/accel.h"
+
+AccelState *current_accel(void)
+{
+    static AccelState *accel;
+
+    if (!accel) {
+        AccelClass *ac = accel_find("tcg");
+
+        g_assert(ac != NULL);
+        accel = ACCEL(object_new_with_class(OBJECT_CLASS(ac)));
+    }
+    return accel;
+}
diff --git a/accel/kvm/kvm-cpus.c b/accel/kvm/kvm-accel-ops.c
similarity index 71%
rename from accel/kvm/kvm-cpus.c
rename to accel/kvm/kvm-accel-ops.c
index d809b1e..7516c67 100644
--- a/accel/kvm/kvm-cpus.c
+++ b/accel/kvm/kvm-accel-ops.c
@@ -74,11 +74,27 @@
                        cpu, QEMU_THREAD_JOINABLE);
 }
 
-const CpusAccel kvm_cpus = {
-    .create_vcpu_thread = kvm_start_vcpu_thread,
+static void kvm_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
-    .synchronize_post_reset = kvm_cpu_synchronize_post_reset,
-    .synchronize_post_init = kvm_cpu_synchronize_post_init,
-    .synchronize_state = kvm_cpu_synchronize_state,
-    .synchronize_pre_loadvm = kvm_cpu_synchronize_pre_loadvm,
+    ops->create_vcpu_thread = kvm_start_vcpu_thread;
+    ops->synchronize_post_reset = kvm_cpu_synchronize_post_reset;
+    ops->synchronize_post_init = kvm_cpu_synchronize_post_init;
+    ops->synchronize_state = kvm_cpu_synchronize_state;
+    ops->synchronize_pre_loadvm = kvm_cpu_synchronize_pre_loadvm;
+}
+
+static const TypeInfo kvm_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("kvm"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = kvm_accel_ops_class_init,
+    .abstract = true,
 };
+
+static void kvm_accel_ops_register_types(void)
+{
+    type_register_static(&kvm_accel_ops_type);
+}
+type_init(kvm_accel_ops_register_types);
diff --git a/accel/kvm/kvm-all.c b/accel/kvm/kvm-all.c
index 3feb17d..5164d83 100644
--- a/accel/kvm/kvm-all.c
+++ b/accel/kvm/kvm-all.c
@@ -2256,8 +2256,6 @@
         ret = ram_block_discard_disable(true);
         assert(!ret);
     }
-
-    cpus_register_accel(&kvm_cpus);
     return 0;
 
 err:
diff --git a/accel/kvm/kvm-cpus.h b/accel/kvm/kvm-cpus.h
index 3df732b..bf0bd1b 100644
--- a/accel/kvm/kvm-cpus.h
+++ b/accel/kvm/kvm-cpus.h
@@ -12,8 +12,6 @@
 
 #include "sysemu/cpus.h"
 
-extern const CpusAccel kvm_cpus;
-
 int kvm_init_vcpu(CPUState *cpu, Error **errp);
 int kvm_cpu_exec(CPUState *cpu);
 void kvm_destroy_vcpu(CPUState *cpu);
diff --git a/accel/kvm/meson.build b/accel/kvm/meson.build
index 7e9dafe..8d219be 100644
--- a/accel/kvm/meson.build
+++ b/accel/kvm/meson.build
@@ -1,7 +1,7 @@
 kvm_ss = ss.source_set()
 kvm_ss.add(files(
   'kvm-all.c',
-  'kvm-cpus.c',
+  'kvm-accel-ops.c',
 ))
 kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c'))
 
diff --git a/accel/meson.build b/accel/meson.build
index b26cca2..b44ba30 100644
--- a/accel/meson.build
+++ b/accel/meson.build
@@ -1,4 +1,6 @@
-softmmu_ss.add(files('accel.c'))
+specific_ss.add(files('accel-common.c'))
+softmmu_ss.add(files('accel-softmmu.c'))
+user_ss.add(files('accel-user.c'))
 
 subdir('qtest')
 subdir('kvm')
diff --git a/accel/qtest/qtest.c b/accel/qtest/qtest.c
index b282cea..edb29f6 100644
--- a/accel/qtest/qtest.c
+++ b/accel/qtest/qtest.c
@@ -17,7 +17,7 @@
 #include "qemu/module.h"
 #include "qemu/option.h"
 #include "qemu/config-file.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/qtest.h"
 #include "sysemu/cpus.h"
 #include "sysemu/cpu-timers.h"
@@ -25,14 +25,8 @@
 #include "qemu/main-loop.h"
 #include "hw/core/cpu.h"
 
-const CpusAccel qtest_cpus = {
-    .create_vcpu_thread = dummy_start_vcpu_thread,
-    .get_virtual_clock = qtest_get_virtual_clock,
-};
-
 static int qtest_init_accel(MachineState *ms)
 {
-    cpus_register_accel(&qtest_cpus);
     return 0;
 }
 
@@ -52,9 +46,26 @@
     .class_init = qtest_accel_class_init,
 };
 
+static void qtest_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
+
+    ops->create_vcpu_thread = dummy_start_vcpu_thread;
+    ops->get_virtual_clock = qtest_get_virtual_clock;
+};
+
+static const TypeInfo qtest_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("qtest"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = qtest_accel_ops_class_init,
+    .abstract = true,
+};
+
 static void qtest_type_init(void)
 {
     type_register_static(&qtest_accel_type);
+    type_register_static(&qtest_accel_ops_type);
 }
 
 type_init(qtest_type_init);
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 6d017e4..d9ef691 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -21,6 +21,7 @@
 #include "qemu-common.h"
 #include "qemu/qemu-print.h"
 #include "cpu.h"
+#include "hw/core/tcg-cpu-ops.h"
 #include "trace.h"
 #include "disas/disas.h"
 #include "exec/exec-all.h"
@@ -213,8 +214,8 @@
                                TARGET_FMT_lx "] %s\n",
                                last_tb->tc.ptr, last_tb->pc,
                                lookup_symbol(last_tb->pc));
-        if (cc->synchronize_from_tb) {
-            cc->synchronize_from_tb(cpu, last_tb);
+        if (cc->tcg_ops->synchronize_from_tb) {
+            cc->tcg_ops->synchronize_from_tb(cpu, last_tb);
         } else {
             assert(cc->set_pc);
             cc->set_pc(cpu, last_tb->pc);
@@ -262,8 +263,8 @@
 {
     CPUClass *cc = CPU_GET_CLASS(cpu);
 
-    if (cc->cpu_exec_enter) {
-        cc->cpu_exec_enter(cpu);
+    if (cc->tcg_ops->cpu_exec_enter) {
+        cc->tcg_ops->cpu_exec_enter(cpu);
     }
 }
 
@@ -271,8 +272,8 @@
 {
     CPUClass *cc = CPU_GET_CLASS(cpu);
 
-    if (cc->cpu_exec_exit) {
-        cc->cpu_exec_exit(cpu);
+    if (cc->tcg_ops->cpu_exec_exit) {
+        cc->tcg_ops->cpu_exec_exit(cpu);
     }
 }
 
@@ -512,8 +513,8 @@
         }
     }
 
-    if (cc->debug_excp_handler) {
-        cc->debug_excp_handler(cpu);
+    if (cc->tcg_ops->debug_excp_handler) {
+        cc->tcg_ops->debug_excp_handler(cpu);
     }
 }
 
@@ -547,7 +548,7 @@
            loop */
 #if defined(TARGET_I386)
         CPUClass *cc = CPU_GET_CLASS(cpu);
-        cc->do_interrupt(cpu);
+        cc->tcg_ops->do_interrupt(cpu);
 #endif
         *ret = cpu->exception_index;
         cpu->exception_index = -1;
@@ -556,7 +557,7 @@
         if (replay_exception()) {
             CPUClass *cc = CPU_GET_CLASS(cpu);
             qemu_mutex_lock_iothread();
-            cc->do_interrupt(cpu);
+            cc->tcg_ops->do_interrupt(cpu);
             qemu_mutex_unlock_iothread();
             cpu->exception_index = -1;
 
@@ -655,8 +656,8 @@
            True when it is, and we should restart on a new TB,
            and via longjmp via cpu_loop_exit.  */
         else {
-            if (cc->cpu_exec_interrupt &&
-                cc->cpu_exec_interrupt(cpu, interrupt_request)) {
+            if (cc->tcg_ops->cpu_exec_interrupt &&
+                cc->tcg_ops->cpu_exec_interrupt(cpu, interrupt_request)) {
                 if (need_replay_interrupt(interrupt_request)) {
                     replay_interrupt();
                 }
@@ -828,6 +829,34 @@
     return ret;
 }
 
+void tcg_exec_realizefn(CPUState *cpu, Error **errp)
+{
+    static bool tcg_target_initialized;
+    CPUClass *cc = CPU_GET_CLASS(cpu);
+
+    if (!tcg_target_initialized) {
+        cc->tcg_ops->initialize();
+        tcg_target_initialized = true;
+    }
+    tlb_init(cpu);
+    qemu_plugin_vcpu_init_hook(cpu);
+
+#ifndef CONFIG_USER_ONLY
+    tcg_iommu_init_notifier_list(cpu);
+#endif /* !CONFIG_USER_ONLY */
+}
+
+/* undo the initializations in reverse order */
+void tcg_exec_unrealizefn(CPUState *cpu)
+{
+#ifndef CONFIG_USER_ONLY
+    tcg_iommu_free_notifier_list(cpu);
+#endif /* !CONFIG_USER_ONLY */
+
+    qemu_plugin_vcpu_exit_hook(cpu);
+    tlb_destroy(cpu);
+}
+
 #ifndef CONFIG_USER_ONLY
 
 void dump_drift_info(void)
diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index 7a69726..8a7b779 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -20,6 +20,7 @@
 #include "qemu/osdep.h"
 #include "qemu/main-loop.h"
 #include "cpu.h"
+#include "hw/core/tcg-cpu-ops.h"
 #include "exec/exec-all.h"
 #include "exec/memory.h"
 #include "exec/address-spaces.h"
@@ -1305,10 +1306,37 @@
      * This is not a probe, so only valid return is success; failure
      * should result in exception + longjmp to the cpu loop.
      */
-    ok = cc->tlb_fill(cpu, addr, size, access_type, mmu_idx, false, retaddr);
+    ok = cc->tcg_ops->tlb_fill(cpu, addr, size,
+                               access_type, mmu_idx, false, retaddr);
     assert(ok);
 }
 
+static inline void cpu_unaligned_access(CPUState *cpu, vaddr addr,
+                                        MMUAccessType access_type,
+                                        int mmu_idx, uintptr_t retaddr)
+{
+    CPUClass *cc = CPU_GET_CLASS(cpu);
+
+    cc->tcg_ops->do_unaligned_access(cpu, addr, access_type, mmu_idx, retaddr);
+}
+
+static inline void cpu_transaction_failed(CPUState *cpu, hwaddr physaddr,
+                                          vaddr addr, unsigned size,
+                                          MMUAccessType access_type,
+                                          int mmu_idx, MemTxAttrs attrs,
+                                          MemTxResult response,
+                                          uintptr_t retaddr)
+{
+    CPUClass *cc = CPU_GET_CLASS(cpu);
+
+    if (!cpu->ignore_memory_transaction_failures &&
+        cc->tcg_ops->do_transaction_failed) {
+        cc->tcg_ops->do_transaction_failed(cpu, physaddr, addr, size,
+                                           access_type, mmu_idx, attrs,
+                                           response, retaddr);
+    }
+}
+
 static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
                          int mmu_idx, target_ulong addr, uintptr_t retaddr,
                          MMUAccessType access_type, MemOp op)
@@ -1576,8 +1604,8 @@
             CPUState *cs = env_cpu(env);
             CPUClass *cc = CPU_GET_CLASS(cs);
 
-            if (!cc->tlb_fill(cs, addr, fault_size, access_type,
-                              mmu_idx, nonfault, retaddr)) {
+            if (!cc->tcg_ops->tlb_fill(cs, addr, fault_size, access_type,
+                                       mmu_idx, nonfault, retaddr)) {
                 /* Non-faulting page table read failed.  */
                 *phost = NULL;
                 return TLB_INVALID_MASK;
diff --git a/accel/tcg/meson.build b/accel/tcg/meson.build
index f39aab0..1236ac7 100644
--- a/accel/tcg/meson.build
+++ b/accel/tcg/meson.build
@@ -1,5 +1,6 @@
 tcg_ss = ss.source_set()
 tcg_ss.add(files(
+  'tcg-all.c',
   'cpu-exec-common.c',
   'cpu-exec.c',
   'tcg-runtime-gvec.c',
@@ -13,10 +14,9 @@
 specific_ss.add_all(when: 'CONFIG_TCG', if_true: tcg_ss)
 
 specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: files(
-  'tcg-all.c',
   'cputlb.c',
-  'tcg-cpus.c',
-  'tcg-cpus-mttcg.c',
-  'tcg-cpus-icount.c',
-  'tcg-cpus-rr.c'
+  'tcg-accel-ops.c',
+  'tcg-accel-ops-mttcg.c',
+  'tcg-accel-ops-icount.c',
+  'tcg-accel-ops-rr.c'
 ))
diff --git a/accel/tcg/tcg-cpus-icount.c b/accel/tcg/tcg-accel-ops-icount.c
similarity index 89%
rename from accel/tcg/tcg-cpus-icount.c
rename to accel/tcg/tcg-accel-ops-icount.c
index 9f45432..87762b4 100644
--- a/accel/tcg/tcg-cpus-icount.c
+++ b/accel/tcg/tcg-accel-ops-icount.c
@@ -32,9 +32,9 @@
 #include "exec/exec-all.h"
 #include "hw/boards.h"
 
-#include "tcg-cpus.h"
-#include "tcg-cpus-icount.h"
-#include "tcg-cpus-rr.h"
+#include "tcg-accel-ops.h"
+#include "tcg-accel-ops-icount.h"
+#include "tcg-accel-ops-rr.h"
 
 static int64_t icount_get_limit(void)
 {
@@ -93,7 +93,7 @@
     /*
      * These should always be cleared by icount_process_data after
      * each vCPU execution. However u16.high can be raised
-     * asynchronously by cpu_exit/cpu_interrupt/tcg_cpus_handle_interrupt
+     * asynchronously by cpu_exit/cpu_interrupt/tcg_handle_interrupt
      */
     g_assert(cpu_neg(cpu)->icount_decr.u16.low == 0);
     g_assert(cpu->icount_extra == 0);
@@ -125,23 +125,14 @@
     replay_mutex_unlock();
 }
 
-static void icount_handle_interrupt(CPUState *cpu, int mask)
+void icount_handle_interrupt(CPUState *cpu, int mask)
 {
     int old_mask = cpu->interrupt_request;
 
-    tcg_cpus_handle_interrupt(cpu, mask);
+    tcg_handle_interrupt(cpu, mask);
     if (qemu_cpu_is_self(cpu) &&
         !cpu->can_do_io
         && (mask & ~old_mask) != 0) {
         cpu_abort(cpu, "Raised interrupt while not in I/O function");
     }
 }
-
-const CpusAccel tcg_cpus_icount = {
-    .create_vcpu_thread = rr_start_vcpu_thread,
-    .kick_vcpu_thread = rr_kick_vcpu_thread,
-
-    .handle_interrupt = icount_handle_interrupt,
-    .get_virtual_clock = icount_get,
-    .get_elapsed_ticks = icount_get,
-};
diff --git a/accel/tcg/tcg-cpus-icount.h b/accel/tcg/tcg-accel-ops-icount.h
similarity index 88%
rename from accel/tcg/tcg-cpus-icount.h
rename to accel/tcg/tcg-accel-ops-icount.h
index b695939..d884aa2 100644
--- a/accel/tcg/tcg-cpus-icount.h
+++ b/accel/tcg/tcg-accel-ops-icount.h
@@ -14,4 +14,6 @@
 void icount_prepare_for_run(CPUState *cpu);
 void icount_process_data(CPUState *cpu);
 
+void icount_handle_interrupt(CPUState *cpu, int mask);
+
 #endif /* TCG_CPUS_ICOUNT_H */
diff --git a/accel/tcg/tcg-cpus-mttcg.c b/accel/tcg/tcg-accel-ops-mttcg.c
similarity index 92%
rename from accel/tcg/tcg-cpus-mttcg.c
rename to accel/tcg/tcg-accel-ops-mttcg.c
index 9c3767d..42973fb 100644
--- a/accel/tcg/tcg-cpus-mttcg.c
+++ b/accel/tcg/tcg-accel-ops-mttcg.c
@@ -32,7 +32,8 @@
 #include "exec/exec-all.h"
 #include "hw/boards.h"
 
-#include "tcg-cpus.h"
+#include "tcg-accel-ops.h"
+#include "tcg-accel-ops-mttcg.h"
 
 /*
  * In the multi-threaded case each vCPU has its own thread. The TLS
@@ -103,12 +104,12 @@
     return NULL;
 }
 
-static void mttcg_kick_vcpu_thread(CPUState *cpu)
+void mttcg_kick_vcpu_thread(CPUState *cpu)
 {
     cpu_exit(cpu);
 }
 
-static void mttcg_start_vcpu_thread(CPUState *cpu)
+void mttcg_start_vcpu_thread(CPUState *cpu)
 {
     char thread_name[VCPU_THREAD_NAME_SIZE];
 
@@ -131,10 +132,3 @@
     cpu->hThread = qemu_thread_get_handle(cpu->thread);
 #endif
 }
-
-const CpusAccel tcg_cpus_mttcg = {
-    .create_vcpu_thread = mttcg_start_vcpu_thread,
-    .kick_vcpu_thread = mttcg_kick_vcpu_thread,
-
-    .handle_interrupt = tcg_cpus_handle_interrupt,
-};
diff --git a/accel/tcg/tcg-accel-ops-mttcg.h b/accel/tcg/tcg-accel-ops-mttcg.h
new file mode 100644
index 0000000..9fdc5a2
--- /dev/null
+++ b/accel/tcg/tcg-accel-ops-mttcg.h
@@ -0,0 +1,19 @@
+/*
+ * QEMU TCG Multi Threaded vCPUs implementation
+ *
+ * Copyright 2021 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef TCG_CPUS_MTTCG_H
+#define TCG_CPUS_MTTCG_H
+
+/* kick MTTCG vCPU thread */
+void mttcg_kick_vcpu_thread(CPUState *cpu);
+
+/* start an mttcg vCPU thread */
+void mttcg_start_vcpu_thread(CPUState *cpu);
+
+#endif /* TCG_CPUS_MTTCG_H */
diff --git a/accel/tcg/tcg-cpus-rr.c b/accel/tcg/tcg-accel-ops-rr.c
similarity index 96%
rename from accel/tcg/tcg-cpus-rr.c
rename to accel/tcg/tcg-accel-ops-rr.c
index 0181d2e..4a66055 100644
--- a/accel/tcg/tcg-cpus-rr.c
+++ b/accel/tcg/tcg-accel-ops-rr.c
@@ -32,9 +32,9 @@
 #include "exec/exec-all.h"
 #include "hw/boards.h"
 
-#include "tcg-cpus.h"
-#include "tcg-cpus-rr.h"
-#include "tcg-cpus-icount.h"
+#include "tcg-accel-ops.h"
+#include "tcg-accel-ops-rr.h"
+#include "tcg-accel-ops-icount.h"
 
 /* Kick all RR vCPUs */
 void rr_kick_vcpu_thread(CPUState *unused)
@@ -296,10 +296,3 @@
         cpu->created = true;
     }
 }
-
-const CpusAccel tcg_cpus_rr = {
-    .create_vcpu_thread = rr_start_vcpu_thread,
-    .kick_vcpu_thread = rr_kick_vcpu_thread,
-
-    .handle_interrupt = tcg_cpus_handle_interrupt,
-};
diff --git a/accel/tcg/tcg-cpus-rr.h b/accel/tcg/tcg-accel-ops-rr.h
similarity index 100%
rename from accel/tcg/tcg-cpus-rr.h
rename to accel/tcg/tcg-accel-ops-rr.h
diff --git a/accel/tcg/tcg-cpus.c b/accel/tcg/tcg-accel-ops.c
similarity index 63%
rename from accel/tcg/tcg-cpus.c
rename to accel/tcg/tcg-accel-ops.c
index e335f9f..6144d9d 100644
--- a/accel/tcg/tcg-cpus.c
+++ b/accel/tcg/tcg-accel-ops.c
@@ -34,7 +34,10 @@
 #include "exec/exec-all.h"
 #include "hw/boards.h"
 
-#include "tcg-cpus.h"
+#include "tcg-accel-ops.h"
+#include "tcg-accel-ops-mttcg.h"
+#include "tcg-accel-ops-rr.h"
+#include "tcg-accel-ops-icount.h"
 
 /* common functionality among all TCG variants */
 
@@ -64,7 +67,7 @@
 }
 
 /* mask must never be zero, except for A20 change call */
-void tcg_cpus_handle_interrupt(CPUState *cpu, int mask)
+void tcg_handle_interrupt(CPUState *cpu, int mask)
 {
     g_assert(qemu_mutex_iothread_locked());
 
@@ -80,3 +83,43 @@
         qatomic_set(&cpu_neg(cpu)->icount_decr.u16.high, -1);
     }
 }
+
+static void tcg_accel_ops_init(AccelOpsClass *ops)
+{
+    if (qemu_tcg_mttcg_enabled()) {
+        ops->create_vcpu_thread = mttcg_start_vcpu_thread;
+        ops->kick_vcpu_thread = mttcg_kick_vcpu_thread;
+        ops->handle_interrupt = tcg_handle_interrupt;
+    } else if (icount_enabled()) {
+        ops->create_vcpu_thread = rr_start_vcpu_thread;
+        ops->kick_vcpu_thread = rr_kick_vcpu_thread;
+        ops->handle_interrupt = icount_handle_interrupt;
+        ops->get_virtual_clock = icount_get;
+        ops->get_elapsed_ticks = icount_get;
+    } else {
+        ops->create_vcpu_thread = rr_start_vcpu_thread;
+        ops->kick_vcpu_thread = rr_kick_vcpu_thread;
+        ops->handle_interrupt = tcg_handle_interrupt;
+    }
+}
+
+static void tcg_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
+
+    ops->ops_init = tcg_accel_ops_init;
+}
+
+static const TypeInfo tcg_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("tcg"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = tcg_accel_ops_class_init,
+    .abstract = true,
+};
+
+static void tcg_accel_ops_register_types(void)
+{
+    type_register_static(&tcg_accel_ops_type);
+}
+type_init(tcg_accel_ops_register_types);
diff --git a/accel/tcg/tcg-cpus.h b/accel/tcg/tcg-accel-ops.h
similarity index 71%
rename from accel/tcg/tcg-cpus.h
rename to accel/tcg/tcg-accel-ops.h
index d6893a3..4813000 100644
--- a/accel/tcg/tcg-cpus.h
+++ b/accel/tcg/tcg-accel-ops.h
@@ -14,12 +14,8 @@
 
 #include "sysemu/cpus.h"
 
-extern const CpusAccel tcg_cpus_mttcg;
-extern const CpusAccel tcg_cpus_icount;
-extern const CpusAccel tcg_cpus_rr;
-
 void tcg_cpus_destroy(CPUState *cpu);
 int tcg_cpus_exec(CPUState *cpu);
-void tcg_cpus_handle_interrupt(CPUState *cpu, int mask);
+void tcg_handle_interrupt(CPUState *cpu, int mask);
 
 #endif /* TCG_CPUS_H */
diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c
index 2eea8c3..e378c2d 100644
--- a/accel/tcg/tcg-all.c
+++ b/accel/tcg/tcg-all.c
@@ -30,9 +30,8 @@
 #include "tcg/tcg.h"
 #include "qapi/error.h"
 #include "qemu/error-report.h"
-#include "hw/boards.h"
+#include "qemu/accel.h"
 #include "qapi/qapi-builtin-visit.h"
-#include "tcg-cpus.h"
 
 struct TCGState {
     AccelState parent_obj;
@@ -97,7 +96,7 @@
     s->mttcg_enabled = default_mttcg_enabled();
 
     /* If debugging enabled, default "auto on", otherwise off. */
-#ifdef CONFIG_DEBUG_TCG
+#if defined(CONFIG_DEBUG_TCG) && !defined(CONFIG_USER_ONLY)
     s->splitwx_enabled = -1;
 #else
     s->splitwx_enabled = 0;
@@ -114,17 +113,15 @@
     mttcg_enabled = s->mttcg_enabled;
 
     /*
-     * Initialize TCG regions
+     * Initialize TCG regions only for softmmu.
+     *
+     * This needs to be done later for user mode, because the prologue
+     * generation needs to be delayed so that GUEST_BASE is already set.
      */
+#ifndef CONFIG_USER_ONLY
     tcg_region_init();
+#endif /* !CONFIG_USER_ONLY */
 
-    if (mttcg_enabled) {
-        cpus_register_accel(&tcg_cpus_mttcg);
-    } else if (icount_enabled()) {
-        cpus_register_accel(&tcg_cpus_icount);
-    } else {
-        cpus_register_accel(&tcg_cpus_rr);
-    }
     return 0;
 }
 
diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c
index 1215b55..0b6f56c 100644
--- a/accel/tcg/user-exec.c
+++ b/accel/tcg/user-exec.c
@@ -18,6 +18,7 @@
  */
 #include "qemu/osdep.h"
 #include "cpu.h"
+#include "hw/core/tcg-cpu-ops.h"
 #include "disas/disas.h"
 #include "exec/exec-all.h"
 #include "tcg/tcg.h"
@@ -187,7 +188,8 @@
     clear_helper_retaddr();
 
     cc = CPU_GET_CLASS(cpu);
-    cc->tlb_fill(cpu, address, 0, access_type, MMU_USER_IDX, false, pc);
+    cc->tcg_ops->tlb_fill(cpu, address, 0, access_type,
+                          MMU_USER_IDX, false, pc);
     g_assert_not_reached();
 }
 
@@ -217,8 +219,8 @@
         } else {
             CPUState *cpu = env_cpu(env);
             CPUClass *cc = CPU_GET_CLASS(cpu);
-            cc->tlb_fill(cpu, addr, fault_size, access_type,
-                         MMU_USER_IDX, false, ra);
+            cc->tcg_ops->tlb_fill(cpu, addr, fault_size, access_type,
+                                  MMU_USER_IDX, false, ra);
             g_assert_not_reached();
         }
     }
diff --git a/accel/xen/xen-all.c b/accel/xen/xen-all.c
index 878a408..e9d2d6a 100644
--- a/accel/xen/xen-all.c
+++ b/accel/xen/xen-all.c
@@ -15,7 +15,7 @@
 #include "hw/xen/xen-legacy-backend.h"
 #include "hw/xen/xen_pt.h"
 #include "chardev/char.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/cpus.h"
 #include "sysemu/xen.h"
 #include "sysemu/runstate.h"
@@ -154,10 +154,6 @@
     }
 }
 
-const CpusAccel xen_cpus = {
-    .create_vcpu_thread = dummy_start_vcpu_thread,
-};
-
 static int xen_init(MachineState *ms)
 {
     MachineClass *mc = MACHINE_GET_CLASS(ms);
@@ -185,9 +181,6 @@
      * opt out of system RAM being allocated by generic code
      */
     mc->default_ram_id = NULL;
-
-    cpus_register_accel(&xen_cpus);
-
     return 0;
 }
 
@@ -222,9 +215,24 @@
     .class_init = xen_accel_class_init,
 };
 
+static void xen_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
+
+    ops->create_vcpu_thread = dummy_start_vcpu_thread;
+}
+
+static const TypeInfo xen_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("xen"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = xen_accel_ops_class_init,
+    .abstract = true,
+};
+
 static void xen_type_init(void)
 {
     type_register_static(&xen_accel_type);
+    type_register_static(&xen_accel_ops_type);
 }
-
 type_init(xen_type_init);
diff --git a/bsd-user/main.c b/bsd-user/main.c
index 65163e1..7cc0802 100644
--- a/bsd-user/main.c
+++ b/bsd-user/main.c
@@ -20,6 +20,7 @@
 #include "qemu/osdep.h"
 #include "qemu-common.h"
 #include "qemu/units.h"
+#include "qemu/accel.h"
 #include "sysemu/tcg.h"
 #include "qemu-version.h"
 #include <machine/trap.h>
@@ -908,10 +909,14 @@
 #endif
     }
 
-    /* init tcg before creating CPUs and to get qemu_host_page_size */
-    tcg_exec_init(0, false);
-
     cpu_type = parse_cpu_option(cpu_model);
+    /* init tcg before creating CPUs and to get qemu_host_page_size */
+    {
+        AccelClass *ac = ACCEL_GET_CLASS(current_accel());
+
+        ac->init_machine(NULL);
+        accel_init_interfaces(ac);
+    }
     cpu = cpu_create(cpu_type);
     env = cpu->env_ptr;
 #if defined(TARGET_SPARC) || defined(TARGET_PPC)
diff --git a/configure b/configure
index e85d6ba..a34f911 100755
--- a/configure
+++ b/configure
@@ -1110,9 +1110,9 @@
   ;;
   --enable-whpx) whpx="enabled"
   ;;
-  --disable-tcg-interpreter) tcg_interpreter="true"
+  --disable-tcg-interpreter) tcg_interpreter="false"
   ;;
-  --enable-tcg-interpreter) tcg_interpreter="false"
+  --enable-tcg-interpreter) tcg_interpreter="true"
   ;;
   --disable-cap-ng)  cap_ng="disabled"
   ;;
@@ -6417,6 +6417,7 @@
         -Dvhost_user_blk_server=$vhost_user_blk_server \
         -Dfuse=$fuse -Dfuse_lseek=$fuse_lseek -Dguest_agent_msi=$guest_agent_msi \
         $(if test "$default_features" = no; then echo "-Dauto_features=disabled"; fi) \
+	-Dtcg_interpreter=$tcg_interpreter \
         $cross_arg \
         "$PWD" "$source_path"
 
diff --git a/cpu.c b/cpu.c
index 0b245cd..bfbe5a6 100644
--- a/cpu.c
+++ b/cpu.c
@@ -124,12 +124,34 @@
 };
 #endif
 
-void cpu_exec_unrealizefn(CPUState *cpu)
+void cpu_exec_realizefn(CPUState *cpu, Error **errp)
 {
     CPUClass *cc = CPU_GET_CLASS(cpu);
 
-    tlb_destroy(cpu);
-    cpu_list_remove(cpu);
+    cpu_list_add(cpu);
+
+#ifdef CONFIG_TCG
+    /* NB: errp parameter is unused currently */
+    if (tcg_enabled()) {
+        tcg_exec_realizefn(cpu, errp);
+    }
+#endif /* CONFIG_TCG */
+
+#ifdef CONFIG_USER_ONLY
+    assert(cc->vmsd == NULL);
+#else
+    if (qdev_get_vmsd(DEVICE(cpu)) == NULL) {
+        vmstate_register(NULL, cpu->cpu_index, &vmstate_cpu_common, cpu);
+    }
+    if (cc->vmsd != NULL) {
+        vmstate_register(NULL, cpu->cpu_index, cc->vmsd, cpu);
+    }
+#endif /* CONFIG_USER_ONLY */
+}
+
+void cpu_exec_unrealizefn(CPUState *cpu)
+{
+    CPUClass *cc = CPU_GET_CLASS(cpu);
 
 #ifdef CONFIG_USER_ONLY
     assert(cc->vmsd == NULL);
@@ -140,8 +162,15 @@
     if (qdev_get_vmsd(DEVICE(cpu)) == NULL) {
         vmstate_unregister(NULL, &vmstate_cpu_common, cpu);
     }
-    tcg_iommu_free_notifier_list(cpu);
 #endif
+#ifdef CONFIG_TCG
+    /* NB: errp parameter is unused currently */
+    if (tcg_enabled()) {
+        tcg_exec_unrealizefn(cpu);
+    }
+#endif /* CONFIG_TCG */
+
+    cpu_list_remove(cpu);
 }
 
 void cpu_exec_initfn(CPUState *cpu)
@@ -156,35 +185,6 @@
 #endif
 }
 
-void cpu_exec_realizefn(CPUState *cpu, Error **errp)
-{
-    CPUClass *cc = CPU_GET_CLASS(cpu);
-    static bool tcg_target_initialized;
-
-    cpu_list_add(cpu);
-
-    if (tcg_enabled() && !tcg_target_initialized) {
-        tcg_target_initialized = true;
-        cc->tcg_initialize();
-    }
-    tlb_init(cpu);
-
-    qemu_plugin_vcpu_init_hook(cpu);
-
-#ifdef CONFIG_USER_ONLY
-    assert(cc->vmsd == NULL);
-#else /* !CONFIG_USER_ONLY */
-    if (qdev_get_vmsd(DEVICE(cpu)) == NULL) {
-        vmstate_register(NULL, cpu->cpu_index, &vmstate_cpu_common, cpu);
-    }
-    if (cc->vmsd != NULL) {
-        vmstate_register(NULL, cpu->cpu_index, cc->vmsd, cpu);
-    }
-
-    tcg_iommu_init_notifier_list(cpu);
-#endif
-}
-
 const char *parse_cpu_option(const char *cpu_option)
 {
     ObjectClass *oc;
diff --git a/hw/core/cpu.c b/hw/core/cpu.c
index 7553411..00330ba 100644
--- a/hw/core/cpu.c
+++ b/hw/core/cpu.c
@@ -186,19 +186,15 @@
     return 0;
 }
 
-static bool cpu_common_debug_check_watchpoint(CPUState *cpu, CPUWatchpoint *wp)
-{
-    /* If no extra check is required, QEMU watchpoint match can be considered
-     * as an architectural match.
-     */
-    return true;
-}
-
 static bool cpu_common_virtio_is_big_endian(CPUState *cpu)
 {
     return target_words_bigendian();
 }
 
+/*
+ * XXX the following #if is always true because this is a common_ss
+ * module, so target CONFIG_* is never defined.
+ */
 #if !defined(CONFIG_USER_ONLY)
 GuestPanicInformation *cpu_get_crash_info(CPUState *cpu)
 {
@@ -340,9 +336,9 @@
 static void cpu_common_unrealizefn(DeviceState *dev)
 {
     CPUState *cpu = CPU(dev);
+
     /* NOTE: latest generic point before the cpu is fully unrealized */
     trace_fini_vcpu(cpu);
-    qemu_plugin_vcpu_exit_hook(cpu);
     cpu_exec_unrealizefn(cpu);
 }
 
@@ -379,11 +375,6 @@
     return cpu->cpu_index;
 }
 
-static vaddr cpu_adjust_watchpoint_address(CPUState *cpu, vaddr addr, int len)
-{
-    return addr;
-}
-
 static Property cpu_common_props[] = {
 #ifndef CONFIG_USER_ONLY
     /* Create a memory property for softmmu CPU object,
@@ -416,8 +407,6 @@
     k->gdb_read_register = cpu_common_gdb_read_register;
     k->gdb_write_register = cpu_common_gdb_write_register;
     k->virtio_is_big_endian = cpu_common_virtio_is_big_endian;
-    k->debug_check_watchpoint = cpu_common_debug_check_watchpoint;
-    k->adjust_watchpoint_address = cpu_adjust_watchpoint_address;
     set_bit(DEVICE_CATEGORY_CPU, dc->categories);
     dc->realize = cpu_common_realizefn;
     dc->unrealize = cpu_common_unrealizefn;
diff --git a/hw/mips/jazz.c b/hw/mips/jazz.c
index f944273..83c8086 100644
--- a/hw/mips/jazz.c
+++ b/hw/mips/jazz.c
@@ -53,6 +53,9 @@
 #include "qapi/error.h"
 #include "qemu/error-report.h"
 #include "qemu/help_option.h"
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+#endif /* CONFIG_TCG */
 
 enum jazz_model_e {
     JAZZ_MAGNUM,
@@ -116,6 +119,8 @@
 #define MAGNUM_BIOS_SIZE_MAX 0x7e000
 #define MAGNUM_BIOS_SIZE                                                       \
         (BIOS_SIZE < MAGNUM_BIOS_SIZE_MAX ? BIOS_SIZE : MAGNUM_BIOS_SIZE_MAX)
+
+#if defined(CONFIG_TCG) && !defined(CONFIG_USER_ONLY)
 static void (*real_do_transaction_failed)(CPUState *cpu, hwaddr physaddr,
                                           vaddr addr, unsigned size,
                                           MMUAccessType access_type,
@@ -137,6 +142,7 @@
     (*real_do_transaction_failed)(cs, physaddr, addr, size, access_type,
                                   mmu_idx, attrs, response, retaddr);
 }
+#endif /* CONFIG_TCG && !CONFIG_USER_ONLY */
 
 static void mips_jazz_init(MachineState *machine,
                            enum jazz_model_e jazz_model)
@@ -205,8 +211,10 @@
      * memory region that catches all memory accesses, as we do on Malta.
      */
     cc = CPU_GET_CLASS(cpu);
-    real_do_transaction_failed = cc->do_transaction_failed;
-    cc->do_transaction_failed = mips_jazz_do_transaction_failed;
+#if defined(CONFIG_TCG) && !defined(CONFIG_USER_ONLY)
+    real_do_transaction_failed = cc->tcg_ops->do_transaction_failed;
+    cc->tcg_ops->do_transaction_failed = mips_jazz_do_transaction_failed;
+#endif /* CONFIG_TCG && !CONFIG_USER_ONLY */
 
     /* allocate RAM */
     memory_region_add_subregion(address_space, 0, machine->ram);
diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
index 4b5408c..cfb1d79 100644
--- a/include/exec/cpu-all.h
+++ b/include/exec/cpu-all.h
@@ -410,19 +410,26 @@
 }
 
 #ifdef CONFIG_TCG
+/* accel/tcg/cpu-exec.c */
 void dump_drift_info(void);
+/* accel/tcg/translate-all.c */
 void dump_exec_info(void);
 void dump_opcount_info(void);
 #endif /* CONFIG_TCG */
 
 #endif /* !CONFIG_USER_ONLY */
 
+#ifdef CONFIG_TCG
+/* accel/tcg/cpu-exec.c */
+int cpu_exec(CPUState *cpu);
+void tcg_exec_realizefn(CPUState *cpu, Error **errp);
+void tcg_exec_unrealizefn(CPUState *cpu);
+#endif /* CONFIG_TCG */
+
 /* Returns: 0 on success, -1 on error */
 int cpu_memory_rw_debug(CPUState *cpu, target_ulong addr,
                         void *ptr, target_ulong len, bool is_write);
 
-int cpu_exec(CPUState *cpu);
-
 /**
  * cpu_set_cpustate_pointers(cpu)
  * @cpu: The cpu object
diff --git a/include/exec/cpu-defs.h b/include/exec/cpu-defs.h
index d1f5e3f..ba3cd32 100644
--- a/include/exec/cpu-defs.h
+++ b/include/exec/cpu-defs.h
@@ -25,9 +25,6 @@
 
 #include "qemu/host-utils.h"
 #include "qemu/thread.h"
-#ifdef CONFIG_TCG
-#include "tcg-target.h"
-#endif
 #ifndef CONFIG_USER_ONLY
 #include "exec/hwaddr.h"
 #endif
diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h
index 125000b..f933c74 100644
--- a/include/exec/exec-all.h
+++ b/include/exec/exec-all.h
@@ -544,7 +544,7 @@
 
 /* GETPC is the true target of the return instruction that we'll execute.  */
 #if defined(CONFIG_TCG_INTERPRETER)
-extern uintptr_t tci_tb_ptr;
+extern __thread uintptr_t tci_tb_ptr;
 # define GETPC() tci_tb_ptr
 #else
 # define GETPC() \
diff --git a/include/hw/boards.h b/include/hw/boards.h
index 17b1f3f..85af4fa 100644
--- a/include/hw/boards.h
+++ b/include/hw/boards.h
@@ -6,7 +6,7 @@
 #include "exec/memory.h"
 #include "sysemu/hostmem.h"
 #include "sysemu/blockdev.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "qapi/qapi-types-machine.h"
 #include "qemu/module.h"
 #include "qom/object.h"
diff --git a/include/hw/core/accel-cpu.h b/include/hw/core/accel-cpu.h
new file mode 100644
index 0000000..24a6697
--- /dev/null
+++ b/include/hw/core/accel-cpu.h
@@ -0,0 +1,38 @@
+/*
+ * Accelerator interface, specializes CPUClass
+ * This header is used only by target-specific code.
+ *
+ * Copyright 2021 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef ACCEL_CPU_H
+#define ACCEL_CPU_H
+
+/*
+ * This header is used to define new accelerator-specific target-specific
+ * accelerator cpu subclasses.
+ * It uses CPU_RESOLVING_TYPE, so this is clearly target-specific.
+ *
+ * Do not try to use for any other purpose than the implementation of new
+ * subclasses in target/, or the accel implementation itself in accel/
+ */
+
+#define TYPE_ACCEL_CPU "accel-" CPU_RESOLVING_TYPE
+#define ACCEL_CPU_NAME(name) (name "-" TYPE_ACCEL_CPU)
+typedef struct AccelCPUClass AccelCPUClass;
+DECLARE_CLASS_CHECKERS(AccelCPUClass, ACCEL_CPU, TYPE_ACCEL_CPU)
+
+typedef struct AccelCPUClass {
+    /*< private >*/
+    ObjectClass parent_class;
+    /*< public >*/
+
+    void (*cpu_class_init)(CPUClass *cc);
+    void (*cpu_instance_init)(CPUState *cpu);
+    void (*cpu_realizefn)(CPUState *cpu, Error **errp);
+} AccelCPUClass;
+
+#endif /* ACCEL_CPU_H */
diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h
index 140fa32..38d813c 100644
--- a/include/hw/core/cpu.h
+++ b/include/hw/core/cpu.h
@@ -76,6 +76,12 @@
 
 struct TranslationBlock;
 
+/* see tcg-cpu-ops.h */
+struct TCGCPUOps;
+
+/* see accel-cpu.h */
+struct AccelCPUClass;
+
 /**
  * CPUClass:
  * @class_by_name: Callback to map -cpu command line model name to an
@@ -83,11 +89,6 @@
  * @parse_features: Callback to parse command line arguments.
  * @reset_dump_flags: #CPUDumpFlags to use for reset logging.
  * @has_work: Callback for checking if there is work to do.
- * @do_interrupt: Callback for interrupt handling.
- * @do_unaligned_access: Callback for unaligned access handling, if
- * the target defines #TARGET_ALIGNED_ONLY.
- * @do_transaction_failed: Callback for handling failed memory transactions
- * (ie bus faults or external aborts; not MMU faults)
  * @virtio_is_big_endian: Callback to return %true if a CPU which supports
  * runtime configurable endianness is currently big-endian. Non-configurable
  * CPUs can use the default implementation of this method. This method should
@@ -106,19 +107,6 @@
  *       If the target behaviour here is anything other than "set
  *       the PC register to the value passed in" then the target must
  *       also implement the synchronize_from_tb hook.
- * @synchronize_from_tb: Callback for synchronizing state from a TCG
- *       #TranslationBlock. This is called when we abandon execution
- *       of a TB before starting it, and must set all parts of the CPU
- *       state which the previous TB in the chain may not have updated.
- *       This always includes at least the program counter; some targets
- *       will need to do more. If this hook is not implemented then the
- *       default is to call @set_pc(tb->pc).
- * @tlb_fill: Callback for handling a softmmu tlb miss or user-only
- *       address fault.  For system mode, if the access is valid, call
- *       tlb_set_page and return true; if the access is invalid, and
- *       probe is true, return false; otherwise raise an exception and
- *       do not return.  For user-only mode, always raise an exception
- *       and do not return.
  * @get_phys_page_debug: Callback for obtaining a physical address.
  * @get_phys_page_attrs_debug: Callback for obtaining a physical address and the
  *       associated memory transaction attributes to use for the access.
@@ -128,9 +116,6 @@
  *       a memory access with the specified memory transaction attributes.
  * @gdb_read_register: Callback for letting GDB read a register.
  * @gdb_write_register: Callback for letting GDB write a register.
- * @debug_check_watchpoint: Callback: return true if the architectural
- *       watchpoint whose address has matched should really fire.
- * @debug_excp_handler: Callback for handling debug exceptions.
  * @write_elf64_note: Callback for writing a CPU-specific ELF note to a
  * 64-bit VM coredump.
  * @write_elf32_qemunote: Callback for writing a CPU- and QEMU-specific ELF
@@ -149,9 +134,6 @@
  * @gdb_get_dynamic_xml: Callback to return dynamically generated XML for the
  *   gdb stub. Returns a pointer to the XML contents for the specified XML file
  *   or NULL if the CPU doesn't have a dynamically generated content for it.
- * @cpu_exec_enter: Callback for cpu_exec preparation.
- * @cpu_exec_exit: Callback for cpu_exec cleanup.
- * @cpu_exec_interrupt: Callback for processing interrupts in cpu_exec.
  * @disas_set_info: Setup architecture specific components of disassembly info
  * @adjust_watchpoint_address: Perform a target-specific adjustment to an
  * address before attempting to match it against watchpoints.
@@ -170,14 +152,6 @@
 
     int reset_dump_flags;
     bool (*has_work)(CPUState *cpu);
-    void (*do_interrupt)(CPUState *cpu);
-    void (*do_unaligned_access)(CPUState *cpu, vaddr addr,
-                                MMUAccessType access_type,
-                                int mmu_idx, uintptr_t retaddr);
-    void (*do_transaction_failed)(CPUState *cpu, hwaddr physaddr, vaddr addr,
-                                  unsigned size, MMUAccessType access_type,
-                                  int mmu_idx, MemTxAttrs attrs,
-                                  MemTxResult response, uintptr_t retaddr);
     bool (*virtio_is_big_endian)(CPUState *cpu);
     int (*memory_rw_debug)(CPUState *cpu, vaddr addr,
                            uint8_t *buf, int len, bool is_write);
@@ -189,19 +163,12 @@
     void (*get_memory_mapping)(CPUState *cpu, MemoryMappingList *list,
                                Error **errp);
     void (*set_pc)(CPUState *cpu, vaddr value);
-    void (*synchronize_from_tb)(CPUState *cpu,
-                                const struct TranslationBlock *tb);
-    bool (*tlb_fill)(CPUState *cpu, vaddr address, int size,
-                     MMUAccessType access_type, int mmu_idx,
-                     bool probe, uintptr_t retaddr);
     hwaddr (*get_phys_page_debug)(CPUState *cpu, vaddr addr);
     hwaddr (*get_phys_page_attrs_debug)(CPUState *cpu, vaddr addr,
                                         MemTxAttrs *attrs);
     int (*asidx_from_attrs)(CPUState *cpu, MemTxAttrs attrs);
     int (*gdb_read_register)(CPUState *cpu, GByteArray *buf, int reg);
     int (*gdb_write_register)(CPUState *cpu, uint8_t *buf, int reg);
-    bool (*debug_check_watchpoint)(CPUState *cpu, CPUWatchpoint *wp);
-    void (*debug_excp_handler)(CPUState *cpu);
 
     int (*write_elf64_note)(WriteCoreDumpFunction f, CPUState *cpu,
                             int cpuid, void *opaque);
@@ -216,18 +183,17 @@
     const char *gdb_core_xml_file;
     gchar * (*gdb_arch_name)(CPUState *cpu);
     const char * (*gdb_get_dynamic_xml)(CPUState *cpu, const char *xmlname);
-    void (*cpu_exec_enter)(CPUState *cpu);
-    void (*cpu_exec_exit)(CPUState *cpu);
-    bool (*cpu_exec_interrupt)(CPUState *cpu, int interrupt_request);
 
     void (*disas_set_info)(CPUState *cpu, disassemble_info *info);
-    vaddr (*adjust_watchpoint_address)(CPUState *cpu, vaddr addr, int len);
-    void (*tcg_initialize)(void);
 
     const char *deprecation_note;
     /* Keep non-pointer data at the end to minimize holes.  */
     int gdb_num_core_regs;
     bool gdb_stop_before_watchpoint;
+    struct AccelCPUClass *accel_cpu;
+
+    /* when TCG is not available, this pointer is NULL */
+    struct TCGCPUOps *tcg_ops;
 };
 
 /*
@@ -858,36 +824,6 @@
 
 void cpu_interrupt(CPUState *cpu, int mask);
 
-#ifdef NEED_CPU_H
-
-#ifdef CONFIG_SOFTMMU
-static inline void cpu_unaligned_access(CPUState *cpu, vaddr addr,
-                                        MMUAccessType access_type,
-                                        int mmu_idx, uintptr_t retaddr)
-{
-    CPUClass *cc = CPU_GET_CLASS(cpu);
-
-    cc->do_unaligned_access(cpu, addr, access_type, mmu_idx, retaddr);
-}
-
-static inline void cpu_transaction_failed(CPUState *cpu, hwaddr physaddr,
-                                          vaddr addr, unsigned size,
-                                          MMUAccessType access_type,
-                                          int mmu_idx, MemTxAttrs attrs,
-                                          MemTxResult response,
-                                          uintptr_t retaddr)
-{
-    CPUClass *cc = CPU_GET_CLASS(cpu);
-
-    if (!cpu->ignore_memory_transaction_failures && cc->do_transaction_failed) {
-        cc->do_transaction_failed(cpu, physaddr, addr, size, access_type,
-                                  mmu_idx, attrs, response, retaddr);
-    }
-}
-#endif
-
-#endif /* NEED_CPU_H */
-
 /**
  * cpu_set_pc:
  * @cpu: The CPU to set the program counter for.
@@ -1112,6 +1048,8 @@
 
 void QEMU_NORETURN cpu_abort(CPUState *cpu, const char *fmt, ...)
     GCC_FMT_ATTR(2, 3);
+
+/* $(top_srcdir)/cpu.c */
 void cpu_exec_initfn(CPUState *cpu);
 void cpu_exec_realizefn(CPUState *cpu, Error **errp);
 void cpu_exec_unrealizefn(CPUState *cpu);
diff --git a/include/hw/core/tcg-cpu-ops.h b/include/hw/core/tcg-cpu-ops.h
new file mode 100644
index 0000000..ccc97d1
--- /dev/null
+++ b/include/hw/core/tcg-cpu-ops.h
@@ -0,0 +1,97 @@
+/*
+ * TCG CPU-specific operations
+ *
+ * Copyright 2021 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef TCG_CPU_OPS_H
+#define TCG_CPU_OPS_H
+
+#include "hw/core/cpu.h"
+
+struct TCGCPUOps {
+    /**
+     * @initialize: Initalize TCG state
+     *
+     * Called when the first CPU is realized.
+     */
+    void (*initialize)(void);
+    /**
+     * @synchronize_from_tb: Synchronize state from a TCG #TranslationBlock
+     *
+     * This is called when we abandon execution of a TB before starting it,
+     * and must set all parts of the CPU state which the previous TB in the
+     * chain may not have updated.
+     * By default, when this is NULL, a call is made to @set_pc(tb->pc).
+     *
+     * If more state needs to be restored, the target must implement a
+     * function to restore all the state, and register it here.
+     */
+    void (*synchronize_from_tb)(CPUState *cpu,
+                                const struct TranslationBlock *tb);
+    /** @cpu_exec_enter: Callback for cpu_exec preparation */
+    void (*cpu_exec_enter)(CPUState *cpu);
+    /** @cpu_exec_exit: Callback for cpu_exec cleanup */
+    void (*cpu_exec_exit)(CPUState *cpu);
+    /** @cpu_exec_interrupt: Callback for processing interrupts in cpu_exec */
+    bool (*cpu_exec_interrupt)(CPUState *cpu, int interrupt_request);
+    /**
+     * @do_interrupt: Callback for interrupt handling.
+     *
+     * note that this is in general SOFTMMU only, but it actually isn't
+     * because of an x86 hack (accel/tcg/cpu-exec.c), so we cannot put it
+     * in the SOFTMMU section in general.
+     */
+    void (*do_interrupt)(CPUState *cpu);
+    /**
+     * @tlb_fill: Handle a softmmu tlb miss or user-only address fault
+     *
+     * For system mode, if the access is valid, call tlb_set_page
+     * and return true; if the access is invalid, and probe is
+     * true, return false; otherwise raise an exception and do
+     * not return.  For user-only mode, always raise an exception
+     * and do not return.
+     */
+    bool (*tlb_fill)(CPUState *cpu, vaddr address, int size,
+                     MMUAccessType access_type, int mmu_idx,
+                     bool probe, uintptr_t retaddr);
+    /** @debug_excp_handler: Callback for handling debug exceptions */
+    void (*debug_excp_handler)(CPUState *cpu);
+
+#ifdef NEED_CPU_H
+#ifdef CONFIG_SOFTMMU
+    /**
+     * @do_transaction_failed: Callback for handling failed memory transactions
+     * (ie bus faults or external aborts; not MMU faults)
+     */
+    void (*do_transaction_failed)(CPUState *cpu, hwaddr physaddr, vaddr addr,
+                                  unsigned size, MMUAccessType access_type,
+                                  int mmu_idx, MemTxAttrs attrs,
+                                  MemTxResult response, uintptr_t retaddr);
+    /**
+     * @do_unaligned_access: Callback for unaligned access handling
+     */
+    void (*do_unaligned_access)(CPUState *cpu, vaddr addr,
+                                MMUAccessType access_type,
+                                int mmu_idx, uintptr_t retaddr);
+
+    /**
+     * @adjust_watchpoint_address: hack for cpu_check_watchpoint used by ARM
+     */
+    vaddr (*adjust_watchpoint_address)(CPUState *cpu, vaddr addr, int len);
+
+    /**
+     * @debug_check_watchpoint: return true if the architectural
+     * watchpoint whose address has matched should really fire, used by ARM
+     */
+    bool (*debug_check_watchpoint)(CPUState *cpu, CPUWatchpoint *wp);
+
+#endif /* CONFIG_SOFTMMU */
+#endif /* NEED_CPU_H */
+
+};
+
+#endif /* TCG_CPU_OPS_H */
diff --git a/include/sysemu/accel.h b/include/qemu/accel.h
similarity index 93%
rename from include/sysemu/accel.h
rename to include/qemu/accel.h
index e08b8ab..b9d6d69 100644
--- a/include/sysemu/accel.h
+++ b/include/qemu/accel.h
@@ -20,8 +20,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  * THE SOFTWARE.
  */
-#ifndef HW_ACCEL_H
-#define HW_ACCEL_H
+#ifndef QEMU_ACCEL_H
+#define QEMU_ACCEL_H
 
 #include "qom/object.h"
 #include "exec/hwaddr.h"
@@ -37,8 +37,8 @@
     /*< public >*/
 
     const char *name;
-#ifndef CONFIG_USER_ONLY
     int (*init_machine)(MachineState *ms);
+#ifndef CONFIG_USER_ONLY
     void (*setup_post)(MachineState *ms, AccelState *accel);
     bool (*has_memory)(MachineState *ms, AddressSpace *as,
                        hwaddr start_addr, hwaddr size);
@@ -67,11 +67,15 @@
     OBJECT_GET_CLASS(AccelClass, (obj), TYPE_ACCEL)
 
 AccelClass *accel_find(const char *opt_name);
+AccelState *current_accel(void);
+
+void accel_init_interfaces(AccelClass *ac);
+
+#ifndef CONFIG_USER_ONLY
 int accel_init_machine(AccelState *accel, MachineState *ms);
 
 /* Called just before os_setup_post (ie just before drop OS privs) */
 void accel_setup_post(MachineState *ms);
+#endif /* !CONFIG_USER_ONLY */
 
-AccelState *current_accel(void);
-
-#endif
+#endif /* QEMU_ACCEL_H */
diff --git a/include/sysemu/accel-ops.h b/include/sysemu/accel-ops.h
new file mode 100644
index 0000000..032f697
--- /dev/null
+++ b/include/sysemu/accel-ops.h
@@ -0,0 +1,45 @@
+/*
+ * Accelerator OPS, used for cpus.c module
+ *
+ * Copyright 2021 SUSE LLC
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef ACCEL_OPS_H
+#define ACCEL_OPS_H
+
+#include "qom/object.h"
+
+#define ACCEL_OPS_SUFFIX "-ops"
+#define TYPE_ACCEL_OPS "accel" ACCEL_OPS_SUFFIX
+#define ACCEL_OPS_NAME(name) (name "-" TYPE_ACCEL_OPS)
+
+typedef struct AccelOpsClass AccelOpsClass;
+DECLARE_CLASS_CHECKERS(AccelOpsClass, ACCEL_OPS, TYPE_ACCEL_OPS)
+
+/* cpus.c operations interface */
+struct AccelOpsClass {
+    /*< private >*/
+    ObjectClass parent_class;
+    /*< public >*/
+
+    /* initialization function called when accel is chosen */
+    void (*ops_init)(AccelOpsClass *ops);
+
+    void (*create_vcpu_thread)(CPUState *cpu); /* MANDATORY NON-NULL */
+    void (*kick_vcpu_thread)(CPUState *cpu);
+
+    void (*synchronize_post_reset)(CPUState *cpu);
+    void (*synchronize_post_init)(CPUState *cpu);
+    void (*synchronize_state)(CPUState *cpu);
+    void (*synchronize_pre_loadvm)(CPUState *cpu);
+
+    void (*handle_interrupt)(CPUState *cpu, int mask);
+
+    int64_t (*get_virtual_clock)(void);
+    int64_t (*get_elapsed_ticks)(void);
+};
+
+#endif /* ACCEL_OPS_H */
diff --git a/include/sysemu/cpus.h b/include/sysemu/cpus.h
index e815672..2cd7439 100644
--- a/include/sysemu/cpus.h
+++ b/include/sysemu/cpus.h
@@ -2,30 +2,14 @@
 #define QEMU_CPUS_H
 
 #include "qemu/timer.h"
+#include "sysemu/accel-ops.h"
 
-/* cpus.c */
+/* register accel-specific operations */
+void cpus_register_accel(const AccelOpsClass *i);
 
-/* CPU execution threads */
+/* accel/dummy-cpus.c */
 
-typedef struct CpusAccel {
-    void (*create_vcpu_thread)(CPUState *cpu); /* MANDATORY */
-    void (*kick_vcpu_thread)(CPUState *cpu);
-
-    void (*synchronize_post_reset)(CPUState *cpu);
-    void (*synchronize_post_init)(CPUState *cpu);
-    void (*synchronize_state)(CPUState *cpu);
-    void (*synchronize_pre_loadvm)(CPUState *cpu);
-
-    void (*handle_interrupt)(CPUState *cpu, int mask);
-
-    int64_t (*get_virtual_clock)(void);
-    int64_t (*get_elapsed_ticks)(void);
-} CpusAccel;
-
-/* register accel-specific cpus interface implementation */
-void cpus_register_accel(const CpusAccel *i);
-
-/* Create a dummy vcpu for CpusAccel->create_vcpu_thread */
+/* Create a dummy vcpu for AccelOpsClass->create_vcpu_thread */
 void dummy_start_vcpu_thread(CPUState *);
 
 /* interface available for cpus accelerator threads */
diff --git a/include/sysemu/hvf.h b/include/sysemu/hvf.h
index f893768..c98636b 100644
--- a/include/sysemu/hvf.h
+++ b/include/sysemu/hvf.h
@@ -13,7 +13,7 @@
 #ifndef HVF_H
 #define HVF_H
 
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "qom/object.h"
 
 #ifdef CONFIG_HVF
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index bb5d5cf..739682f 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -17,7 +17,7 @@
 #include "qemu/queue.h"
 #include "hw/core/cpu.h"
 #include "exec/memattrs.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "qom/object.h"
 
 #ifdef NEED_CPU_H
diff --git a/include/sysemu/kvm_int.h b/include/sysemu/kvm_int.h
index 6574080..ccb8869 100644
--- a/include/sysemu/kvm_int.h
+++ b/include/sysemu/kvm_int.h
@@ -10,7 +10,7 @@
 #define QEMU_KVM_INT_H
 
 #include "exec/memory.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/kvm.h"
 
 typedef struct KVMSlot
diff --git a/linux-user/main.c b/linux-user/main.c
index bb4e55e..2e3c169 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -20,6 +20,7 @@
 #include "qemu/osdep.h"
 #include "qemu-common.h"
 #include "qemu/units.h"
+#include "qemu/accel.h"
 #include "sysemu/tcg.h"
 #include "qemu-version.h"
 #include <sys/syscall.h>
@@ -701,8 +702,12 @@
     cpu_type = parse_cpu_option(cpu_model);
 
     /* init tcg before creating CPUs and to get qemu_host_page_size */
-    tcg_exec_init(0, false);
+    {
+        AccelClass *ac = ACCEL_GET_CLASS(current_accel());
 
+        ac->init_machine(NULL);
+        accel_init_interfaces(ac);
+    }
     cpu = cpu_create(cpu_type);
     env = cpu->env_ptr;
     cpu_reset(cpu);
diff --git a/softmmu/cpus.c b/softmmu/cpus.c
index 1dc20b9..112eba9 100644
--- a/softmmu/cpus.c
+++ b/softmmu/cpus.c
@@ -128,7 +128,7 @@
 /*
  * The chosen accelerator is supposed to register this.
  */
-static const CpusAccel *cpus_accel;
+static const AccelOpsClass *cpus_accel;
 
 void cpu_synchronize_all_states(void)
 {
@@ -594,11 +594,11 @@
     qemu_mutex_lock_iothread();
 }
 
-void cpus_register_accel(const CpusAccel *ca)
+void cpus_register_accel(const AccelOpsClass *ops)
 {
-    assert(ca != NULL);
-    assert(ca->create_vcpu_thread != NULL); /* mandatory */
-    cpus_accel = ca;
+    assert(ops != NULL);
+    assert(ops->create_vcpu_thread != NULL); /* mandatory */
+    cpus_accel = ops;
 }
 
 void qemu_init_vcpu(CPUState *cpu)
@@ -618,7 +618,7 @@
         cpu_address_space_init(cpu, 0, "cpu-memory", cpu->memory);
     }
 
-    /* accelerators all implement the CpusAccel interface */
+    /* accelerators all implement the AccelOpsClass */
     g_assert(cpus_accel != NULL && cpus_accel->create_vcpu_thread != NULL);
     cpus_accel->create_vcpu_thread(cpu);
 
diff --git a/softmmu/memory.c b/softmmu/memory.c
index 676c298..c0c814f 100644
--- a/softmmu/memory.c
+++ b/softmmu/memory.c
@@ -32,7 +32,7 @@
 #include "sysemu/kvm.h"
 #include "sysemu/runstate.h"
 #include "sysemu/tcg.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "hw/boards.h"
 #include "migration/vmstate.h"
 
diff --git a/softmmu/physmem.c b/softmmu/physmem.c
index 60760a3..243c309 100644
--- a/softmmu/physmem.c
+++ b/softmmu/physmem.c
@@ -24,6 +24,11 @@
 #include "qemu/cutils.h"
 #include "qemu/cacheflush.h"
 #include "cpu.h"
+
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+#endif /* CONFIG_TCG */
+
 #include "exec/exec-all.h"
 #include "exec/target_page.h"
 #include "hw/qdev-core.h"
@@ -840,6 +845,7 @@
     }
 }
 
+#ifdef CONFIG_TCG
 /* Return true if this watchpoint address matches the specified
  * access (ie the address range covered by the watchpoint overlaps
  * partially or completely with the address range covered by the
@@ -873,6 +879,80 @@
     return ret;
 }
 
+/* Generate a debug exception if a watchpoint has been hit.  */
+void cpu_check_watchpoint(CPUState *cpu, vaddr addr, vaddr len,
+                          MemTxAttrs attrs, int flags, uintptr_t ra)
+{
+    CPUClass *cc = CPU_GET_CLASS(cpu);
+    CPUWatchpoint *wp;
+
+    assert(tcg_enabled());
+    if (cpu->watchpoint_hit) {
+        /*
+         * We re-entered the check after replacing the TB.
+         * Now raise the debug interrupt so that it will
+         * trigger after the current instruction.
+         */
+        qemu_mutex_lock_iothread();
+        cpu_interrupt(cpu, CPU_INTERRUPT_DEBUG);
+        qemu_mutex_unlock_iothread();
+        return;
+    }
+
+    if (cc->tcg_ops->adjust_watchpoint_address) {
+        /* this is currently used only by ARM BE32 */
+        addr = cc->tcg_ops->adjust_watchpoint_address(cpu, addr, len);
+    }
+    QTAILQ_FOREACH(wp, &cpu->watchpoints, entry) {
+        if (watchpoint_address_matches(wp, addr, len)
+            && (wp->flags & flags)) {
+            if (replay_running_debug()) {
+                /*
+                 * Don't process the watchpoints when we are
+                 * in a reverse debugging operation.
+                 */
+                replay_breakpoint();
+                return;
+            }
+            if (flags == BP_MEM_READ) {
+                wp->flags |= BP_WATCHPOINT_HIT_READ;
+            } else {
+                wp->flags |= BP_WATCHPOINT_HIT_WRITE;
+            }
+            wp->hitaddr = MAX(addr, wp->vaddr);
+            wp->hitattrs = attrs;
+            if (!cpu->watchpoint_hit) {
+                if (wp->flags & BP_CPU && cc->tcg_ops->debug_check_watchpoint &&
+                    !cc->tcg_ops->debug_check_watchpoint(cpu, wp)) {
+                    wp->flags &= ~BP_WATCHPOINT_HIT;
+                    continue;
+                }
+                cpu->watchpoint_hit = wp;
+
+                mmap_lock();
+                tb_check_watchpoint(cpu, ra);
+                if (wp->flags & BP_STOP_BEFORE_ACCESS) {
+                    cpu->exception_index = EXCP_DEBUG;
+                    mmap_unlock();
+                    cpu_loop_exit_restore(cpu, ra);
+                } else {
+                    /* Force execution of one insn next time.  */
+                    cpu->cflags_next_tb = 1 | curr_cflags();
+                    mmap_unlock();
+                    if (ra) {
+                        cpu_restore_state(cpu, ra, true);
+                    }
+                    cpu_loop_exit_noexc(cpu);
+                }
+            }
+        } else {
+            wp->flags &= ~BP_WATCHPOINT_HIT;
+        }
+    }
+}
+
+#endif /* CONFIG_TCG */
+
 /* Called from RCU critical section */
 static RAMBlock *qemu_get_ram_block(ram_addr_t addr)
 {
@@ -2359,75 +2439,6 @@
     return block->offset + offset;
 }
 
-/* Generate a debug exception if a watchpoint has been hit.  */
-void cpu_check_watchpoint(CPUState *cpu, vaddr addr, vaddr len,
-                          MemTxAttrs attrs, int flags, uintptr_t ra)
-{
-    CPUClass *cc = CPU_GET_CLASS(cpu);
-    CPUWatchpoint *wp;
-
-    assert(tcg_enabled());
-    if (cpu->watchpoint_hit) {
-        /*
-         * We re-entered the check after replacing the TB.
-         * Now raise the debug interrupt so that it will
-         * trigger after the current instruction.
-         */
-        qemu_mutex_lock_iothread();
-        cpu_interrupt(cpu, CPU_INTERRUPT_DEBUG);
-        qemu_mutex_unlock_iothread();
-        return;
-    }
-
-    addr = cc->adjust_watchpoint_address(cpu, addr, len);
-    QTAILQ_FOREACH(wp, &cpu->watchpoints, entry) {
-        if (watchpoint_address_matches(wp, addr, len)
-            && (wp->flags & flags)) {
-            if (replay_running_debug()) {
-                /*
-                 * Don't process the watchpoints when we are
-                 * in a reverse debugging operation.
-                 */
-                replay_breakpoint();
-                return;
-            }
-            if (flags == BP_MEM_READ) {
-                wp->flags |= BP_WATCHPOINT_HIT_READ;
-            } else {
-                wp->flags |= BP_WATCHPOINT_HIT_WRITE;
-            }
-            wp->hitaddr = MAX(addr, wp->vaddr);
-            wp->hitattrs = attrs;
-            if (!cpu->watchpoint_hit) {
-                if (wp->flags & BP_CPU &&
-                    !cc->debug_check_watchpoint(cpu, wp)) {
-                    wp->flags &= ~BP_WATCHPOINT_HIT;
-                    continue;
-                }
-                cpu->watchpoint_hit = wp;
-
-                mmap_lock();
-                tb_check_watchpoint(cpu, ra);
-                if (wp->flags & BP_STOP_BEFORE_ACCESS) {
-                    cpu->exception_index = EXCP_DEBUG;
-                    mmap_unlock();
-                    cpu_loop_exit_restore(cpu, ra);
-                } else {
-                    /* Force execution of one insn next time.  */
-                    cpu->cflags_next_tb = 1 | curr_cflags();
-                    mmap_unlock();
-                    if (ra) {
-                        cpu_restore_state(cpu, ra, true);
-                    }
-                    cpu_loop_exit_noexc(cpu);
-                }
-            }
-        } else {
-            wp->flags &= ~BP_WATCHPOINT_HIT;
-        }
-    }
-}
-
 static MemTxResult flatview_read(FlatView *fv, hwaddr addr,
                                  MemTxAttrs attrs, void *buf, hwaddr len);
 static MemTxResult flatview_write(FlatView *fv, hwaddr addr, MemTxAttrs attrs,
diff --git a/softmmu/qtest.c b/softmmu/qtest.c
index 7965dc9..130c366 100644
--- a/softmmu/qtest.c
+++ b/softmmu/qtest.c
@@ -20,7 +20,7 @@
 #include "exec/ioport.h"
 #include "exec/memory.h"
 #include "hw/irq.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/cpu-timers.h"
 #include "qemu/config-file.h"
 #include "qemu/option.h"
diff --git a/softmmu/vl.c b/softmmu/vl.c
index bd55468..2bf94ec 100644
--- a/softmmu/vl.c
+++ b/softmmu/vl.c
@@ -44,7 +44,7 @@
 
 #include "qemu/error-report.h"
 #include "qemu/sockets.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "hw/usb.h"
 #include "hw/isa/isa.h"
 #include "hw/scsi/scsi.h"
@@ -1726,7 +1726,8 @@
         return false;
     }
 
-    /* Allocation of large amounts of memory may delay
+    /*
+     * Allocation of large amounts of memory may delay
      * chardev initialization for too long, and trigger timeouts
      * on software that waits for a monitor socket to be created
      * (e.g. libvirt).
@@ -3497,7 +3498,7 @@
      *
      * Machine compat properties: object_set_machine_compat_props().
      * Accelerator compat props: object_set_accelerator_compat_props(),
-     * called from configure_accelerator().
+     * called from do_configure_accelerator().
      */
 
     machine_class = MACHINE_GET_CLASS(current_machine);
@@ -3519,6 +3520,8 @@
     if (cpu_option) {
         current_machine->cpu_type = parse_cpu_option(cpu_option);
     }
+    /* NB: for machine none cpu_type could STILL be NULL here! */
+    accel_init_interfaces(ACCEL_GET_CLASS(current_machine->accelerator));
 
     qemu_resolve_machine_memdev();
     parse_numa_opts(current_machine);
diff --git a/target/alpha/cpu.c b/target/alpha/cpu.c
index b3fd664..27192b6 100644
--- a/target/alpha/cpu.c
+++ b/target/alpha/cpu.c
@@ -206,6 +206,20 @@
 #endif
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps alpha_tcg_ops = {
+    .initialize = alpha_translate_init,
+    .cpu_exec_interrupt = alpha_cpu_exec_interrupt,
+    .tlb_fill = alpha_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = alpha_cpu_do_interrupt,
+    .do_transaction_failed = alpha_cpu_do_transaction_failed,
+    .do_unaligned_access = alpha_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void alpha_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -217,22 +231,17 @@
 
     cc->class_by_name = alpha_cpu_class_by_name;
     cc->has_work = alpha_cpu_has_work;
-    cc->do_interrupt = alpha_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = alpha_cpu_exec_interrupt;
     cc->dump_state = alpha_cpu_dump_state;
     cc->set_pc = alpha_cpu_set_pc;
     cc->gdb_read_register = alpha_cpu_gdb_read_register;
     cc->gdb_write_register = alpha_cpu_gdb_write_register;
-    cc->tlb_fill = alpha_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
-    cc->do_transaction_failed = alpha_cpu_do_transaction_failed;
-    cc->do_unaligned_access = alpha_cpu_do_unaligned_access;
     cc->get_phys_page_debug = alpha_cpu_get_phys_page_debug;
     dc->vmsd = &vmstate_alpha_cpu;
 #endif
     cc->disas_set_info = alpha_cpu_disas_set_info;
-    cc->tcg_initialize = alpha_translate_init;
 
+    cc->tcg_ops = &alpha_tcg_ops;
     cc->gdb_num_core_regs = 67;
 }
 
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 40142ac..8ddb255 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -26,6 +26,9 @@
 #include "qapi/error.h"
 #include "qapi/visitor.h"
 #include "cpu.h"
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+#endif /* CONFIG_TCG */
 #include "internals.h"
 #include "exec/exec-all.h"
 #include "hw/qdev-properties.h"
@@ -54,8 +57,9 @@
     }
 }
 
-static void arm_cpu_synchronize_from_tb(CPUState *cs,
-                                        const TranslationBlock *tb)
+#ifdef CONFIG_TCG
+void arm_cpu_synchronize_from_tb(CPUState *cs,
+                                 const TranslationBlock *tb)
 {
     ARMCPU *cpu = ARM_CPU(cs);
     CPUARMState *env = &cpu->env;
@@ -70,6 +74,7 @@
         env->regs[15] = tb->pc;
     }
 }
+#endif /* CONFIG_TCG */
 
 static bool arm_cpu_has_work(CPUState *cs)
 {
@@ -588,7 +593,7 @@
  found:
     cs->exception_index = excp_idx;
     env->exception.target_el = target_el;
-    cc->do_interrupt(cs);
+    cc->tcg_ops->do_interrupt(cs);
     return true;
 }
 
@@ -2240,6 +2245,24 @@
     return g_strdup("arm");
 }
 
+#ifdef CONFIG_TCG
+static struct TCGCPUOps arm_tcg_ops = {
+    .initialize = arm_translate_init,
+    .synchronize_from_tb = arm_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = arm_cpu_exec_interrupt,
+    .tlb_fill = arm_cpu_tlb_fill,
+    .debug_excp_handler = arm_debug_excp_handler,
+
+#if !defined(CONFIG_USER_ONLY)
+    .do_interrupt = arm_cpu_do_interrupt,
+    .do_transaction_failed = arm_cpu_do_transaction_failed,
+    .do_unaligned_access = arm_cpu_do_unaligned_access,
+    .adjust_watchpoint_address = arm_adjust_watchpoint_address,
+    .debug_check_watchpoint = arm_debug_check_watchpoint,
+#endif /* !CONFIG_USER_ONLY */
+};
+#endif /* CONFIG_TCG */
+
 static void arm_cpu_class_init(ObjectClass *oc, void *data)
 {
     ARMCPUClass *acc = ARM_CPU_CLASS(oc);
@@ -2254,14 +2277,11 @@
 
     cc->class_by_name = arm_cpu_class_by_name;
     cc->has_work = arm_cpu_has_work;
-    cc->cpu_exec_interrupt = arm_cpu_exec_interrupt;
     cc->dump_state = arm_cpu_dump_state;
     cc->set_pc = arm_cpu_set_pc;
-    cc->synchronize_from_tb = arm_cpu_synchronize_from_tb;
     cc->gdb_read_register = arm_cpu_gdb_read_register;
     cc->gdb_write_register = arm_cpu_gdb_write_register;
 #ifndef CONFIG_USER_ONLY
-    cc->do_interrupt = arm_cpu_do_interrupt;
     cc->get_phys_page_attrs_debug = arm_cpu_get_phys_page_attrs_debug;
     cc->asidx_from_attrs = arm_asidx_from_attrs;
     cc->vmsd = &vmstate_arm_cpu;
@@ -2275,17 +2295,10 @@
     cc->gdb_get_dynamic_xml = arm_gdb_get_dynamic_xml;
     cc->gdb_stop_before_watchpoint = true;
     cc->disas_set_info = arm_disas_set_info;
+
 #ifdef CONFIG_TCG
-    cc->tcg_initialize = arm_translate_init;
-    cc->tlb_fill = arm_cpu_tlb_fill;
-    cc->debug_excp_handler = arm_debug_excp_handler;
-    cc->debug_check_watchpoint = arm_debug_check_watchpoint;
-    cc->do_unaligned_access = arm_cpu_do_unaligned_access;
-#if !defined(CONFIG_USER_ONLY)
-    cc->do_transaction_failed = arm_cpu_do_transaction_failed;
-    cc->adjust_watchpoint_address = arm_adjust_watchpoint_address;
-#endif /* CONFIG_TCG && !CONFIG_USER_ONLY */
-#endif
+    cc->tcg_ops = &arm_tcg_ops;
+#endif /* CONFIG_TCG */
 }
 
 #ifdef CONFIG_KVM
diff --git a/target/arm/cpu64.c b/target/arm/cpu64.c
index 5e85102..10c5118 100644
--- a/target/arm/cpu64.c
+++ b/target/arm/cpu64.c
@@ -21,6 +21,9 @@
 #include "qemu/osdep.h"
 #include "qapi/error.h"
 #include "cpu.h"
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+#endif /* CONFIG_TCG */
 #include "qemu/module.h"
 #if !defined(CONFIG_USER_ONLY)
 #include "hw/loader.h"
@@ -805,7 +808,6 @@
 {
     CPUClass *cc = CPU_CLASS(oc);
 
-    cc->cpu_exec_interrupt = arm_cpu_exec_interrupt;
     cc->gdb_read_register = aarch64_cpu_gdb_read_register;
     cc->gdb_write_register = aarch64_cpu_gdb_write_register;
     cc->gdb_num_core_regs = 34;
diff --git a/target/arm/cpu_tcg.c b/target/arm/cpu_tcg.c
index 98544db..c29b434 100644
--- a/target/arm/cpu_tcg.c
+++ b/target/arm/cpu_tcg.c
@@ -10,11 +10,15 @@
 
 #include "qemu/osdep.h"
 #include "cpu.h"
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+#endif /* CONFIG_TCG */
 #include "internals.h"
 
 /* CPU models. These are not needed for the AArch64 linux-user build. */
 #if !defined(CONFIG_USER_ONLY) || !defined(TARGET_AARCH64)
 
+#ifdef CONFIG_TCG
 static bool arm_v7m_cpu_exec_interrupt(CPUState *cs, int interrupt_request)
 {
     CPUClass *cc = CPU_GET_CLASS(cs);
@@ -33,11 +37,12 @@
     if (interrupt_request & CPU_INTERRUPT_HARD
         && (armv7m_nvic_can_take_pending_exception(env->nvic))) {
         cs->exception_index = EXCP_IRQ;
-        cc->do_interrupt(cs);
+        cc->tcg_ops->do_interrupt(cs);
         ret = true;
     }
     return ret;
 }
+#endif /* CONFIG_TCG */
 
 static void arm926_initfn(Object *obj)
 {
@@ -658,17 +663,34 @@
     cpu->reset_sctlr = 0x00000078;
 }
 
+#ifdef CONFIG_TCG
+static struct TCGCPUOps arm_v7m_tcg_ops = {
+    .initialize = arm_translate_init,
+    .synchronize_from_tb = arm_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = arm_v7m_cpu_exec_interrupt,
+    .tlb_fill = arm_cpu_tlb_fill,
+    .debug_excp_handler = arm_debug_excp_handler,
+
+#if !defined(CONFIG_USER_ONLY)
+    .do_interrupt = arm_v7m_cpu_do_interrupt,
+    .do_transaction_failed = arm_cpu_do_transaction_failed,
+    .do_unaligned_access = arm_cpu_do_unaligned_access,
+    .adjust_watchpoint_address = arm_adjust_watchpoint_address,
+    .debug_check_watchpoint = arm_debug_check_watchpoint,
+#endif /* !CONFIG_USER_ONLY */
+};
+#endif /* CONFIG_TCG */
+
 static void arm_v7m_class_init(ObjectClass *oc, void *data)
 {
     ARMCPUClass *acc = ARM_CPU_CLASS(oc);
     CPUClass *cc = CPU_CLASS(oc);
 
     acc->info = data;
-#ifndef CONFIG_USER_ONLY
-    cc->do_interrupt = arm_v7m_cpu_do_interrupt;
-#endif
+#ifdef CONFIG_TCG
+    cc->tcg_ops = &arm_v7m_tcg_ops;
+#endif /* CONFIG_TCG */
 
-    cc->cpu_exec_interrupt = arm_v7m_cpu_exec_interrupt;
     cc->gdb_core_xml_file = "arm-m-profile.xml";
 }
 
diff --git a/target/arm/helper.c b/target/arm/helper.c
index 47e266d..1a64bd7 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -9969,6 +9969,10 @@
  * Do any appropriate logging, handle PSCI calls, and then hand off
  * to the AArch64-entry or AArch32-entry function depending on the
  * target exception level's register width.
+ *
+ * Note: this is used for both TCG (as the do_interrupt tcg op),
+ *       and KVM to re-inject guest debug exceptions, and to
+ *       inject a Synchronous-External-Abort.
  */
 void arm_cpu_do_interrupt(CPUState *cs)
 {
diff --git a/target/arm/internals.h b/target/arm/internals.h
index 853fa88..448982d 100644
--- a/target/arm/internals.h
+++ b/target/arm/internals.h
@@ -171,6 +171,12 @@
 void arm_cpu_register_gdb_regs_for_features(ARMCPU *cpu);
 void arm_translate_init(void);
 
+#ifdef CONFIG_TCG
+void arm_cpu_synchronize_from_tb(CPUState *cs,
+                                 const struct TranslationBlock *tb);
+#endif /* CONFIG_TCG */
+
+
 enum arm_fprounding {
     FPROUNDING_TIEEVEN,
     FPROUNDING_POSINF,
diff --git a/target/arm/kvm64.c b/target/arm/kvm64.c
index 3c37fc4..dff85f6 100644
--- a/target/arm/kvm64.c
+++ b/target/arm/kvm64.c
@@ -946,7 +946,6 @@
 {
     ARMCPU *cpu = ARM_CPU(c);
     CPUARMState *env = &cpu->env;
-    CPUClass *cc = CPU_GET_CLASS(c);
     uint32_t esr;
     bool same_el;
 
@@ -962,7 +961,7 @@
 
     env->exception.syndrome = esr;
 
-    cc->do_interrupt(c);
+    arm_cpu_do_interrupt(c);
 }
 
 #define AARCH64_CORE_REG(x)   (KVM_REG_ARM64 | KVM_REG_SIZE_U64 | \
@@ -1493,7 +1492,6 @@
 {
     int hsr_ec = syn_get_ec(debug_exit->hsr);
     ARMCPU *cpu = ARM_CPU(cs);
-    CPUClass *cc = CPU_GET_CLASS(cs);
     CPUARMState *env = &cpu->env;
 
     /* Ensure PC is synchronised */
@@ -1547,7 +1545,7 @@
     env->exception.vaddress = debug_exit->far;
     env->exception.target_el = 1;
     qemu_mutex_lock_iothread();
-    cc->do_interrupt(cs);
+    arm_cpu_do_interrupt(cs);
     qemu_mutex_unlock_iothread();
 
     return false;
diff --git a/target/avr/cpu.c b/target/avr/cpu.c
index 6f3d5a9..fa0f8e0 100644
--- a/target/avr/cpu.c
+++ b/target/avr/cpu.c
@@ -184,6 +184,19 @@
     qemu_fprintf(f, "\n");
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps avr_tcg_ops = {
+    .initialize = avr_cpu_tcg_init,
+    .synchronize_from_tb = avr_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = avr_cpu_exec_interrupt,
+    .tlb_fill = avr_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = avr_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void avr_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -198,21 +211,17 @@
     cc->class_by_name = avr_cpu_class_by_name;
 
     cc->has_work = avr_cpu_has_work;
-    cc->do_interrupt = avr_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = avr_cpu_exec_interrupt;
     cc->dump_state = avr_cpu_dump_state;
     cc->set_pc = avr_cpu_set_pc;
     cc->memory_rw_debug = avr_cpu_memory_rw_debug;
     cc->get_phys_page_debug = avr_cpu_get_phys_page_debug;
-    cc->tlb_fill = avr_cpu_tlb_fill;
     cc->vmsd = &vms_avr_cpu;
     cc->disas_set_info = avr_cpu_disas_set_info;
-    cc->tcg_initialize = avr_cpu_tcg_init;
-    cc->synchronize_from_tb = avr_cpu_synchronize_from_tb;
     cc->gdb_read_register = avr_cpu_gdb_read_register;
     cc->gdb_write_register = avr_cpu_gdb_write_register;
     cc->gdb_num_core_regs = 35;
     cc->gdb_core_xml_file = "avr-cpu.xml";
+    cc->tcg_ops = &avr_tcg_ops;
 }
 
 /*
diff --git a/target/avr/helper.c b/target/avr/helper.c
index d96d143..65880b9 100644
--- a/target/avr/helper.c
+++ b/target/avr/helper.c
@@ -20,6 +20,7 @@
 
 #include "qemu/osdep.h"
 #include "cpu.h"
+#include "hw/core/tcg-cpu-ops.h"
 #include "exec/exec-all.h"
 #include "exec/address-spaces.h"
 #include "exec/helper-proto.h"
@@ -34,7 +35,7 @@
     if (interrupt_request & CPU_INTERRUPT_RESET) {
         if (cpu_interrupts_enabled(env)) {
             cs->exception_index = EXCP_RESET;
-            cc->do_interrupt(cs);
+            cc->tcg_ops->do_interrupt(cs);
 
             cs->interrupt_request &= ~CPU_INTERRUPT_RESET;
 
@@ -45,7 +46,7 @@
         if (cpu_interrupts_enabled(env) && env->intsrc != 0) {
             int index = ctz32(env->intsrc);
             cs->exception_index = EXCP_INT(index);
-            cc->do_interrupt(cs);
+            cc->tcg_ops->do_interrupt(cs);
 
             env->intsrc &= env->intsrc - 1; /* clear the interrupt */
             cs->interrupt_request &= ~CPU_INTERRUPT_HARD;
diff --git a/target/cris/cpu.c b/target/cris/cpu.c
index cff6b9e..ed98338 100644
--- a/target/cris/cpu.c
+++ b/target/cris/cpu.c
@@ -193,15 +193,36 @@
 #endif
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps crisv10_tcg_ops = {
+    .initialize = cris_initialize_crisv10_tcg,
+    .cpu_exec_interrupt = cris_cpu_exec_interrupt,
+    .tlb_fill = cris_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = crisv10_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
+static struct TCGCPUOps crisv32_tcg_ops = {
+    .initialize = cris_initialize_tcg,
+    .cpu_exec_interrupt = cris_cpu_exec_interrupt,
+    .tlb_fill = cris_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = cris_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void crisv8_cpu_class_init(ObjectClass *oc, void *data)
 {
     CPUClass *cc = CPU_CLASS(oc);
     CRISCPUClass *ccc = CRIS_CPU_CLASS(oc);
 
     ccc->vr = 8;
-    cc->do_interrupt = crisv10_cpu_do_interrupt;
     cc->gdb_read_register = crisv10_cpu_gdb_read_register;
-    cc->tcg_initialize = cris_initialize_crisv10_tcg;
+    cc->tcg_ops = &crisv10_tcg_ops;
 }
 
 static void crisv9_cpu_class_init(ObjectClass *oc, void *data)
@@ -210,9 +231,8 @@
     CRISCPUClass *ccc = CRIS_CPU_CLASS(oc);
 
     ccc->vr = 9;
-    cc->do_interrupt = crisv10_cpu_do_interrupt;
     cc->gdb_read_register = crisv10_cpu_gdb_read_register;
-    cc->tcg_initialize = cris_initialize_crisv10_tcg;
+    cc->tcg_ops = &crisv10_tcg_ops;
 }
 
 static void crisv10_cpu_class_init(ObjectClass *oc, void *data)
@@ -221,9 +241,8 @@
     CRISCPUClass *ccc = CRIS_CPU_CLASS(oc);
 
     ccc->vr = 10;
-    cc->do_interrupt = crisv10_cpu_do_interrupt;
     cc->gdb_read_register = crisv10_cpu_gdb_read_register;
-    cc->tcg_initialize = cris_initialize_crisv10_tcg;
+    cc->tcg_ops = &crisv10_tcg_ops;
 }
 
 static void crisv11_cpu_class_init(ObjectClass *oc, void *data)
@@ -232,9 +251,8 @@
     CRISCPUClass *ccc = CRIS_CPU_CLASS(oc);
 
     ccc->vr = 11;
-    cc->do_interrupt = crisv10_cpu_do_interrupt;
     cc->gdb_read_register = crisv10_cpu_gdb_read_register;
-    cc->tcg_initialize = cris_initialize_crisv10_tcg;
+    cc->tcg_ops = &crisv10_tcg_ops;
 }
 
 static void crisv17_cpu_class_init(ObjectClass *oc, void *data)
@@ -243,16 +261,17 @@
     CRISCPUClass *ccc = CRIS_CPU_CLASS(oc);
 
     ccc->vr = 17;
-    cc->do_interrupt = crisv10_cpu_do_interrupt;
     cc->gdb_read_register = crisv10_cpu_gdb_read_register;
-    cc->tcg_initialize = cris_initialize_crisv10_tcg;
+    cc->tcg_ops = &crisv10_tcg_ops;
 }
 
 static void crisv32_cpu_class_init(ObjectClass *oc, void *data)
 {
+    CPUClass *cc = CPU_CLASS(oc);
     CRISCPUClass *ccc = CRIS_CPU_CLASS(oc);
 
     ccc->vr = 32;
+    cc->tcg_ops = &crisv32_tcg_ops;
 }
 
 static void cris_cpu_class_init(ObjectClass *oc, void *data)
@@ -268,13 +287,10 @@
 
     cc->class_by_name = cris_cpu_class_by_name;
     cc->has_work = cris_cpu_has_work;
-    cc->do_interrupt = cris_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = cris_cpu_exec_interrupt;
     cc->dump_state = cris_cpu_dump_state;
     cc->set_pc = cris_cpu_set_pc;
     cc->gdb_read_register = cris_cpu_gdb_read_register;
     cc->gdb_write_register = cris_cpu_gdb_write_register;
-    cc->tlb_fill = cris_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
     cc->get_phys_page_debug = cris_cpu_get_phys_page_debug;
     dc->vmsd = &vmstate_cris_cpu;
@@ -284,7 +300,6 @@
     cc->gdb_stop_before_watchpoint = true;
 
     cc->disas_set_info = cris_disas_set_info;
-    cc->tcg_initialize = cris_initialize_tcg;
 }
 
 #define DEFINE_CRIS_CPU_TYPE(cpu_model, initfn) \
diff --git a/target/cris/helper.c b/target/cris/helper.c
index ed45c3d..7e3bb58 100644
--- a/target/cris/helper.c
+++ b/target/cris/helper.c
@@ -20,6 +20,7 @@
 
 #include "qemu/osdep.h"
 #include "cpu.h"
+#include "hw/core/tcg-cpu-ops.h"
 #include "mmu.h"
 #include "qemu/host-utils.h"
 #include "exec/exec-all.h"
@@ -299,7 +300,7 @@
         && (env->pregs[PR_CCS] & I_FLAG)
         && !env->locked_irq) {
         cs->exception_index = EXCP_IRQ;
-        cc->do_interrupt(cs);
+        cc->tcg_ops->do_interrupt(cs);
         ret = true;
     }
     if (interrupt_request & CPU_INTERRUPT_NMI) {
@@ -311,7 +312,7 @@
         }
         if ((env->pregs[PR_CCS] & m_flag_archval)) {
             cs->exception_index = EXCP_NMI;
-            cc->do_interrupt(cs);
+            cc->tcg_ops->do_interrupt(cs);
             ret = true;
         }
     }
diff --git a/target/hppa/cpu.c b/target/hppa/cpu.c
index e28f047..d8fad52 100644
--- a/target/hppa/cpu.c
+++ b/target/hppa/cpu.c
@@ -71,6 +71,7 @@
     info->print_insn = print_insn_hppa;
 }
 
+#ifndef CONFIG_USER_ONLY
 static void hppa_cpu_do_unaligned_access(CPUState *cs, vaddr addr,
                                          MMUAccessType access_type,
                                          int mmu_idx, uintptr_t retaddr)
@@ -87,6 +88,7 @@
 
     cpu_loop_exit_restore(cs, retaddr);
 }
+#endif /* CONFIG_USER_ONLY */
 
 static void hppa_cpu_realizefn(DeviceState *dev, Error **errp)
 {
@@ -129,6 +131,20 @@
     return object_class_by_name(TYPE_HPPA_CPU);
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps hppa_tcg_ops = {
+    .initialize = hppa_translate_init,
+    .synchronize_from_tb = hppa_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = hppa_cpu_exec_interrupt,
+    .tlb_fill = hppa_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = hppa_cpu_do_interrupt,
+    .do_unaligned_access = hppa_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void hppa_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -140,23 +156,17 @@
 
     cc->class_by_name = hppa_cpu_class_by_name;
     cc->has_work = hppa_cpu_has_work;
-    cc->do_interrupt = hppa_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = hppa_cpu_exec_interrupt;
     cc->dump_state = hppa_cpu_dump_state;
     cc->set_pc = hppa_cpu_set_pc;
-    cc->synchronize_from_tb = hppa_cpu_synchronize_from_tb;
     cc->gdb_read_register = hppa_cpu_gdb_read_register;
     cc->gdb_write_register = hppa_cpu_gdb_write_register;
-    cc->tlb_fill = hppa_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
     cc->get_phys_page_debug = hppa_cpu_get_phys_page_debug;
     dc->vmsd = &vmstate_hppa_cpu;
 #endif
-    cc->do_unaligned_access = hppa_cpu_do_unaligned_access;
     cc->disas_set_info = hppa_cpu_disas_set_info;
-    cc->tcg_initialize = hppa_translate_init;
-
     cc->gdb_num_core_regs = 128;
+    cc->tcg_ops = &hppa_tcg_ops;
 }
 
 static const TypeInfo hppa_cpu_type_info = {
diff --git a/target/i386/hax/hax-cpus.c b/target/i386/hax/hax-accel-ops.c
similarity index 69%
rename from target/i386/hax/hax-cpus.c
rename to target/i386/hax/hax-accel-ops.c
index f72c85b..136630e 100644
--- a/target/i386/hax/hax-cpus.c
+++ b/target/i386/hax/hax-accel-ops.c
@@ -26,7 +26,7 @@
 #include "sysemu/cpus.h"
 #include "qemu/guest-random.h"
 
-#include "hax-cpus.h"
+#include "hax-accel-ops.h"
 
 static void *hax_cpu_thread_fn(void *arg)
 {
@@ -74,12 +74,29 @@
 #endif
 }
 
-const CpusAccel hax_cpus = {
-    .create_vcpu_thread = hax_start_vcpu_thread,
-    .kick_vcpu_thread = hax_kick_vcpu_thread,
+static void hax_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
-    .synchronize_post_reset = hax_cpu_synchronize_post_reset,
-    .synchronize_post_init = hax_cpu_synchronize_post_init,
-    .synchronize_state = hax_cpu_synchronize_state,
-    .synchronize_pre_loadvm = hax_cpu_synchronize_pre_loadvm,
+    ops->create_vcpu_thread = hax_start_vcpu_thread;
+    ops->kick_vcpu_thread = hax_kick_vcpu_thread;
+
+    ops->synchronize_post_reset = hax_cpu_synchronize_post_reset;
+    ops->synchronize_post_init = hax_cpu_synchronize_post_init;
+    ops->synchronize_state = hax_cpu_synchronize_state;
+    ops->synchronize_pre_loadvm = hax_cpu_synchronize_pre_loadvm;
+}
+
+static const TypeInfo hax_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("hax"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = hax_accel_ops_class_init,
+    .abstract = true,
 };
+
+static void hax_accel_ops_register_types(void)
+{
+    type_register_static(&hax_accel_ops_type);
+}
+type_init(hax_accel_ops_register_types);
diff --git a/target/i386/hax/hax-cpus.h b/target/i386/hax/hax-accel-ops.h
similarity index 95%
rename from target/i386/hax/hax-cpus.h
rename to target/i386/hax/hax-accel-ops.h
index ee8ab7a..c769851 100644
--- a/target/i386/hax/hax-cpus.h
+++ b/target/i386/hax/hax-accel-ops.h
@@ -12,8 +12,6 @@
 
 #include "sysemu/cpus.h"
 
-extern const CpusAccel hax_cpus;
-
 #include "hax-interface.h"
 #include "hax-i386.h"
 
diff --git a/target/i386/hax/hax-all.c b/target/i386/hax/hax-all.c
index fecfe8c..bf65ed6 100644
--- a/target/i386/hax/hax-all.c
+++ b/target/i386/hax/hax-all.c
@@ -28,12 +28,12 @@
 #include "exec/address-spaces.h"
 
 #include "qemu-common.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/reset.h"
 #include "sysemu/runstate.h"
 #include "hw/boards.h"
 
-#include "hax-cpus.h"
+#include "hax-accel-ops.h"
 
 #define DEBUG_HAX 0
 
@@ -364,9 +364,6 @@
                 !ret ? "working" : "not working",
                 !ret ? "fast virt" : "emulation");
     }
-    if (ret == 0) {
-        cpus_register_accel(&hax_cpus);
-    }
     return ret;
 }
 
diff --git a/target/i386/hax/hax-mem.c b/target/i386/hax/hax-mem.c
index 71e637c..35495f5 100644
--- a/target/i386/hax/hax-mem.c
+++ b/target/i386/hax/hax-mem.c
@@ -13,7 +13,7 @@
 #include "exec/address-spaces.h"
 #include "qemu/error-report.h"
 
-#include "hax-cpus.h"
+#include "hax-accel-ops.h"
 #include "qemu/queue.h"
 
 #define DEBUG_HAX_MEM 0
diff --git a/target/i386/hax/hax-posix.c b/target/i386/hax/hax-posix.c
index 735a749..ac1a510 100644
--- a/target/i386/hax/hax-posix.c
+++ b/target/i386/hax/hax-posix.c
@@ -15,7 +15,7 @@
 #include <sys/ioctl.h>
 
 #include "sysemu/cpus.h"
-#include "hax-cpus.h"
+#include "hax-accel-ops.h"
 
 hax_fd hax_mod_open(void)
 {
diff --git a/target/i386/hax/hax-windows.c b/target/i386/hax/hax-windows.c
index 6c82dfb..59afa21 100644
--- a/target/i386/hax/hax-windows.c
+++ b/target/i386/hax/hax-windows.c
@@ -12,7 +12,7 @@
 
 #include "qemu/osdep.h"
 #include "cpu.h"
-#include "hax-cpus.h"
+#include "hax-accel-ops.h"
 
 /*
  * return 0 when success, -1 when driver not loaded,
diff --git a/target/i386/hax/hax-windows.h b/target/i386/hax/hax-windows.h
index a5ce12d..b1f5d4f3 100644
--- a/target/i386/hax/hax-windows.h
+++ b/target/i386/hax/hax-windows.h
@@ -23,7 +23,7 @@
 #include <winioctl.h>
 #include <windef.h>
 
-#include "hax-cpus.h"
+#include "hax-accel-ops.h"
 
 #define HAX_INVALID_FD INVALID_HANDLE_VALUE
 
diff --git a/target/i386/hax/meson.build b/target/i386/hax/meson.build
index 77ea431..d6c520f 100644
--- a/target/i386/hax/meson.build
+++ b/target/i386/hax/meson.build
@@ -1,7 +1,7 @@
 i386_softmmu_ss.add(when: 'CONFIG_HAX', if_true: files(
   'hax-all.c',
   'hax-mem.c',
-  'hax-cpus.c',
+  'hax-accel-ops.c',
 ))
 i386_softmmu_ss.add(when: ['CONFIG_HAX', 'CONFIG_POSIX'], if_true: files('hax-posix.c'))
 i386_softmmu_ss.add(when: ['CONFIG_HAX', 'CONFIG_WIN32'], if_true: files('hax-windows.c'))
diff --git a/target/i386/hvf/hvf-cpus.c b/target/i386/hvf/hvf-accel-ops.c
similarity index 83%
rename from target/i386/hvf/hvf-cpus.c
rename to target/i386/hvf/hvf-accel-ops.c
index 817b3d7..cbaad23 100644
--- a/target/i386/hvf/hvf-cpus.c
+++ b/target/i386/hvf/hvf-accel-ops.c
@@ -55,7 +55,7 @@
 #include "target/i386/cpu.h"
 #include "qemu/guest-random.h"
 
-#include "hvf-cpus.h"
+#include "hvf-accel-ops.h"
 
 /*
  * The HVF-specific vCPU thread function. This one should only run when the host
@@ -121,11 +121,26 @@
                        cpu, QEMU_THREAD_JOINABLE);
 }
 
-const CpusAccel hvf_cpus = {
-    .create_vcpu_thread = hvf_start_vcpu_thread,
+static void hvf_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
-    .synchronize_post_reset = hvf_cpu_synchronize_post_reset,
-    .synchronize_post_init = hvf_cpu_synchronize_post_init,
-    .synchronize_state = hvf_cpu_synchronize_state,
-    .synchronize_pre_loadvm = hvf_cpu_synchronize_pre_loadvm,
+    ops->create_vcpu_thread = hvf_start_vcpu_thread;
+
+    ops->synchronize_post_reset = hvf_cpu_synchronize_post_reset;
+    ops->synchronize_post_init = hvf_cpu_synchronize_post_init;
+    ops->synchronize_state = hvf_cpu_synchronize_state;
+    ops->synchronize_pre_loadvm = hvf_cpu_synchronize_pre_loadvm;
 };
+static const TypeInfo hvf_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("hvf"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = hvf_accel_ops_class_init,
+    .abstract = true,
+};
+static void hvf_accel_ops_register_types(void)
+{
+    type_register_static(&hvf_accel_ops_type);
+}
+type_init(hvf_accel_ops_register_types);
diff --git a/target/i386/hvf/hvf-cpus.h b/target/i386/hvf/hvf-accel-ops.h
similarity index 94%
rename from target/i386/hvf/hvf-cpus.h
rename to target/i386/hvf/hvf-accel-ops.h
index ced31b8..8f992da 100644
--- a/target/i386/hvf/hvf-cpus.h
+++ b/target/i386/hvf/hvf-accel-ops.h
@@ -12,8 +12,6 @@
 
 #include "sysemu/cpus.h"
 
-extern const CpusAccel hvf_cpus;
-
 int hvf_init_vcpu(CPUState *);
 int hvf_vcpu_exec(CPUState *);
 void hvf_cpu_synchronize_state(CPUState *);
diff --git a/target/i386/hvf/hvf-i386.h b/target/i386/hvf/hvf-i386.h
index e0edffd..50b914f 100644
--- a/target/i386/hvf/hvf-i386.h
+++ b/target/i386/hvf/hvf-i386.h
@@ -16,7 +16,7 @@
 #ifndef HVF_I386_H
 #define HVF_I386_H
 
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/hvf.h"
 #include "cpu.h"
 #include "x86.h"
diff --git a/target/i386/hvf/hvf.c b/target/i386/hvf/hvf.c
index ed93565..5b90dcd 100644
--- a/target/i386/hvf/hvf.c
+++ b/target/i386/hvf/hvf.c
@@ -69,10 +69,10 @@
 #include "exec/address-spaces.h"
 #include "hw/i386/apic_internal.h"
 #include "qemu/main-loop.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "target/i386/cpu.h"
 
-#include "hvf-cpus.h"
+#include "hvf-accel-ops.h"
 
 HVFState *hvf_state;
 
@@ -887,7 +887,6 @@
   
     hvf_state = s;
     memory_listener_register(&hvf_memory_listener, &address_space_memory);
-    cpus_register_accel(&hvf_cpus);
     return 0;
 }
 
diff --git a/target/i386/hvf/meson.build b/target/i386/hvf/meson.build
index 409c9a3..e9eb5a5 100644
--- a/target/i386/hvf/meson.build
+++ b/target/i386/hvf/meson.build
@@ -1,6 +1,6 @@
 i386_softmmu_ss.add(when: [hvf, 'CONFIG_HVF'], if_true: files(
   'hvf.c',
-  'hvf-cpus.c',
+  'hvf-accel-ops.c',
   'x86.c',
   'x86_cpuid.c',
   'x86_decode.c',
diff --git a/target/i386/hvf/x86_task.c b/target/i386/hvf/x86_task.c
index 6f04478..d66dfd7 100644
--- a/target/i386/hvf/x86_task.c
+++ b/target/i386/hvf/x86_task.c
@@ -28,7 +28,7 @@
 
 #include "hw/i386/apic_internal.h"
 #include "qemu/main-loop.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "target/i386/cpu.h"
 
 // TODO: taskswitch handling
diff --git a/target/i386/hvf/x86hvf.c b/target/i386/hvf/x86hvf.c
index bbec412..0d75337 100644
--- a/target/i386/hvf/x86hvf.c
+++ b/target/i386/hvf/x86hvf.c
@@ -32,7 +32,7 @@
 #include <Hypervisor/hv.h>
 #include <Hypervisor/hv_vmx.h>
 
-#include "hvf-cpus.h"
+#include "hvf-accel-ops.h"
 
 void hvf_set_segment(struct CPUState *cpu, struct vmx_segment *vmx_seg,
                      SegmentCache *qseg, bool is_tr)
diff --git a/target/i386/tcg/tcg-cpu.c b/target/i386/tcg/tcg-cpu.c
index 4fa0137..1e125d2 100644
--- a/target/i386/tcg/tcg-cpu.c
+++ b/target/i386/tcg/tcg-cpu.c
@@ -57,16 +57,22 @@
     cpu->env.eip = tb->pc - tb->cs_base;
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps x86_tcg_ops = {
+    .initialize = tcg_x86_init,
+    .synchronize_from_tb = x86_cpu_synchronize_from_tb,
+    .cpu_exec_enter = x86_cpu_exec_enter,
+    .cpu_exec_exit = x86_cpu_exec_exit,
+    .cpu_exec_interrupt = x86_cpu_exec_interrupt,
+    .do_interrupt = x86_cpu_do_interrupt,
+    .tlb_fill = x86_cpu_tlb_fill,
+#ifndef CONFIG_USER_ONLY
+    .debug_excp_handler = breakpoint_handler,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 void tcg_cpu_common_class_init(CPUClass *cc)
 {
-    cc->do_interrupt = x86_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = x86_cpu_exec_interrupt;
-    cc->synchronize_from_tb = x86_cpu_synchronize_from_tb;
-    cc->cpu_exec_enter = x86_cpu_exec_enter;
-    cc->cpu_exec_exit = x86_cpu_exec_exit;
-    cc->tcg_initialize = tcg_x86_init;
-    cc->tlb_fill = x86_cpu_tlb_fill;
-#ifndef CONFIG_USER_ONLY
-    cc->debug_excp_handler = breakpoint_handler;
-#endif
+    cc->tcg_ops = &x86_tcg_ops;
 }
diff --git a/target/i386/whpx/meson.build b/target/i386/whpx/meson.build
index d8aa683..95fc31e 100644
--- a/target/i386/whpx/meson.build
+++ b/target/i386/whpx/meson.build
@@ -1,5 +1,5 @@
 i386_softmmu_ss.add(when: 'CONFIG_WHPX', if_true: files(
   'whpx-all.c',
   'whpx-apic.c',
-  'whpx-cpus.c',
+  'whpx-accel-ops.c',
 ))
diff --git a/target/i386/whpx/whpx-cpus.c b/target/i386/whpx/whpx-accel-ops.c
similarity index 71%
rename from target/i386/whpx/whpx-cpus.c
rename to target/i386/whpx/whpx-accel-ops.c
index f7e6988..6bc47c5 100644
--- a/target/i386/whpx/whpx-cpus.c
+++ b/target/i386/whpx/whpx-accel-ops.c
@@ -16,7 +16,7 @@
 
 #include "sysemu/whpx.h"
 #include "whpx-internal.h"
-#include "whpx-cpus.h"
+#include "whpx-accel-ops.h"
 
 static void *whpx_cpu_thread_fn(void *arg)
 {
@@ -83,12 +83,29 @@
     }
 }
 
-const CpusAccel whpx_cpus = {
-    .create_vcpu_thread = whpx_start_vcpu_thread,
-    .kick_vcpu_thread = whpx_kick_vcpu_thread,
+static void whpx_accel_ops_class_init(ObjectClass *oc, void *data)
+{
+    AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
-    .synchronize_post_reset = whpx_cpu_synchronize_post_reset,
-    .synchronize_post_init = whpx_cpu_synchronize_post_init,
-    .synchronize_state = whpx_cpu_synchronize_state,
-    .synchronize_pre_loadvm = whpx_cpu_synchronize_pre_loadvm,
+    ops->create_vcpu_thread = whpx_start_vcpu_thread;
+    ops->kick_vcpu_thread = whpx_kick_vcpu_thread;
+
+    ops->synchronize_post_reset = whpx_cpu_synchronize_post_reset;
+    ops->synchronize_post_init = whpx_cpu_synchronize_post_init;
+    ops->synchronize_state = whpx_cpu_synchronize_state;
+    ops->synchronize_pre_loadvm = whpx_cpu_synchronize_pre_loadvm;
+}
+
+static const TypeInfo whpx_accel_ops_type = {
+    .name = ACCEL_OPS_NAME("whpx"),
+
+    .parent = TYPE_ACCEL_OPS,
+    .class_init = whpx_accel_ops_class_init,
+    .abstract = true,
 };
+
+static void whpx_accel_ops_register_types(void)
+{
+    type_register_static(&whpx_accel_ops_type);
+}
+type_init(whpx_accel_ops_register_types);
diff --git a/target/i386/whpx/whpx-cpus.h b/target/i386/whpx/whpx-accel-ops.h
similarity index 95%
rename from target/i386/whpx/whpx-cpus.h
rename to target/i386/whpx/whpx-accel-ops.h
index bdb367d..2dee6d6 100644
--- a/target/i386/whpx/whpx-cpus.h
+++ b/target/i386/whpx/whpx-accel-ops.h
@@ -12,8 +12,6 @@
 
 #include "sysemu/cpus.h"
 
-extern const CpusAccel whpx_cpus;
-
 int whpx_init_vcpu(CPUState *cpu);
 int whpx_vcpu_exec(CPUState *cpu);
 void whpx_destroy_vcpu(CPUState *cpu);
diff --git a/target/i386/whpx/whpx-all.c b/target/i386/whpx/whpx-all.c
index 985ceba..f0a35df 100644
--- a/target/i386/whpx/whpx-all.c
+++ b/target/i386/whpx/whpx-all.c
@@ -13,7 +13,7 @@
 #include "exec/address-spaces.h"
 #include "exec/ioport.h"
 #include "qemu-common.h"
-#include "sysemu/accel.h"
+#include "qemu/accel.h"
 #include "sysemu/whpx.h"
 #include "sysemu/cpus.h"
 #include "sysemu/runstate.h"
@@ -28,8 +28,11 @@
 #include "migration/blocker.h"
 #include <winerror.h>
 
-#include "whpx-cpus.h"
 #include "whpx-internal.h"
+#include "whpx-accel-ops.h"
+
+#include <WinHvPlatform.h>
+#include <WinHvEmulation.h>
 
 #define HYPERV_APIC_BUS_FREQUENCY      (200000000ULL)
 
@@ -1846,8 +1849,6 @@
 
     whpx_memory_init();
 
-    cpus_register_accel(&whpx_cpus);
-
     printf("Windows Hypervisor Platform accelerator is operational\n");
     return 0;
 
diff --git a/target/lm32/cpu.c b/target/lm32/cpu.c
index c50ad5f..c23d728 100644
--- a/target/lm32/cpu.c
+++ b/target/lm32/cpu.c
@@ -210,6 +210,19 @@
     return oc;
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps lm32_tcg_ops = {
+    .initialize = lm32_translate_init,
+    .cpu_exec_interrupt = lm32_cpu_exec_interrupt,
+    .tlb_fill = lm32_cpu_tlb_fill,
+    .debug_excp_handler = lm32_debug_excp_handler,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = lm32_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void lm32_cpu_class_init(ObjectClass *oc, void *data)
 {
     LM32CPUClass *lcc = LM32_CPU_CLASS(oc);
@@ -222,22 +235,18 @@
 
     cc->class_by_name = lm32_cpu_class_by_name;
     cc->has_work = lm32_cpu_has_work;
-    cc->do_interrupt = lm32_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = lm32_cpu_exec_interrupt;
     cc->dump_state = lm32_cpu_dump_state;
     cc->set_pc = lm32_cpu_set_pc;
     cc->gdb_read_register = lm32_cpu_gdb_read_register;
     cc->gdb_write_register = lm32_cpu_gdb_write_register;
-    cc->tlb_fill = lm32_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
     cc->get_phys_page_debug = lm32_cpu_get_phys_page_debug;
     cc->vmsd = &vmstate_lm32_cpu;
 #endif
     cc->gdb_num_core_regs = 32 + 7;
     cc->gdb_stop_before_watchpoint = true;
-    cc->debug_excp_handler = lm32_debug_excp_handler;
     cc->disas_set_info = lm32_cpu_disas_set_info;
-    cc->tcg_initialize = lm32_translate_init;
+    cc->tcg_ops = &lm32_tcg_ops;
 }
 
 #define DEFINE_LM32_CPU_TYPE(cpu_model, initfn) \
diff --git a/target/m68k/cpu.c b/target/m68k/cpu.c
index b811a0b..c6fde81 100644
--- a/target/m68k/cpu.c
+++ b/target/m68k/cpu.c
@@ -453,6 +453,19 @@
 };
 #endif
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps m68k_tcg_ops = {
+    .initialize = m68k_tcg_init,
+    .cpu_exec_interrupt = m68k_cpu_exec_interrupt,
+    .tlb_fill = m68k_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = m68k_cpu_do_interrupt,
+    .do_transaction_failed = m68k_cpu_transaction_failed,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void m68k_cpu_class_init(ObjectClass *c, void *data)
 {
     M68kCPUClass *mcc = M68K_CPU_CLASS(c);
@@ -465,22 +478,18 @@
 
     cc->class_by_name = m68k_cpu_class_by_name;
     cc->has_work = m68k_cpu_has_work;
-    cc->do_interrupt = m68k_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = m68k_cpu_exec_interrupt;
     cc->dump_state = m68k_cpu_dump_state;
     cc->set_pc = m68k_cpu_set_pc;
     cc->gdb_read_register = m68k_cpu_gdb_read_register;
     cc->gdb_write_register = m68k_cpu_gdb_write_register;
-    cc->tlb_fill = m68k_cpu_tlb_fill;
 #if defined(CONFIG_SOFTMMU)
-    cc->do_transaction_failed = m68k_cpu_transaction_failed;
     cc->get_phys_page_debug = m68k_cpu_get_phys_page_debug;
     dc->vmsd = &vmstate_m68k_cpu;
 #endif
     cc->disas_set_info = m68k_cpu_disas_set_info;
-    cc->tcg_initialize = m68k_tcg_init;
 
     cc->gdb_num_core_regs = 18;
+    cc->tcg_ops = &m68k_tcg_ops;
 }
 
 static void m68k_cpu_class_init_cf_core(ObjectClass *c, void *data)
diff --git a/target/microblaze/cpu.c b/target/microblaze/cpu.c
index d5e8bfe..433ba20 100644
--- a/target/microblaze/cpu.c
+++ b/target/microblaze/cpu.c
@@ -352,6 +352,21 @@
     return object_class_by_name(TYPE_MICROBLAZE_CPU);
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps mb_tcg_ops = {
+    .initialize = mb_tcg_init,
+    .synchronize_from_tb = mb_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = mb_cpu_exec_interrupt,
+    .tlb_fill = mb_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = mb_cpu_do_interrupt,
+    .do_transaction_failed = mb_cpu_transaction_failed,
+    .do_unaligned_access = mb_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void mb_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -364,17 +379,13 @@
 
     cc->class_by_name = mb_cpu_class_by_name;
     cc->has_work = mb_cpu_has_work;
-    cc->do_interrupt = mb_cpu_do_interrupt;
-    cc->do_unaligned_access = mb_cpu_do_unaligned_access;
-    cc->cpu_exec_interrupt = mb_cpu_exec_interrupt;
+
     cc->dump_state = mb_cpu_dump_state;
     cc->set_pc = mb_cpu_set_pc;
-    cc->synchronize_from_tb = mb_cpu_synchronize_from_tb;
     cc->gdb_read_register = mb_cpu_gdb_read_register;
     cc->gdb_write_register = mb_cpu_gdb_write_register;
-    cc->tlb_fill = mb_cpu_tlb_fill;
+
 #ifndef CONFIG_USER_ONLY
-    cc->do_transaction_failed = mb_cpu_transaction_failed;
     cc->get_phys_page_attrs_debug = mb_cpu_get_phys_page_attrs_debug;
     dc->vmsd = &vmstate_mb_cpu;
 #endif
@@ -382,7 +393,7 @@
     cc->gdb_num_core_regs = 32 + 27;
 
     cc->disas_set_info = mb_disas_set_info;
-    cc->tcg_initialize = mb_tcg_init;
+    cc->tcg_ops = &mb_tcg_ops;
 }
 
 static const TypeInfo mb_cpu_type_info = {
diff --git a/target/mips/cpu.c b/target/mips/cpu.c
index 4c45482..ad163ea 100644
--- a/target/mips/cpu.c
+++ b/target/mips/cpu.c
@@ -257,6 +257,7 @@
     }
 }
 
+#ifdef CONFIG_TCG
 static void mips_cpu_synchronize_from_tb(CPUState *cs,
                                          const TranslationBlock *tb)
 {
@@ -267,6 +268,7 @@
     env->hflags &= ~MIPS_HFLAG_BMASK;
     env->hflags |= tb->flags & MIPS_HFLAG_BMASK;
 }
+#endif /* CONFIG_TCG */
 
 static bool mips_cpu_has_work(CPUState *cs)
 {
@@ -661,6 +663,26 @@
     DEFINE_PROP_END_OF_LIST()
 };
 
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+/*
+ * NB: cannot be const, as some elements are changed for specific
+ * mips hardware (see hw/mips/jazz.c).
+ */
+static struct TCGCPUOps mips_tcg_ops = {
+    .initialize = mips_tcg_init,
+    .synchronize_from_tb = mips_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = mips_cpu_exec_interrupt,
+    .tlb_fill = mips_cpu_tlb_fill,
+
+#if !defined(CONFIG_USER_ONLY)
+    .do_interrupt = mips_cpu_do_interrupt,
+    .do_transaction_failed = mips_cpu_do_transaction_failed,
+    .do_unaligned_access = mips_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+#endif /* CONFIG_TCG */
+
 static void mips_cpu_class_init(ObjectClass *c, void *data)
 {
     MIPSCPUClass *mcc = MIPS_CPU_CLASS(c);
@@ -674,27 +696,20 @@
 
     cc->class_by_name = mips_cpu_class_by_name;
     cc->has_work = mips_cpu_has_work;
-    cc->do_interrupt = mips_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = mips_cpu_exec_interrupt;
     cc->dump_state = mips_cpu_dump_state;
     cc->set_pc = mips_cpu_set_pc;
-    cc->synchronize_from_tb = mips_cpu_synchronize_from_tb;
     cc->gdb_read_register = mips_cpu_gdb_read_register;
     cc->gdb_write_register = mips_cpu_gdb_write_register;
 #ifndef CONFIG_USER_ONLY
-    cc->do_transaction_failed = mips_cpu_do_transaction_failed;
-    cc->do_unaligned_access = mips_cpu_do_unaligned_access;
     cc->get_phys_page_debug = mips_cpu_get_phys_page_debug;
     cc->vmsd = &vmstate_mips_cpu;
 #endif
     cc->disas_set_info = mips_cpu_disas_set_info;
-#ifdef CONFIG_TCG
-    cc->tcg_initialize = mips_tcg_init;
-    cc->tlb_fill = mips_cpu_tlb_fill;
-#endif
-
     cc->gdb_num_core_regs = 73;
     cc->gdb_stop_before_watchpoint = true;
+#ifdef CONFIG_TCG
+    cc->tcg_ops = &mips_tcg_ops;
+#endif /* CONFIG_TCG */
 }
 
 static const TypeInfo mips_cpu_type_info = {
diff --git a/target/moxie/cpu.c b/target/moxie/cpu.c
index 6e0443c..83bec34d 100644
--- a/target/moxie/cpu.c
+++ b/target/moxie/cpu.c
@@ -94,6 +94,17 @@
     return oc;
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps moxie_tcg_ops = {
+    .initialize = moxie_translate_init,
+    .tlb_fill = moxie_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = moxie_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void moxie_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -107,16 +118,14 @@
     cc->class_by_name = moxie_cpu_class_by_name;
 
     cc->has_work = moxie_cpu_has_work;
-    cc->do_interrupt = moxie_cpu_do_interrupt;
     cc->dump_state = moxie_cpu_dump_state;
     cc->set_pc = moxie_cpu_set_pc;
-    cc->tlb_fill = moxie_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
     cc->get_phys_page_debug = moxie_cpu_get_phys_page_debug;
     cc->vmsd = &vmstate_moxie_cpu;
 #endif
     cc->disas_set_info = moxie_cpu_disas_set_info;
-    cc->tcg_initialize = moxie_translate_init;
+    cc->tcg_ops = &moxie_tcg_ops;
 }
 
 static void moxielite_initfn(Object *obj)
diff --git a/target/nios2/cpu.c b/target/nios2/cpu.c
index 58688e1..e9c9fc3 100644
--- a/target/nios2/cpu.c
+++ b/target/nios2/cpu.c
@@ -207,6 +207,18 @@
     DEFINE_PROP_END_OF_LIST(),
 };
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps nios2_tcg_ops = {
+    .initialize = nios2_tcg_init,
+    .cpu_exec_interrupt = nios2_cpu_exec_interrupt,
+    .tlb_fill = nios2_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = nios2_cpu_do_interrupt,
+    .do_unaligned_access = nios2_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
 
 static void nios2_cpu_class_init(ObjectClass *oc, void *data)
 {
@@ -221,20 +233,16 @@
 
     cc->class_by_name = nios2_cpu_class_by_name;
     cc->has_work = nios2_cpu_has_work;
-    cc->do_interrupt = nios2_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = nios2_cpu_exec_interrupt;
     cc->dump_state = nios2_cpu_dump_state;
     cc->set_pc = nios2_cpu_set_pc;
     cc->disas_set_info = nios2_cpu_disas_set_info;
-    cc->tlb_fill = nios2_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
-    cc->do_unaligned_access = nios2_cpu_do_unaligned_access;
     cc->get_phys_page_debug = nios2_cpu_get_phys_page_debug;
 #endif
     cc->gdb_read_register = nios2_cpu_gdb_read_register;
     cc->gdb_write_register = nios2_cpu_gdb_write_register;
     cc->gdb_num_core_regs = 49;
-    cc->tcg_initialize = nios2_tcg_init;
+    cc->tcg_ops = &nios2_tcg_ops;
 }
 
 static const TypeInfo nios2_cpu_type_info = {
diff --git a/target/openrisc/cpu.c b/target/openrisc/cpu.c
index b0bdfbe..2c64842 100644
--- a/target/openrisc/cpu.c
+++ b/target/openrisc/cpu.c
@@ -174,6 +174,18 @@
                       | (IMMUCFGR_NTS & (ctz32(TLB_SIZE) << 2));
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps openrisc_tcg_ops = {
+    .initialize = openrisc_translate_init,
+    .cpu_exec_interrupt = openrisc_cpu_exec_interrupt,
+    .tlb_fill = openrisc_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = openrisc_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void openrisc_cpu_class_init(ObjectClass *oc, void *data)
 {
     OpenRISCCPUClass *occ = OPENRISC_CPU_CLASS(oc);
@@ -186,20 +198,17 @@
 
     cc->class_by_name = openrisc_cpu_class_by_name;
     cc->has_work = openrisc_cpu_has_work;
-    cc->do_interrupt = openrisc_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = openrisc_cpu_exec_interrupt;
     cc->dump_state = openrisc_cpu_dump_state;
     cc->set_pc = openrisc_cpu_set_pc;
     cc->gdb_read_register = openrisc_cpu_gdb_read_register;
     cc->gdb_write_register = openrisc_cpu_gdb_write_register;
-    cc->tlb_fill = openrisc_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
     cc->get_phys_page_debug = openrisc_cpu_get_phys_page_debug;
     dc->vmsd = &vmstate_openrisc_cpu;
 #endif
     cc->gdb_num_core_regs = 32 + 3;
-    cc->tcg_initialize = openrisc_translate_init;
     cc->disas_set_info = openrisc_disas_set_info;
+    cc->tcg_ops = &openrisc_tcg_ops;
 }
 
 /* Sort alphabetically by type name, except for "any". */
diff --git a/target/ppc/translate_init.c.inc b/target/ppc/translate_init.c.inc
index 3c05a17..9867d0a 100644
--- a/target/ppc/translate_init.c.inc
+++ b/target/ppc/translate_init.c.inc
@@ -10700,6 +10700,7 @@
 }
 
 #ifndef CONFIG_USER_ONLY
+
 static bool ppc_cpu_is_big_endian(CPUState *cs)
 {
     PowerPCCPU *cpu = POWERPC_CPU(cs);
@@ -10710,6 +10711,7 @@
     return !msr_le;
 }
 
+#ifdef CONFIG_TCG
 static void ppc_cpu_exec_enter(CPUState *cs)
 {
     PowerPCCPU *cpu = POWERPC_CPU(cs);
@@ -10731,7 +10733,9 @@
         vhc->cpu_exec_exit(cpu->vhyp, cpu);
     }
 }
-#endif
+#endif /* CONFIG_TCG */
+
+#endif /* !CONFIG_USER_ONLY */
 
 static void ppc_cpu_instance_init(Object *obj)
 {
@@ -10823,6 +10827,23 @@
     DEFINE_PROP_END_OF_LIST(),
 };
 
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps ppc_tcg_ops = {
+  .initialize = ppc_translate_init,
+  .cpu_exec_interrupt = ppc_cpu_exec_interrupt,
+  .tlb_fill = ppc_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+  .do_interrupt = ppc_cpu_do_interrupt,
+  .cpu_exec_enter = ppc_cpu_exec_enter,
+  .cpu_exec_exit = ppc_cpu_exec_exit,
+  .do_unaligned_access = ppc_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+#endif /* CONFIG_TCG */
+
 static void ppc_cpu_class_init(ObjectClass *oc, void *data)
 {
     PowerPCCPUClass *pcc = POWERPC_CPU_CLASS(oc);
@@ -10841,14 +10862,11 @@
 
     cc->class_by_name = ppc_cpu_class_by_name;
     cc->has_work = ppc_cpu_has_work;
-    cc->do_interrupt = ppc_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = ppc_cpu_exec_interrupt;
     cc->dump_state = ppc_cpu_dump_state;
     cc->dump_statistics = ppc_cpu_dump_statistics;
     cc->set_pc = ppc_cpu_set_pc;
     cc->gdb_read_register = ppc_cpu_gdb_read_register;
     cc->gdb_write_register = ppc_cpu_gdb_write_register;
-    cc->do_unaligned_access = ppc_cpu_do_unaligned_access;
 #ifndef CONFIG_USER_ONLY
     cc->get_phys_page_debug = ppc_cpu_get_phys_page_debug;
     cc->vmsd = &vmstate_ppc_cpu;
@@ -10877,18 +10895,13 @@
 #ifndef CONFIG_USER_ONLY
     cc->virtio_is_big_endian = ppc_cpu_is_big_endian;
 #endif
-#ifdef CONFIG_TCG
-    cc->tcg_initialize = ppc_translate_init;
-    cc->tlb_fill = ppc_cpu_tlb_fill;
-#endif
-#ifndef CONFIG_USER_ONLY
-    cc->cpu_exec_enter = ppc_cpu_exec_enter;
-    cc->cpu_exec_exit = ppc_cpu_exec_exit;
-#endif
-
     cc->disas_set_info = ppc_disas_set_info;
 
     dc->fw_name = "PowerPC,UNKNOWN";
+
+#ifdef CONFIG_TCG
+    cc->tcg_ops = &ppc_tcg_ops;
+#endif /* CONFIG_TCG */
 }
 
 static const TypeInfo ppc_cpu_type_info = {
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 2778802..16f1a34 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -580,6 +580,21 @@
     return NULL;
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps riscv_tcg_ops = {
+    .initialize = riscv_translate_init,
+    .synchronize_from_tb = riscv_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = riscv_cpu_exec_interrupt,
+    .tlb_fill = riscv_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = riscv_cpu_do_interrupt,
+    .do_transaction_failed = riscv_cpu_do_transaction_failed,
+    .do_unaligned_access = riscv_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void riscv_cpu_class_init(ObjectClass *c, void *data)
 {
     RISCVCPUClass *mcc = RISCV_CPU_CLASS(c);
@@ -593,11 +608,8 @@
 
     cc->class_by_name = riscv_cpu_class_by_name;
     cc->has_work = riscv_cpu_has_work;
-    cc->do_interrupt = riscv_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = riscv_cpu_exec_interrupt;
     cc->dump_state = riscv_cpu_dump_state;
     cc->set_pc = riscv_cpu_set_pc;
-    cc->synchronize_from_tb = riscv_cpu_synchronize_from_tb;
     cc->gdb_read_register = riscv_cpu_gdb_read_register;
     cc->gdb_write_register = riscv_cpu_gdb_write_register;
     cc->gdb_num_core_regs = 33;
@@ -609,18 +621,14 @@
     cc->gdb_stop_before_watchpoint = true;
     cc->disas_set_info = riscv_cpu_disas_set_info;
 #ifndef CONFIG_USER_ONLY
-    cc->do_transaction_failed = riscv_cpu_do_transaction_failed;
-    cc->do_unaligned_access = riscv_cpu_do_unaligned_access;
     cc->get_phys_page_debug = riscv_cpu_get_phys_page_debug;
     /* For now, mark unmigratable: */
     cc->vmsd = &vmstate_riscv_cpu;
 #endif
     cc->gdb_arch_name = riscv_gdb_arch_name;
     cc->gdb_get_dynamic_xml = riscv_gdb_get_dynamic_xml;
-#ifdef CONFIG_TCG
-    cc->tcg_initialize = riscv_translate_init;
-    cc->tlb_fill = riscv_cpu_tlb_fill;
-#endif
+    cc->tcg_ops = &riscv_tcg_ops;
+
     device_class_set_props(dc, riscv_cpu_properties);
 }
 
diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c
index f8350f5..2f43939 100644
--- a/target/riscv/cpu_helper.c
+++ b/target/riscv/cpu_helper.c
@@ -671,7 +671,7 @@
     env->badaddr = addr;
     riscv_raise_exception(env, cs->exception_index, retaddr);
 }
-#endif
+#endif /* !CONFIG_USER_ONLY */
 
 bool riscv_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
                         MMUAccessType access_type, int mmu_idx,
diff --git a/target/rx/cpu.c b/target/rx/cpu.c
index 2bb1414..7ac6618 100644
--- a/target/rx/cpu.c
+++ b/target/rx/cpu.c
@@ -173,6 +173,19 @@
     qdev_init_gpio_in(DEVICE(cpu), rx_cpu_set_irq, 2);
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps rx_tcg_ops = {
+    .initialize = rx_translate_init,
+    .synchronize_from_tb = rx_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = rx_cpu_exec_interrupt,
+    .tlb_fill = rx_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = rx_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void rx_cpu_class_init(ObjectClass *klass, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(klass);
@@ -186,20 +199,17 @@
 
     cc->class_by_name = rx_cpu_class_by_name;
     cc->has_work = rx_cpu_has_work;
-    cc->do_interrupt = rx_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = rx_cpu_exec_interrupt;
     cc->dump_state = rx_cpu_dump_state;
     cc->set_pc = rx_cpu_set_pc;
-    cc->synchronize_from_tb = rx_cpu_synchronize_from_tb;
+
     cc->gdb_read_register = rx_cpu_gdb_read_register;
     cc->gdb_write_register = rx_cpu_gdb_write_register;
     cc->get_phys_page_debug = rx_cpu_get_phys_page_debug;
     cc->disas_set_info = rx_cpu_disas_set_info;
-    cc->tcg_initialize = rx_translate_init;
-    cc->tlb_fill = rx_cpu_tlb_fill;
 
     cc->gdb_num_core_regs = 26;
     cc->gdb_core_xml_file = "rx-core.xml";
+    cc->tcg_ops = &rx_tcg_ops;
 }
 
 static const TypeInfo rx_cpu_info = {
diff --git a/target/s390x/cpu.c b/target/s390x/cpu.c
index 7da70af..d35eb39 100644
--- a/target/s390x/cpu.c
+++ b/target/s390x/cpu.c
@@ -477,6 +477,22 @@
     return s390_cpu_reset(s, S390_CPU_RESET_CLEAR);
 }
 
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps s390_tcg_ops = {
+    .initialize = s390x_translate_init,
+    .tlb_fill = s390_cpu_tlb_fill,
+
+#if !defined(CONFIG_USER_ONLY)
+    .cpu_exec_interrupt = s390_cpu_exec_interrupt,
+    .do_interrupt = s390_cpu_do_interrupt,
+    .debug_excp_handler = s390x_cpu_debug_excp_handler,
+    .do_unaligned_access = s390x_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+#endif /* CONFIG_TCG */
+
 static void s390_cpu_class_init(ObjectClass *oc, void *data)
 {
     S390CPUClass *scc = S390_CPU_CLASS(oc);
@@ -495,9 +511,6 @@
     scc->reset = s390_cpu_reset;
     cc->class_by_name = s390_cpu_class_by_name,
     cc->has_work = s390_cpu_has_work;
-#ifdef CONFIG_TCG
-    cc->do_interrupt = s390_cpu_do_interrupt;
-#endif
     cc->dump_state = s390_cpu_dump_state;
     cc->set_pc = s390_cpu_set_pc;
     cc->gdb_read_register = s390_cpu_gdb_read_register;
@@ -507,23 +520,17 @@
     cc->vmsd = &vmstate_s390_cpu;
     cc->get_crash_info = s390_cpu_get_crash_info;
     cc->write_elf64_note = s390_cpu_write_elf64_note;
-#ifdef CONFIG_TCG
-    cc->cpu_exec_interrupt = s390_cpu_exec_interrupt;
-    cc->debug_excp_handler = s390x_cpu_debug_excp_handler;
-    cc->do_unaligned_access = s390x_cpu_do_unaligned_access;
-#endif
 #endif
     cc->disas_set_info = s390_cpu_disas_set_info;
-#ifdef CONFIG_TCG
-    cc->tcg_initialize = s390x_translate_init;
-    cc->tlb_fill = s390_cpu_tlb_fill;
-#endif
-
     cc->gdb_num_core_regs = S390_NUM_CORE_REGS;
     cc->gdb_core_xml_file = "s390x-core64.xml";
     cc->gdb_arch_name = s390_gdb_arch_name;
 
     s390_cpu_model_class_register_props(oc);
+
+#ifdef CONFIG_TCG
+    cc->tcg_ops = &s390_tcg_ops;
+#endif /* CONFIG_TCG */
 }
 
 static const TypeInfo s390_cpu_type_info = {
diff --git a/target/s390x/excp_helper.c b/target/s390x/excp_helper.c
index 9cf66d3..ce16af3 100644
--- a/target/s390x/excp_helper.c
+++ b/target/s390x/excp_helper.c
@@ -634,4 +634,4 @@
     }
 }
 
-#endif /* CONFIG_USER_ONLY */
+#endif /* !CONFIG_USER_ONLY */
diff --git a/target/sh4/cpu.c b/target/sh4/cpu.c
index 1e0f05a..a78d283 100644
--- a/target/sh4/cpu.c
+++ b/target/sh4/cpu.c
@@ -206,6 +206,20 @@
     .unmigratable = 1,
 };
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps superh_tcg_ops = {
+    .initialize = sh4_translate_init,
+    .synchronize_from_tb = superh_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = superh_cpu_exec_interrupt,
+    .tlb_fill = superh_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = superh_cpu_do_interrupt,
+    .do_unaligned_access = superh_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void superh_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -219,24 +233,19 @@
 
     cc->class_by_name = superh_cpu_class_by_name;
     cc->has_work = superh_cpu_has_work;
-    cc->do_interrupt = superh_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = superh_cpu_exec_interrupt;
     cc->dump_state = superh_cpu_dump_state;
     cc->set_pc = superh_cpu_set_pc;
-    cc->synchronize_from_tb = superh_cpu_synchronize_from_tb;
     cc->gdb_read_register = superh_cpu_gdb_read_register;
     cc->gdb_write_register = superh_cpu_gdb_write_register;
-    cc->tlb_fill = superh_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
-    cc->do_unaligned_access = superh_cpu_do_unaligned_access;
     cc->get_phys_page_debug = superh_cpu_get_phys_page_debug;
 #endif
     cc->disas_set_info = superh_cpu_disas_set_info;
-    cc->tcg_initialize = sh4_translate_init;
 
     cc->gdb_num_core_regs = 59;
 
     dc->vmsd = &vmstate_sh_cpu;
+    cc->tcg_ops = &superh_tcg_ops;
 }
 
 #define DEFINE_SUPERH_CPU_TYPE(type_name, cinit, initfn) \
diff --git a/target/sparc/cpu.c b/target/sparc/cpu.c
index 6f14e37..aece2c7 100644
--- a/target/sparc/cpu.c
+++ b/target/sparc/cpu.c
@@ -848,6 +848,23 @@
     DEFINE_PROP_END_OF_LIST()
 };
 
+#ifdef CONFIG_TCG
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps sparc_tcg_ops = {
+    .initialize = sparc_tcg_init,
+    .synchronize_from_tb = sparc_cpu_synchronize_from_tb,
+    .cpu_exec_interrupt = sparc_cpu_exec_interrupt,
+    .tlb_fill = sparc_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = sparc_cpu_do_interrupt,
+    .do_transaction_failed = sparc_cpu_do_transaction_failed,
+    .do_unaligned_access = sparc_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+#endif /* CONFIG_TCG */
+
 static void sparc_cpu_class_init(ObjectClass *oc, void *data)
 {
     SPARCCPUClass *scc = SPARC_CPU_CLASS(oc);
@@ -863,31 +880,25 @@
     cc->class_by_name = sparc_cpu_class_by_name;
     cc->parse_features = sparc_cpu_parse_features;
     cc->has_work = sparc_cpu_has_work;
-    cc->do_interrupt = sparc_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = sparc_cpu_exec_interrupt;
     cc->dump_state = sparc_cpu_dump_state;
 #if !defined(TARGET_SPARC64) && !defined(CONFIG_USER_ONLY)
     cc->memory_rw_debug = sparc_cpu_memory_rw_debug;
 #endif
     cc->set_pc = sparc_cpu_set_pc;
-    cc->synchronize_from_tb = sparc_cpu_synchronize_from_tb;
     cc->gdb_read_register = sparc_cpu_gdb_read_register;
     cc->gdb_write_register = sparc_cpu_gdb_write_register;
-    cc->tlb_fill = sparc_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
-    cc->do_transaction_failed = sparc_cpu_do_transaction_failed;
-    cc->do_unaligned_access = sparc_cpu_do_unaligned_access;
     cc->get_phys_page_debug = sparc_cpu_get_phys_page_debug;
     cc->vmsd = &vmstate_sparc_cpu;
 #endif
     cc->disas_set_info = cpu_sparc_disas_set_info;
-    cc->tcg_initialize = sparc_tcg_init;
 
 #if defined(TARGET_SPARC64) && !defined(TARGET_ABI32)
     cc->gdb_num_core_regs = 86;
 #else
     cc->gdb_num_core_regs = 72;
 #endif
+    cc->tcg_ops = &sparc_tcg_ops;
 }
 
 static const TypeInfo sparc_cpu_type_info = {
diff --git a/target/tilegx/cpu.c b/target/tilegx/cpu.c
index 1fee87c..d969c2f 100644
--- a/target/tilegx/cpu.c
+++ b/target/tilegx/cpu.c
@@ -134,6 +134,18 @@
     return false;
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps tilegx_tcg_ops = {
+    .initialize = tilegx_tcg_init,
+    .cpu_exec_interrupt = tilegx_cpu_exec_interrupt,
+    .tlb_fill = tilegx_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = tilegx_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void tilegx_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -147,13 +159,10 @@
 
     cc->class_by_name = tilegx_cpu_class_by_name;
     cc->has_work = tilegx_cpu_has_work;
-    cc->do_interrupt = tilegx_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = tilegx_cpu_exec_interrupt;
     cc->dump_state = tilegx_cpu_dump_state;
     cc->set_pc = tilegx_cpu_set_pc;
-    cc->tlb_fill = tilegx_cpu_tlb_fill;
     cc->gdb_num_core_regs = 0;
-    cc->tcg_initialize = tilegx_tcg_init;
+    cc->tcg_ops = &tilegx_tcg_ops;
 }
 
 static const TypeInfo tilegx_cpu_type_info = {
diff --git a/target/tricore/cpu.c b/target/tricore/cpu.c
index 4bff1d4..0b1e139 100644
--- a/target/tricore/cpu.c
+++ b/target/tricore/cpu.c
@@ -142,6 +142,14 @@
     set_feature(&cpu->env, TRICORE_FEATURE_161);
 }
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps tricore_tcg_ops = {
+    .initialize = tricore_tcg_init,
+    .synchronize_from_tb = tricore_cpu_synchronize_from_tb,
+    .tlb_fill = tricore_cpu_tlb_fill,
+};
+
 static void tricore_cpu_class_init(ObjectClass *c, void *data)
 {
     TriCoreCPUClass *mcc = TRICORE_CPU_CLASS(c);
@@ -162,10 +170,8 @@
 
     cc->dump_state = tricore_cpu_dump_state;
     cc->set_pc = tricore_cpu_set_pc;
-    cc->synchronize_from_tb = tricore_cpu_synchronize_from_tb;
     cc->get_phys_page_debug = tricore_cpu_get_phys_page_debug;
-    cc->tcg_initialize = tricore_tcg_init;
-    cc->tlb_fill = tricore_cpu_tlb_fill;
+    cc->tcg_ops = &tricore_tcg_ops;
 }
 
 #define DEFINE_TRICORE_CPU_TYPE(cpu_model, initfn) \
diff --git a/target/unicore32/cpu.c b/target/unicore32/cpu.c
index b27fb96..0258884 100644
--- a/target/unicore32/cpu.c
+++ b/target/unicore32/cpu.c
@@ -120,6 +120,18 @@
     .unmigratable = 1,
 };
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps uc32_tcg_ops = {
+    .initialize = uc32_translate_init,
+    .cpu_exec_interrupt = uc32_cpu_exec_interrupt,
+    .tlb_fill = uc32_cpu_tlb_fill,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = uc32_cpu_do_interrupt,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void uc32_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -131,14 +143,11 @@
 
     cc->class_by_name = uc32_cpu_class_by_name;
     cc->has_work = uc32_cpu_has_work;
-    cc->do_interrupt = uc32_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = uc32_cpu_exec_interrupt;
     cc->dump_state = uc32_cpu_dump_state;
     cc->set_pc = uc32_cpu_set_pc;
-    cc->tlb_fill = uc32_cpu_tlb_fill;
     cc->get_phys_page_debug = uc32_cpu_get_phys_page_debug;
-    cc->tcg_initialize = uc32_translate_init;
     dc->vmsd = &vmstate_uc32_cpu;
+    cc->tcg_ops = &uc32_tcg_ops;
 }
 
 #define DEFINE_UNICORE32_CPU_TYPE(cpu_model, initfn) \
diff --git a/target/xtensa/cpu.c b/target/xtensa/cpu.c
index 88a3226..e2b2c7a 100644
--- a/target/xtensa/cpu.c
+++ b/target/xtensa/cpu.c
@@ -181,6 +181,21 @@
     .unmigratable = 1,
 };
 
+#include "hw/core/tcg-cpu-ops.h"
+
+static struct TCGCPUOps xtensa_tcg_ops = {
+    .initialize = xtensa_translate_init,
+    .cpu_exec_interrupt = xtensa_cpu_exec_interrupt,
+    .tlb_fill = xtensa_cpu_tlb_fill,
+    .debug_excp_handler = xtensa_breakpoint_handler,
+
+#ifndef CONFIG_USER_ONLY
+    .do_interrupt = xtensa_cpu_do_interrupt,
+    .do_transaction_failed = xtensa_cpu_do_transaction_failed,
+    .do_unaligned_access = xtensa_cpu_do_unaligned_access,
+#endif /* !CONFIG_USER_ONLY */
+};
+
 static void xtensa_cpu_class_init(ObjectClass *oc, void *data)
 {
     DeviceClass *dc = DEVICE_CLASS(oc);
@@ -194,23 +209,17 @@
 
     cc->class_by_name = xtensa_cpu_class_by_name;
     cc->has_work = xtensa_cpu_has_work;
-    cc->do_interrupt = xtensa_cpu_do_interrupt;
-    cc->cpu_exec_interrupt = xtensa_cpu_exec_interrupt;
     cc->dump_state = xtensa_cpu_dump_state;
     cc->set_pc = xtensa_cpu_set_pc;
     cc->gdb_read_register = xtensa_cpu_gdb_read_register;
     cc->gdb_write_register = xtensa_cpu_gdb_write_register;
     cc->gdb_stop_before_watchpoint = true;
-    cc->tlb_fill = xtensa_cpu_tlb_fill;
 #ifndef CONFIG_USER_ONLY
-    cc->do_unaligned_access = xtensa_cpu_do_unaligned_access;
     cc->get_phys_page_debug = xtensa_cpu_get_phys_page_debug;
-    cc->do_transaction_failed = xtensa_cpu_do_transaction_failed;
 #endif
-    cc->debug_excp_handler = xtensa_breakpoint_handler;
     cc->disas_set_info = xtensa_cpu_disas_set_info;
-    cc->tcg_initialize = xtensa_translate_init;
     dc->vmsd = &vmstate_xtensa_cpu;
+    cc->tcg_ops = &xtensa_tcg_ops;
 }
 
 static const TypeInfo xtensa_cpu_type_info = {
diff --git a/target/xtensa/helper.c b/target/xtensa/helper.c
index 05e2b7f..eeffee2 100644
--- a/target/xtensa/helper.c
+++ b/target/xtensa/helper.c
@@ -261,7 +261,7 @@
     cpu_loop_exit_restore(cs, retaddr);
 }
 
-#else
+#else /* !CONFIG_USER_ONLY */
 
 void xtensa_cpu_do_unaligned_access(CPUState *cs,
                                     vaddr addr, MMUAccessType access_type,
@@ -337,4 +337,4 @@
         qemu_cpu_kick(cpu);
     }
 }
-#endif
+#endif /* !CONFIG_USER_ONLY */
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 3c1ee39..1376cdc 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -2488,7 +2488,7 @@
     v0 = temp_tcgv_vec(arg_temp(a0));
     v1 = temp_tcgv_vec(arg_temp(va_arg(va, TCGArg)));
     a2 = va_arg(va, TCGArg);
-    v2 = temp_tcgv_vec(arg_temp(a2));
+    va_end(va);
 
     switch (opc) {
     case INDEX_op_rotli_vec:
@@ -2502,6 +2502,7 @@
     case INDEX_op_shrv_vec:
     case INDEX_op_sarv_vec:
         /* Right shifts are negative left shifts for AArch64.  */
+        v2 = temp_tcgv_vec(arg_temp(a2));
         t1 = tcg_temp_new_vec(type);
         tcg_gen_neg_vec(vece, t1, v2);
         opc = (opc == INDEX_op_shrv_vec
@@ -2512,6 +2513,7 @@
         break;
 
     case INDEX_op_rotlv_vec:
+        v2 = temp_tcgv_vec(arg_temp(a2));
         t1 = tcg_temp_new_vec(type);
         c1 = tcg_constant_vec(type, vece, 8 << vece);
         tcg_gen_sub_vec(vece, t1, v2, c1);
@@ -2525,6 +2527,7 @@
         break;
 
     case INDEX_op_rotrv_vec:
+        v2 = temp_tcgv_vec(arg_temp(a2));
         t1 = tcg_temp_new_vec(type);
         t2 = tcg_temp_new_vec(type);
         c1 = tcg_constant_vec(type, vece, 8 << vece);
@@ -2543,8 +2546,6 @@
     default:
         g_assert_not_reached();
     }
-
-    va_end(va);
 }
 
 static TCGConstraintSetIndex tcg_target_op_def(TCGOpcode op)
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index b674701..695d7ee 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -1067,7 +1067,7 @@
                 op = (is_unsigned ? RIL_CLFI : RIL_CFI);
                 tcg_out_insn_RIL(s, op, r1, c2);
                 goto exit;
-            } else if (c2 == (is_unsigned ? (uint32_t)c2 : (int32_t)c2)) {
+            } else if (c2 == (is_unsigned ? (TCGArg)(uint32_t)c2 : (TCGArg)(int32_t)c2)) {
                 op = (is_unsigned ? RIL_CLGFI : RIL_CGFI);
                 tcg_out_insn_RIL(s, op, r1, c2);
                 goto exit;
diff --git a/tcg/tcg-common.c b/tcg/tcg-common.c
index 7e1992e..aa0c4f6 100644
--- a/tcg/tcg-common.c
+++ b/tcg/tcg-common.c
@@ -25,10 +25,6 @@
 #include "qemu/osdep.h"
 #include "tcg/tcg.h"
 
-#if defined(CONFIG_TCG_INTERPRETER)
-uintptr_t tci_tb_ptr;
-#endif
-
 TCGOpDef tcg_op_defs[] = {
 #define DEF(s, oargs, iargs, cargs, flags) \
          { #s, oargs, iargs, cargs, iargs + oargs + cargs, flags },
diff --git a/tcg/tci.c b/tcg/tci.c
index 3fc82d3..fb3c97a 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -33,14 +33,6 @@
 #include "tcg/tcg-op.h"
 #include "qemu/compiler.h"
 
-/* Marker for missing code. */
-#define TODO() \
-    do { \
-        fprintf(stderr, "TODO %s:%u: %s()\n", \
-                __FILE__, __LINE__, __func__); \
-        tcg_abort(); \
-    } while (0)
-
 #if MAX_OPC_PARAM_IARGS != 6
 # error Fix needed, number of supported input arguments changed!
 #endif
@@ -57,6 +49,8 @@
                                     tcg_target_ulong, tcg_target_ulong);
 #endif
 
+__thread uintptr_t tci_tb_ptr;
+
 static tcg_target_ulong tci_read_reg(const tcg_target_ulong *regs, TCGReg index)
 {
     tci_assert(index < TCG_TARGET_NB_REGS);
@@ -115,33 +109,6 @@
     regs[index] = value;
 }
 
-#if TCG_TARGET_REG_BITS == 64
-static void
-tci_write_reg32s(tcg_target_ulong *regs, TCGReg index, int32_t value)
-{
-    tci_write_reg(regs, index, value);
-}
-#endif
-
-static void tci_write_reg8(tcg_target_ulong *regs, TCGReg index, uint8_t value)
-{
-    tci_write_reg(regs, index, value);
-}
-
-#if TCG_TARGET_REG_BITS == 64
-static void
-tci_write_reg16(tcg_target_ulong *regs, TCGReg index, uint16_t value)
-{
-    tci_write_reg(regs, index, value);
-}
-#endif
-
-static void
-tci_write_reg32(tcg_target_ulong *regs, TCGReg index, uint32_t value)
-{
-    tci_write_reg(regs, index, value);
-}
-
 #if TCG_TARGET_REG_BITS == 32
 static void tci_write_reg64(tcg_target_ulong *regs, uint32_t high_index,
                             uint32_t low_index, uint64_t value)
@@ -149,12 +116,6 @@
     tci_write_reg(regs, low_index, value);
     tci_write_reg(regs, high_index, value >> 32);
 }
-#elif TCG_TARGET_REG_BITS == 64
-static void
-tci_write_reg64(tcg_target_ulong *regs, TCGReg index, uint64_t value)
-{
-    tci_write_reg(regs, index, value);
-}
 #endif
 
 #if TCG_TARGET_REG_BITS == 32
@@ -294,61 +255,6 @@
     return taddr;
 }
 
-/* Read indexed register or constant (native size) from bytecode. */
-static tcg_target_ulong
-tci_read_ri(const tcg_target_ulong *regs, const uint8_t **tb_ptr)
-{
-    tcg_target_ulong value;
-    TCGReg r = **tb_ptr;
-    *tb_ptr += 1;
-    if (r == TCG_CONST) {
-        value = tci_read_i(tb_ptr);
-    } else {
-        value = tci_read_reg(regs, r);
-    }
-    return value;
-}
-
-/* Read indexed register or constant (32 bit) from bytecode. */
-static uint32_t tci_read_ri32(const tcg_target_ulong *regs,
-                              const uint8_t **tb_ptr)
-{
-    uint32_t value;
-    TCGReg r = **tb_ptr;
-    *tb_ptr += 1;
-    if (r == TCG_CONST) {
-        value = tci_read_i32(tb_ptr);
-    } else {
-        value = tci_read_reg32(regs, r);
-    }
-    return value;
-}
-
-#if TCG_TARGET_REG_BITS == 32
-/* Read two indexed registers or constants (2 * 32 bit) from bytecode. */
-static uint64_t tci_read_ri64(const tcg_target_ulong *regs,
-                              const uint8_t **tb_ptr)
-{
-    uint32_t low = tci_read_ri32(regs, tb_ptr);
-    return tci_uint64(tci_read_ri32(regs, tb_ptr), low);
-}
-#elif TCG_TARGET_REG_BITS == 64
-/* Read indexed register or constant (64 bit) from bytecode. */
-static uint64_t tci_read_ri64(const tcg_target_ulong *regs,
-                              const uint8_t **tb_ptr)
-{
-    uint64_t value;
-    TCGReg r = **tb_ptr;
-    *tb_ptr += 1;
-    if (r == TCG_CONST) {
-        value = tci_read_i64(tb_ptr);
-    } else {
-        value = tci_read_reg64(regs, r);
-    }
-    return value;
-}
-#endif
-
 static tcg_target_ulong tci_read_label(const uint8_t **tb_ptr)
 {
     tcg_target_ulong label = tci_read_i(tb_ptr);
@@ -393,7 +299,7 @@
         result = (u0 > u1);
         break;
     default:
-        TODO();
+        g_assert_not_reached();
     }
     return result;
 }
@@ -435,7 +341,7 @@
         result = (u0 > u1);
         break;
     default:
-        TODO();
+        g_assert_not_reached();
     }
     return result;
 }
@@ -486,6 +392,18 @@
 # define qemu_st_beq(X)  stq_be_p(g2h(taddr), X)
 #endif
 
+#if TCG_TARGET_REG_BITS == 64
+# define CASE_32_64(x) \
+        case glue(glue(INDEX_op_, x), _i64): \
+        case glue(glue(INDEX_op_, x), _i32):
+# define CASE_64(x) \
+        case glue(glue(INDEX_op_, x), _i64):
+#else
+# define CASE_32_64(x) \
+        case glue(glue(INDEX_op_, x), _i32):
+# define CASE_64(x)
+#endif
+
 /* Interpret pseudo code in tb. */
 /*
  * Disable CFI checks.
@@ -526,29 +444,26 @@
 #endif
         TCGMemOpIdx oi;
 
-#if defined(GETPC)
-        tci_tb_ptr = (uintptr_t)tb_ptr;
-#endif
-
         /* Skip opcode and size entry. */
         tb_ptr += 2;
 
         switch (opc) {
         case INDEX_op_call:
-            t0 = tci_read_ri(regs, &tb_ptr);
+            t0 = tci_read_i(&tb_ptr);
+            tci_tb_ptr = (uintptr_t)tb_ptr;
 #if TCG_TARGET_REG_BITS == 32
             tmp64 = ((helper_function)t0)(tci_read_reg(regs, TCG_REG_R0),
                                           tci_read_reg(regs, TCG_REG_R1),
                                           tci_read_reg(regs, TCG_REG_R2),
                                           tci_read_reg(regs, TCG_REG_R3),
+                                          tci_read_reg(regs, TCG_REG_R4),
                                           tci_read_reg(regs, TCG_REG_R5),
                                           tci_read_reg(regs, TCG_REG_R6),
                                           tci_read_reg(regs, TCG_REG_R7),
                                           tci_read_reg(regs, TCG_REG_R8),
                                           tci_read_reg(regs, TCG_REG_R9),
                                           tci_read_reg(regs, TCG_REG_R10),
-                                          tci_read_reg(regs, TCG_REG_R11),
-                                          tci_read_reg(regs, TCG_REG_R12));
+                                          tci_read_reg(regs, TCG_REG_R11));
             tci_write_reg(regs, TCG_REG_R0, tmp64);
             tci_write_reg(regs, TCG_REG_R1, tmp64 >> 32);
 #else
@@ -556,8 +471,8 @@
                                           tci_read_reg(regs, TCG_REG_R1),
                                           tci_read_reg(regs, TCG_REG_R2),
                                           tci_read_reg(regs, TCG_REG_R3),
-                                          tci_read_reg(regs, TCG_REG_R5),
-                                          tci_read_reg(regs, TCG_REG_R6));
+                                          tci_read_reg(regs, TCG_REG_R4),
+                                          tci_read_reg(regs, TCG_REG_R5));
             tci_write_reg(regs, TCG_REG_R0, tmp64);
 #endif
             break;
@@ -569,78 +484,88 @@
         case INDEX_op_setcond_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
             condition = *tb_ptr++;
-            tci_write_reg32(regs, t0, tci_compare32(t1, t2, condition));
+            tci_write_reg(regs, t0, tci_compare32(t1, t2, condition));
             break;
 #if TCG_TARGET_REG_BITS == 32
         case INDEX_op_setcond2_i32:
             t0 = *tb_ptr++;
             tmp64 = tci_read_r64(regs, &tb_ptr);
-            v64 = tci_read_ri64(regs, &tb_ptr);
+            v64 = tci_read_r64(regs, &tb_ptr);
             condition = *tb_ptr++;
-            tci_write_reg32(regs, t0, tci_compare64(tmp64, v64, condition));
+            tci_write_reg(regs, t0, tci_compare64(tmp64, v64, condition));
             break;
 #elif TCG_TARGET_REG_BITS == 64
         case INDEX_op_setcond_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
             condition = *tb_ptr++;
-            tci_write_reg64(regs, t0, tci_compare64(t1, t2, condition));
+            tci_write_reg(regs, t0, tci_compare64(t1, t2, condition));
             break;
 #endif
         case INDEX_op_mov_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
         case INDEX_op_tci_movi_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_i32(&tb_ptr);
-            tci_write_reg32(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 
             /* Load/store operations (32 bit). */
 
-        case INDEX_op_ld8u_i32:
+        CASE_32_64(ld8u)
             t0 = *tb_ptr++;
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg8(regs, t0, *(uint8_t *)(t1 + t2));
+            tci_write_reg(regs, t0, *(uint8_t *)(t1 + t2));
             break;
-        case INDEX_op_ld8s_i32:
-            TODO();
+        CASE_32_64(ld8s)
+            t0 = *tb_ptr++;
+            t1 = tci_read_r(regs, &tb_ptr);
+            t2 = tci_read_s32(&tb_ptr);
+            tci_write_reg(regs, t0, *(int8_t *)(t1 + t2));
             break;
-        case INDEX_op_ld16u_i32:
-            TODO();
+        CASE_32_64(ld16u)
+            t0 = *tb_ptr++;
+            t1 = tci_read_r(regs, &tb_ptr);
+            t2 = tci_read_s32(&tb_ptr);
+            tci_write_reg(regs, t0, *(uint16_t *)(t1 + t2));
             break;
-        case INDEX_op_ld16s_i32:
-            TODO();
+        CASE_32_64(ld16s)
+            t0 = *tb_ptr++;
+            t1 = tci_read_r(regs, &tb_ptr);
+            t2 = tci_read_s32(&tb_ptr);
+            tci_write_reg(regs, t0, *(int16_t *)(t1 + t2));
             break;
         case INDEX_op_ld_i32:
+        CASE_64(ld32u)
             t0 = *tb_ptr++;
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg32(regs, t0, *(uint32_t *)(t1 + t2));
+            tci_write_reg(regs, t0, *(uint32_t *)(t1 + t2));
             break;
-        case INDEX_op_st8_i32:
+        CASE_32_64(st8)
             t0 = tci_read_r8(regs, &tb_ptr);
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
             *(uint8_t *)(t1 + t2) = t0;
             break;
-        case INDEX_op_st16_i32:
+        CASE_32_64(st16)
             t0 = tci_read_r16(regs, &tb_ptr);
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
             *(uint16_t *)(t1 + t2) = t0;
             break;
         case INDEX_op_st_i32:
+        CASE_64(st32)
             t0 = tci_read_r32(regs, &tb_ptr);
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
-            tci_assert(t1 != sp_value || (int32_t)t2 < 0);
             *(uint32_t *)(t1 + t2) = t0;
             break;
 
@@ -648,104 +573,97 @@
 
         case INDEX_op_add_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 + t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 + t2);
             break;
         case INDEX_op_sub_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 - t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 - t2);
             break;
         case INDEX_op_mul_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 * t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 * t2);
             break;
-#if TCG_TARGET_HAS_div_i32
         case INDEX_op_div_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, (int32_t)t1 / (int32_t)t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, (int32_t)t1 / (int32_t)t2);
             break;
         case INDEX_op_divu_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 / t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 / t2);
             break;
         case INDEX_op_rem_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, (int32_t)t1 % (int32_t)t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, (int32_t)t1 % (int32_t)t2);
             break;
         case INDEX_op_remu_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 % t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 % t2);
             break;
-#elif TCG_TARGET_HAS_div2_i32
-        case INDEX_op_div2_i32:
-        case INDEX_op_divu2_i32:
-            TODO();
-            break;
-#endif
         case INDEX_op_and_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 & t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 & t2);
             break;
         case INDEX_op_or_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 | t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 | t2);
             break;
         case INDEX_op_xor_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 ^ t2);
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 ^ t2);
             break;
 
             /* Shift/rotate operations (32 bit). */
 
         case INDEX_op_shl_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 << (t2 & 31));
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 << (t2 & 31));
             break;
         case INDEX_op_shr_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1 >> (t2 & 31));
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 >> (t2 & 31));
             break;
         case INDEX_op_sar_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, ((int32_t)t1 >> (t2 & 31)));
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, ((int32_t)t1 >> (t2 & 31)));
             break;
 #if TCG_TARGET_HAS_rot_i32
         case INDEX_op_rotl_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, rol32(t1, t2 & 31));
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, rol32(t1, t2 & 31));
             break;
         case INDEX_op_rotr_i32:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri32(regs, &tb_ptr);
-            t2 = tci_read_ri32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, ror32(t1, t2 & 31));
+            t1 = tci_read_r32(regs, &tb_ptr);
+            t2 = tci_read_r32(regs, &tb_ptr);
+            tci_write_reg(regs, t0, ror32(t1, t2 & 31));
             break;
 #endif
 #if TCG_TARGET_HAS_deposit_i32
@@ -756,12 +674,12 @@
             tmp16 = *tb_ptr++;
             tmp8 = *tb_ptr++;
             tmp32 = (((1 << tmp8) - 1) << tmp16);
-            tci_write_reg32(regs, t0, (t1 & ~tmp32) | ((t2 << tmp16) & tmp32));
+            tci_write_reg(regs, t0, (t1 & ~tmp32) | ((t2 << tmp16) & tmp32));
             break;
 #endif
         case INDEX_op_brcond_i32:
             t0 = tci_read_r32(regs, &tb_ptr);
-            t1 = tci_read_ri32(regs, &tb_ptr);
+            t1 = tci_read_r32(regs, &tb_ptr);
             condition = *tb_ptr++;
             label = tci_read_label(&tb_ptr);
             if (tci_compare32(t0, t1, condition)) {
@@ -787,7 +705,7 @@
             break;
         case INDEX_op_brcond2_i32:
             tmp64 = tci_read_r64(regs, &tb_ptr);
-            v64 = tci_read_ri64(regs, &tb_ptr);
+            v64 = tci_read_r64(regs, &tb_ptr);
             condition = *tb_ptr++;
             label = tci_read_label(&tb_ptr);
             if (tci_compare64(tmp64, v64, condition)) {
@@ -808,131 +726,88 @@
         case INDEX_op_ext8s_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r8s(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext16s_i32
         case INDEX_op_ext16s_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r16s(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext8u_i32
         case INDEX_op_ext8u_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r8(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext16u_i32
         case INDEX_op_ext16u_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r16(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_bswap16_i32
         case INDEX_op_bswap16_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r16(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, bswap16(t1));
+            tci_write_reg(regs, t0, bswap16(t1));
             break;
 #endif
 #if TCG_TARGET_HAS_bswap32_i32
         case INDEX_op_bswap32_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, bswap32(t1));
+            tci_write_reg(regs, t0, bswap32(t1));
             break;
 #endif
 #if TCG_TARGET_HAS_not_i32
         case INDEX_op_not_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, ~t1);
+            tci_write_reg(regs, t0, ~t1);
             break;
 #endif
 #if TCG_TARGET_HAS_neg_i32
         case INDEX_op_neg_i32:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            tci_write_reg32(regs, t0, -t1);
+            tci_write_reg(regs, t0, -t1);
             break;
 #endif
 #if TCG_TARGET_REG_BITS == 64
         case INDEX_op_mov_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
         case INDEX_op_tci_movi_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_i64(&tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 
             /* Load/store operations (64 bit). */
 
-        case INDEX_op_ld8u_i64:
-            t0 = *tb_ptr++;
-            t1 = tci_read_r(regs, &tb_ptr);
-            t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg8(regs, t0, *(uint8_t *)(t1 + t2));
-            break;
-        case INDEX_op_ld8s_i64:
-            TODO();
-            break;
-        case INDEX_op_ld16u_i64:
-            t0 = *tb_ptr++;
-            t1 = tci_read_r(regs, &tb_ptr);
-            t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg16(regs, t0, *(uint16_t *)(t1 + t2));
-            break;
-        case INDEX_op_ld16s_i64:
-            TODO();
-            break;
-        case INDEX_op_ld32u_i64:
-            t0 = *tb_ptr++;
-            t1 = tci_read_r(regs, &tb_ptr);
-            t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg32(regs, t0, *(uint32_t *)(t1 + t2));
-            break;
         case INDEX_op_ld32s_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg32s(regs, t0, *(int32_t *)(t1 + t2));
+            tci_write_reg(regs, t0, *(int32_t *)(t1 + t2));
             break;
         case INDEX_op_ld_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
-            tci_write_reg64(regs, t0, *(uint64_t *)(t1 + t2));
-            break;
-        case INDEX_op_st8_i64:
-            t0 = tci_read_r8(regs, &tb_ptr);
-            t1 = tci_read_r(regs, &tb_ptr);
-            t2 = tci_read_s32(&tb_ptr);
-            *(uint8_t *)(t1 + t2) = t0;
-            break;
-        case INDEX_op_st16_i64:
-            t0 = tci_read_r16(regs, &tb_ptr);
-            t1 = tci_read_r(regs, &tb_ptr);
-            t2 = tci_read_s32(&tb_ptr);
-            *(uint16_t *)(t1 + t2) = t0;
-            break;
-        case INDEX_op_st32_i64:
-            t0 = tci_read_r32(regs, &tb_ptr);
-            t1 = tci_read_r(regs, &tb_ptr);
-            t2 = tci_read_s32(&tb_ptr);
-            *(uint32_t *)(t1 + t2) = t0;
+            tci_write_reg(regs, t0, *(uint64_t *)(t1 + t2));
             break;
         case INDEX_op_st_i64:
             t0 = tci_read_r64(regs, &tb_ptr);
             t1 = tci_read_r(regs, &tb_ptr);
             t2 = tci_read_s32(&tb_ptr);
-            tci_assert(t1 != sp_value || (int32_t)t2 < 0);
             *(uint64_t *)(t1 + t2) = t0;
             break;
 
@@ -940,86 +815,97 @@
 
         case INDEX_op_add_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 + t2);
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 + t2);
             break;
         case INDEX_op_sub_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 - t2);
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 - t2);
             break;
         case INDEX_op_mul_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 * t2);
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 * t2);
             break;
-#if TCG_TARGET_HAS_div_i64
         case INDEX_op_div_i64:
+            t0 = *tb_ptr++;
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, (int64_t)t1 / (int64_t)t2);
+            break;
         case INDEX_op_divu_i64:
+            t0 = *tb_ptr++;
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, (uint64_t)t1 / (uint64_t)t2);
+            break;
         case INDEX_op_rem_i64:
+            t0 = *tb_ptr++;
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, (int64_t)t1 % (int64_t)t2);
+            break;
         case INDEX_op_remu_i64:
-            TODO();
+            t0 = *tb_ptr++;
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, (uint64_t)t1 % (uint64_t)t2);
             break;
-#elif TCG_TARGET_HAS_div2_i64
-        case INDEX_op_div2_i64:
-        case INDEX_op_divu2_i64:
-            TODO();
-            break;
-#endif
         case INDEX_op_and_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 & t2);
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 & t2);
             break;
         case INDEX_op_or_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 | t2);
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 | t2);
             break;
         case INDEX_op_xor_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 ^ t2);
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 ^ t2);
             break;
 
             /* Shift/rotate operations (64 bit). */
 
         case INDEX_op_shl_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 << (t2 & 63));
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 << (t2 & 63));
             break;
         case INDEX_op_shr_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1 >> (t2 & 63));
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, t1 >> (t2 & 63));
             break;
         case INDEX_op_sar_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, ((int64_t)t1 >> (t2 & 63)));
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, ((int64_t)t1 >> (t2 & 63)));
             break;
 #if TCG_TARGET_HAS_rot_i64
         case INDEX_op_rotl_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, rol64(t1, t2 & 63));
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, rol64(t1, t2 & 63));
             break;
         case INDEX_op_rotr_i64:
             t0 = *tb_ptr++;
-            t1 = tci_read_ri64(regs, &tb_ptr);
-            t2 = tci_read_ri64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, ror64(t1, t2 & 63));
+            t1 = tci_read_r64(regs, &tb_ptr);
+            t2 = tci_read_r64(regs, &tb_ptr);
+            tci_write_reg(regs, t0, ror64(t1, t2 & 63));
             break;
 #endif
 #if TCG_TARGET_HAS_deposit_i64
@@ -1030,12 +916,12 @@
             tmp16 = *tb_ptr++;
             tmp8 = *tb_ptr++;
             tmp64 = (((1ULL << tmp8) - 1) << tmp16);
-            tci_write_reg64(regs, t0, (t1 & ~tmp64) | ((t2 << tmp16) & tmp64));
+            tci_write_reg(regs, t0, (t1 & ~tmp64) | ((t2 << tmp16) & tmp64));
             break;
 #endif
         case INDEX_op_brcond_i64:
             t0 = tci_read_r64(regs, &tb_ptr);
-            t1 = tci_read_ri64(regs, &tb_ptr);
+            t1 = tci_read_r64(regs, &tb_ptr);
             condition = *tb_ptr++;
             label = tci_read_label(&tb_ptr);
             if (tci_compare64(t0, t1, condition)) {
@@ -1048,28 +934,28 @@
         case INDEX_op_ext8u_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r8(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext8s_i64
         case INDEX_op_ext8s_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r8s(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext16s_i64
         case INDEX_op_ext16s_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r16s(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext16u_i64
         case INDEX_op_ext16u_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r16(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #endif
 #if TCG_TARGET_HAS_ext32s_i64
@@ -1078,7 +964,7 @@
         case INDEX_op_ext_i32_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r32s(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #if TCG_TARGET_HAS_ext32u_i64
         case INDEX_op_ext32u_i64:
@@ -1086,41 +972,41 @@
         case INDEX_op_extu_i32_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, t1);
+            tci_write_reg(regs, t0, t1);
             break;
 #if TCG_TARGET_HAS_bswap16_i64
         case INDEX_op_bswap16_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r16(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, bswap16(t1));
+            tci_write_reg(regs, t0, bswap16(t1));
             break;
 #endif
 #if TCG_TARGET_HAS_bswap32_i64
         case INDEX_op_bswap32_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r32(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, bswap32(t1));
+            tci_write_reg(regs, t0, bswap32(t1));
             break;
 #endif
 #if TCG_TARGET_HAS_bswap64_i64
         case INDEX_op_bswap64_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, bswap64(t1));
+            tci_write_reg(regs, t0, bswap64(t1));
             break;
 #endif
 #if TCG_TARGET_HAS_not_i64
         case INDEX_op_not_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, ~t1);
+            tci_write_reg(regs, t0, ~t1);
             break;
 #endif
 #if TCG_TARGET_HAS_neg_i64
         case INDEX_op_neg_i64:
             t0 = *tb_ptr++;
             t1 = tci_read_r64(regs, &tb_ptr);
-            tci_write_reg64(regs, t0, -t1);
+            tci_write_reg(regs, t0, -t1);
             break;
 #endif
 #endif /* TCG_TARGET_REG_BITS == 64 */
@@ -1169,7 +1055,7 @@
                 tmp32 = qemu_ld_beul;
                 break;
             default:
-                tcg_abort();
+                g_assert_not_reached();
             }
             tci_write_reg(regs, t0, tmp32);
             break;
@@ -1218,7 +1104,7 @@
                 tmp64 = qemu_ld_beq;
                 break;
             default:
-                tcg_abort();
+                g_assert_not_reached();
             }
             tci_write_reg(regs, t0, tmp64);
             if (TCG_TARGET_REG_BITS == 32) {
@@ -1246,7 +1132,7 @@
                 qemu_st_bel(t0);
                 break;
             default:
-                tcg_abort();
+                g_assert_not_reached();
             }
             break;
         case INDEX_op_qemu_st_i64:
@@ -1276,7 +1162,7 @@
                 qemu_st_beq(tmp64);
                 break;
             default:
-                tcg_abort();
+                g_assert_not_reached();
             }
             break;
         case INDEX_op_mb:
@@ -1284,8 +1170,7 @@
             smp_mb();
             break;
         default:
-            TODO();
-            break;
+            g_assert_not_reached();
         }
         tci_assert(tb_ptr == old_code_ptr + op_size);
     }
diff --git a/tcg/tci/tcg-target-con-set.h b/tcg/tci/tcg-target-con-set.h
index 38e82f7..f51b7bc 100644
--- a/tcg/tci/tcg-target-con-set.h
+++ b/tcg/tci/tcg-target-con-set.h
@@ -10,16 +10,12 @@
  * tcg-target-con-str.h; the constraint combination is inclusive or.
  */
 C_O0_I2(r, r)
-C_O0_I2(r, ri)
 C_O0_I3(r, r, r)
-C_O0_I4(r, r, ri, ri)
 C_O0_I4(r, r, r, r)
 C_O1_I1(r, r)
 C_O1_I2(r, 0, r)
-C_O1_I2(r, ri, ri)
 C_O1_I2(r, r, r)
-C_O1_I2(r, r, ri)
-C_O1_I4(r, r, r, ri, ri)
+C_O1_I4(r, r, r, r, r)
 C_O2_I1(r, r, r)
 C_O2_I2(r, r, r, r)
 C_O2_I4(r, r, r, r, r, r)
diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc
index f0f6b13..feac465 100644
--- a/tcg/tci/tcg-target.c.inc
+++ b/tcg/tci/tcg-target.c.inc
@@ -92,8 +92,6 @@
     case INDEX_op_rem_i64:
     case INDEX_op_remu_i32:
     case INDEX_op_remu_i64:
-        return C_O1_I2(r, r, r);
-
     case INDEX_op_add_i32:
     case INDEX_op_add_i64:
     case INDEX_op_sub_i32:
@@ -126,8 +124,9 @@
     case INDEX_op_rotl_i64:
     case INDEX_op_rotr_i32:
     case INDEX_op_rotr_i64:
-        /* TODO: Does R, RI, RI result in faster code than R, R, RI? */
-        return C_O1_I2(r, ri, ri);
+    case INDEX_op_setcond_i32:
+    case INDEX_op_setcond_i64:
+        return C_O1_I2(r, r, r);
 
     case INDEX_op_deposit_i32:
     case INDEX_op_deposit_i64:
@@ -135,11 +134,7 @@
 
     case INDEX_op_brcond_i32:
     case INDEX_op_brcond_i64:
-        return C_O0_I2(r, ri);
-
-    case INDEX_op_setcond_i32:
-    case INDEX_op_setcond_i64:
-        return C_O1_I2(r, r, ri);
+        return C_O0_I2(r, r);
 
 #if TCG_TARGET_REG_BITS == 32
     /* TODO: Support R, R, R, R, RI, RI? Will it be faster? */
@@ -147,11 +142,11 @@
     case INDEX_op_sub2_i32:
         return C_O2_I4(r, r, r, r, r, r);
     case INDEX_op_brcond2_i32:
-        return C_O0_I4(r, r, ri, ri);
+        return C_O0_I4(r, r, r, r);
     case INDEX_op_mulu2_i32:
         return C_O2_I2(r, r, r, r);
     case INDEX_op_setcond2_i32:
-        return C_O1_I4(r, r, r, ri, ri);
+        return C_O1_I4(r, r, r, r, r);
 #endif
 
     case INDEX_op_qemu_ld_i32:
@@ -181,13 +176,10 @@
     TCG_REG_R1,
     TCG_REG_R2,
     TCG_REG_R3,
-#if 0 /* used for TCG_REG_CALL_STACK */
     TCG_REG_R4,
-#endif
     TCG_REG_R5,
     TCG_REG_R6,
     TCG_REG_R7,
-#if TCG_TARGET_NB_REGS >= 16
     TCG_REG_R8,
     TCG_REG_R9,
     TCG_REG_R10,
@@ -196,7 +188,6 @@
     TCG_REG_R13,
     TCG_REG_R14,
     TCG_REG_R15,
-#endif
 };
 
 #if MAX_OPC_PARAM_IARGS != 6
@@ -208,23 +199,16 @@
     TCG_REG_R1,
     TCG_REG_R2,
     TCG_REG_R3,
-#if 0 /* used for TCG_REG_CALL_STACK */
     TCG_REG_R4,
-#endif
     TCG_REG_R5,
-    TCG_REG_R6,
 #if TCG_TARGET_REG_BITS == 32
     /* 32 bit hosts need 2 * MAX_OPC_PARAM_IARGS registers. */
+    TCG_REG_R6,
     TCG_REG_R7,
-#if TCG_TARGET_NB_REGS >= 16
     TCG_REG_R8,
     TCG_REG_R9,
     TCG_REG_R10,
     TCG_REG_R11,
-    TCG_REG_R12,
-#else
-# error Too few input registers available
-#endif
 #endif
 };
 
@@ -245,7 +229,6 @@
     "r05",
     "r06",
     "r07",
-#if TCG_TARGET_NB_REGS >= 16
     "r08",
     "r09",
     "r10",
@@ -254,25 +237,6 @@
     "r13",
     "r14",
     "r15",
-#if TCG_TARGET_NB_REGS >= 32
-    "r16",
-    "r17",
-    "r18",
-    "r19",
-    "r20",
-    "r21",
-    "r22",
-    "r23",
-    "r24",
-    "r25",
-    "r26",
-    "r27",
-    "r28",
-    "r29",
-    "r30",
-    "r31"
-#endif
-#endif
 };
 #endif
 
@@ -325,44 +289,6 @@
     tcg_out8(s, t0);
 }
 
-/* Write register or constant (native size). */
-static void tcg_out_ri(TCGContext *s, int const_arg, TCGArg arg)
-{
-    if (const_arg) {
-        tcg_debug_assert(const_arg == 1);
-        tcg_out8(s, TCG_CONST);
-        tcg_out_i(s, arg);
-    } else {
-        tcg_out_r(s, arg);
-    }
-}
-
-/* Write register or constant (32 bit). */
-static void tcg_out_ri32(TCGContext *s, int const_arg, TCGArg arg)
-{
-    if (const_arg) {
-        tcg_debug_assert(const_arg == 1);
-        tcg_out8(s, TCG_CONST);
-        tcg_out32(s, arg);
-    } else {
-        tcg_out_r(s, arg);
-    }
-}
-
-#if TCG_TARGET_REG_BITS == 64
-/* Write register or constant (64 bit). */
-static void tcg_out_ri64(TCGContext *s, int const_arg, TCGArg arg)
-{
-    if (const_arg) {
-        tcg_debug_assert(const_arg == 1);
-        tcg_out8(s, TCG_CONST);
-        tcg_out64(s, arg);
-    } else {
-        tcg_out_r(s, arg);
-    }
-}
-#endif
-
 /* Write label. */
 static void tci_out_label(TCGContext *s, TCGLabel *label)
 {
@@ -375,10 +301,20 @@
     }
 }
 
+static void stack_bounds_check(TCGReg base, target_long offset)
+{
+    if (base == TCG_REG_CALL_STACK) {
+        tcg_debug_assert(offset < 0);
+        tcg_debug_assert(offset >= -(CPU_TEMP_BUF_NLONGS * sizeof(long)));
+    }
+}
+
 static void tcg_out_ld(TCGContext *s, TCGType type, TCGReg ret, TCGReg arg1,
                        intptr_t arg2)
 {
     uint8_t *old_code_ptr = s->code_ptr;
+
+    stack_bounds_check(arg1, arg2);
     if (type == TCG_TYPE_I32) {
         tcg_out_op_t(s, INDEX_op_ld_i32);
         tcg_out_r(s, ret);
@@ -440,7 +376,7 @@
 {
     uint8_t *old_code_ptr = s->code_ptr;
     tcg_out_op_t(s, INDEX_op_call);
-    tcg_out_ri(s, 1, (uintptr_t)arg);
+    tcg_out_i(s, (uintptr_t)arg);
     old_code_ptr[1] = s->code_ptr - old_code_ptr;
 }
 
@@ -474,7 +410,7 @@
     case INDEX_op_setcond_i32:
         tcg_out_r(s, args[0]);
         tcg_out_r(s, args[1]);
-        tcg_out_ri32(s, const_args[2], args[2]);
+        tcg_out_r(s, args[2]);
         tcg_out8(s, args[3]);   /* condition */
         break;
 #if TCG_TARGET_REG_BITS == 32
@@ -483,15 +419,15 @@
         tcg_out_r(s, args[0]);
         tcg_out_r(s, args[1]);
         tcg_out_r(s, args[2]);
-        tcg_out_ri32(s, const_args[3], args[3]);
-        tcg_out_ri32(s, const_args[4], args[4]);
+        tcg_out_r(s, args[3]);
+        tcg_out_r(s, args[4]);
         tcg_out8(s, args[5]);   /* condition */
         break;
 #elif TCG_TARGET_REG_BITS == 64
     case INDEX_op_setcond_i64:
         tcg_out_r(s, args[0]);
         tcg_out_r(s, args[1]);
-        tcg_out_ri64(s, const_args[2], args[2]);
+        tcg_out_r(s, args[2]);
         tcg_out8(s, args[3]);   /* condition */
         break;
 #endif
@@ -514,6 +450,7 @@
     case INDEX_op_st16_i64:
     case INDEX_op_st32_i64:
     case INDEX_op_st_i64:
+        stack_bounds_check(args[1], args[2]);
         tcg_out_r(s, args[0]);
         tcg_out_r(s, args[1]);
         tcg_debug_assert(args[2] == (int32_t)args[2]);
@@ -536,8 +473,8 @@
     case INDEX_op_rotl_i32:     /* Optional (TCG_TARGET_HAS_rot_i32). */
     case INDEX_op_rotr_i32:     /* Optional (TCG_TARGET_HAS_rot_i32). */
         tcg_out_r(s, args[0]);
-        tcg_out_ri32(s, const_args[1], args[1]);
-        tcg_out_ri32(s, const_args[2], args[2]);
+        tcg_out_r(s, args[1]);
+        tcg_out_r(s, args[2]);
         break;
     case INDEX_op_deposit_i32:  /* Optional (TCG_TARGET_HAS_deposit_i32). */
         tcg_out_r(s, args[0]);
@@ -566,9 +503,13 @@
     case INDEX_op_sar_i64:
     case INDEX_op_rotl_i64:     /* Optional (TCG_TARGET_HAS_rot_i64). */
     case INDEX_op_rotr_i64:     /* Optional (TCG_TARGET_HAS_rot_i64). */
+    case INDEX_op_div_i64:      /* Optional (TCG_TARGET_HAS_div_i64). */
+    case INDEX_op_divu_i64:     /* Optional (TCG_TARGET_HAS_div_i64). */
+    case INDEX_op_rem_i64:      /* Optional (TCG_TARGET_HAS_div_i64). */
+    case INDEX_op_remu_i64:     /* Optional (TCG_TARGET_HAS_div_i64). */
         tcg_out_r(s, args[0]);
-        tcg_out_ri64(s, const_args[1], args[1]);
-        tcg_out_ri64(s, const_args[2], args[2]);
+        tcg_out_r(s, args[1]);
+        tcg_out_r(s, args[2]);
         break;
     case INDEX_op_deposit_i64:  /* Optional (TCG_TARGET_HAS_deposit_i64). */
         tcg_out_r(s, args[0]);
@@ -579,19 +520,9 @@
         tcg_debug_assert(args[4] <= UINT8_MAX);
         tcg_out8(s, args[4]);
         break;
-    case INDEX_op_div_i64:      /* Optional (TCG_TARGET_HAS_div_i64). */
-    case INDEX_op_divu_i64:     /* Optional (TCG_TARGET_HAS_div_i64). */
-    case INDEX_op_rem_i64:      /* Optional (TCG_TARGET_HAS_div_i64). */
-    case INDEX_op_remu_i64:     /* Optional (TCG_TARGET_HAS_div_i64). */
-        TODO();
-        break;
-    case INDEX_op_div2_i64:     /* Optional (TCG_TARGET_HAS_div2_i64). */
-    case INDEX_op_divu2_i64:    /* Optional (TCG_TARGET_HAS_div2_i64). */
-        TODO();
-        break;
     case INDEX_op_brcond_i64:
         tcg_out_r(s, args[0]);
-        tcg_out_ri64(s, const_args[1], args[1]);
+        tcg_out_r(s, args[1]);
         tcg_out8(s, args[2]);           /* condition */
         tci_out_label(s, arg_label(args[3]));
         break;
@@ -625,12 +556,8 @@
     case INDEX_op_rem_i32:      /* Optional (TCG_TARGET_HAS_div_i32). */
     case INDEX_op_remu_i32:     /* Optional (TCG_TARGET_HAS_div_i32). */
         tcg_out_r(s, args[0]);
-        tcg_out_ri32(s, const_args[1], args[1]);
-        tcg_out_ri32(s, const_args[2], args[2]);
-        break;
-    case INDEX_op_div2_i32:     /* Optional (TCG_TARGET_HAS_div2_i32). */
-    case INDEX_op_divu2_i32:    /* Optional (TCG_TARGET_HAS_div2_i32). */
-        TODO();
+        tcg_out_r(s, args[1]);
+        tcg_out_r(s, args[2]);
         break;
 #if TCG_TARGET_REG_BITS == 32
     case INDEX_op_add2_i32:
@@ -645,8 +572,8 @@
     case INDEX_op_brcond2_i32:
         tcg_out_r(s, args[0]);
         tcg_out_r(s, args[1]);
-        tcg_out_ri32(s, const_args[2], args[2]);
-        tcg_out_ri32(s, const_args[3], args[3]);
+        tcg_out_r(s, args[2]);
+        tcg_out_r(s, args[3]);
         tcg_out8(s, args[4]);           /* condition */
         tci_out_label(s, arg_label(args[5]));
         break;
@@ -659,7 +586,7 @@
 #endif
     case INDEX_op_brcond_i32:
         tcg_out_r(s, args[0]);
-        tcg_out_ri32(s, const_args[1], args[1]);
+        tcg_out_r(s, args[1]);
         tcg_out8(s, args[2]);           /* condition */
         tci_out_label(s, arg_label(args[3]));
         break;
@@ -716,6 +643,8 @@
                        intptr_t arg2)
 {
     uint8_t *old_code_ptr = s->code_ptr;
+
+    stack_bounds_check(arg1, arg2);
     if (type == TCG_TYPE_I32) {
         tcg_out_op_t(s, INDEX_op_st_i32);
         tcg_out_r(s, arg);
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index bb784e0..9c0021a 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -100,8 +100,8 @@
 #define TCG_TARGET_HAS_extract_i64      0
 #define TCG_TARGET_HAS_sextract_i64     0
 #define TCG_TARGET_HAS_extract2_i64     0
-#define TCG_TARGET_HAS_div_i64          0
-#define TCG_TARGET_HAS_rem_i64          0
+#define TCG_TARGET_HAS_div_i64          1
+#define TCG_TARGET_HAS_rem_i64          1
 #define TCG_TARGET_HAS_ext8s_i64        1
 #define TCG_TARGET_HAS_ext16s_i64       1
 #define TCG_TARGET_HAS_ext32s_i64       1
@@ -133,11 +133,8 @@
 #define TCG_TARGET_HAS_mulu2_i32        1
 #endif /* TCG_TARGET_REG_BITS == 64 */
 
-/* Number of registers available.
-   For 32 bit hosts, we need more than 8 registers (call arguments). */
-/* #define TCG_TARGET_NB_REGS 8 */
+/* Number of registers available. */
 #define TCG_TARGET_NB_REGS 16
-/* #define TCG_TARGET_NB_REGS 32 */
 
 /* List of registers which are used by TCG. */
 typedef enum {
@@ -149,7 +146,6 @@
     TCG_REG_R5,
     TCG_REG_R6,
     TCG_REG_R7,
-#if TCG_TARGET_NB_REGS >= 16
     TCG_REG_R8,
     TCG_REG_R9,
     TCG_REG_R10,
@@ -158,33 +154,12 @@
     TCG_REG_R13,
     TCG_REG_R14,
     TCG_REG_R15,
-#if TCG_TARGET_NB_REGS >= 32
-    TCG_REG_R16,
-    TCG_REG_R17,
-    TCG_REG_R18,
-    TCG_REG_R19,
-    TCG_REG_R20,
-    TCG_REG_R21,
-    TCG_REG_R22,
-    TCG_REG_R23,
-    TCG_REG_R24,
-    TCG_REG_R25,
-    TCG_REG_R26,
-    TCG_REG_R27,
-    TCG_REG_R28,
-    TCG_REG_R29,
-    TCG_REG_R30,
-    TCG_REG_R31,
-#endif
-#endif
-    /* Special value UINT8_MAX is used by TCI to encode constant values. */
-    TCG_CONST = UINT8_MAX
+
+    TCG_AREG0 = TCG_REG_R14,
+    TCG_REG_CALL_STACK = TCG_REG_R15,
 } TCGReg;
 
-#define TCG_AREG0                       (TCG_TARGET_NB_REGS - 2)
-
 /* Used for function call generation. */
-#define TCG_REG_CALL_STACK              (TCG_TARGET_NB_REGS - 1)
 #define TCG_TARGET_CALL_STACK_OFFSET    0
 #define TCG_TARGET_STACK_ALIGN          16