Merge remote-tracking branch 'remotes/philmd/tags/abstract-arch-cpu-20220307' into staging

- Re-org accel/ and softmmu/ to have more target-agnostic objects.

- Use CPUArchState as an abstract type, defined by each target
  (CPUState is our interface with generic code, CPUArchState is
   our interface with target-specific code).

# gpg: Signature made Sun 06 Mar 2022 23:23:19 GMT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.org>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* remotes/philmd/tags/abstract-arch-cpu-20220307: (33 commits)
  accel/tcg: Remove pointless CPUArchState casts
  target/i386: Remove pointless CPUArchState casts
  target: Use ArchCPU as interface to target CPU
  target: Introduce and use OBJECT_DECLARE_CPU_TYPE() macro
  target: Use CPUArchState as interface to target-specific CPU state
  target: Use forward declared type instead of structure type
  target/hexagon: Add missing 'hw/core/cpu.h' include
  target: Include missing 'cpu.h'
  Hexagon (target/hexagon) convert to OBJECT_DECLARE_TYPE
  target/i386/tcg/sysemu: Include missing 'exec/exec-all.h' header
  cpu: Add missing 'exec/exec-all.h' and 'qemu/accel.h' headers
  exec/cpu_ldst: Include 'cpu.h' to get target_ulong definition
  meson: Display libfdt as disabled when system emulation is disabled
  softmmu: Build target-agnostic objects once
  softmmu: Add qemu_init_arch_modules()
  exec/cpu: Make address_space_init/reloading_memory_map target agnostic
  exec/gdbstub: Make gdb_exit() / gdb_set_stop_cpu() target agnostic
  misc: Add missing "sysemu/cpu-timers.h" include
  misc: Remove unnecessary "sysemu/cpu-timers.h" include
  softmmu/cpu-timers: Remove unused 'exec/exec-all.h' header
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

87 files changed, 371 insertions, 402 deletions

diff --git a/accel/kvm/kvm-accel-ops.c b/accel/kvm/kvm-accel-ops.c
index 7516c67a3f..c4244a23c6 100644
--- a/accel/kvm/kvm-accel-ops.c
+++ b/accel/kvm/kvm-accel-ops.c
@@ -74,11 +74,23 @@ static void kvm_start_vcpu_thread(CPUState *cpu)
                        cpu, QEMU_THREAD_JOINABLE);
 }
 
+static bool kvm_vcpu_thread_is_idle(CPUState *cpu)
+{
+    return !kvm_halt_in_kernel();
+}
+
+static bool kvm_cpus_are_resettable(void)
+{
+    return !kvm_enabled() || kvm_cpu_check_are_resettable();
+}
+
 static void kvm_accel_ops_class_init(ObjectClass *oc, void *data)
 {
     AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
     ops->create_vcpu_thread = kvm_start_vcpu_thread;
+    ops->cpu_thread_is_idle = kvm_vcpu_thread_is_idle;
+    ops->cpus_are_resettable = kvm_cpus_are_resettable;
     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;
diff --git a/accel/meson.build b/accel/meson.build
index dfd808d2c8..b9a963cf80 100644
--- a/accel/meson.build
+++ b/accel/meson.build
@@ -2,12 +2,14 @@ specific_ss.add(files('accel-common.c'))
 softmmu_ss.add(files('accel-softmmu.c'))
 user_ss.add(files('accel-user.c'))
 
-subdir('hvf')
-subdir('qtest')
-subdir('kvm')
 subdir('tcg')
-subdir('xen')
-subdir('stubs')
+if have_system
+  subdir('hvf')
+  subdir('qtest')
+  subdir('kvm')
+  subdir('xen')
+  subdir('stubs')
+endif
 
 dummy_ss = ss.source_set()
 dummy_ss.add(files(
diff --git a/accel/qtest/qtest.c b/accel/qtest/qtest.c
index 7e6b8110d5..f6056ac836 100644
--- a/accel/qtest/qtest.c
+++ b/accel/qtest/qtest.c
@@ -20,7 +20,6 @@
 #include "qemu/accel.h"
 #include "sysemu/qtest.h"
 #include "sysemu/cpus.h"
-#include "sysemu/cpu-timers.h"
 #include "qemu/guest-random.h"
 #include "qemu/main-loop.h"
 #include "hw/core/cpu.h"
diff --git a/accel/stubs/hax-stub.c b/accel/stubs/hax-stub.c
index 49077f88e3..2fe31aaa9a 100644
--- a/accel/stubs/hax-stub.c
+++ b/accel/stubs/hax-stub.c
@@ -16,6 +16,8 @@
 #include "qemu/osdep.h"
 #include "sysemu/hax.h"
 
+bool hax_allowed;
+
 int hax_sync_vcpus(void)
 {
     return 0;
diff --git a/accel/stubs/kvm-stub.c b/accel/stubs/kvm-stub.c
index 5319573e00..7e0fb884b9 100644
--- a/accel/stubs/kvm-stub.c
+++ b/accel/stubs/kvm-stub.c
@@ -12,10 +12,7 @@
 
 #include "qemu/osdep.h"
 #include "sysemu/kvm.h"
-
-#ifndef CONFIG_USER_ONLY
 #include "hw/pci/msi.h"
-#endif
 
 KVMState *kvm_state;
 bool kvm_kernel_irqchip;
@@ -80,7 +77,6 @@ int kvm_on_sigbus(int code, void *addr)
     return 1;
 }
 
-#ifndef CONFIG_USER_ONLY
 int kvm_irqchip_add_msi_route(KVMState *s, int vector, PCIDevice *dev)
 {
     return -ENOSYS;
@@ -152,4 +148,3 @@ bool kvm_dirty_ring_enabled(void)
 {
     return false;
 }
-#endif
diff --git a/accel/stubs/meson.build b/accel/stubs/meson.build
index 12dd1539af..0249b9258f 100644
--- a/accel/stubs/meson.build
+++ b/accel/stubs/meson.build
@@ -1,4 +1,7 @@
-specific_ss.add(when: 'CONFIG_HAX', if_false: files('hax-stub.c'))
-specific_ss.add(when: 'CONFIG_XEN', if_false: files('xen-stub.c'))
-specific_ss.add(when: 'CONFIG_KVM', if_false: files('kvm-stub.c'))
-specific_ss.add(when: 'CONFIG_TCG', if_false: files('tcg-stub.c'))
+sysemu_stubs_ss = ss.source_set()
+sysemu_stubs_ss.add(when: 'CONFIG_HAX', if_false: files('hax-stub.c'))
+sysemu_stubs_ss.add(when: 'CONFIG_XEN', if_false: files('xen-stub.c'))
+sysemu_stubs_ss.add(when: 'CONFIG_KVM', if_false: files('kvm-stub.c'))
+sysemu_stubs_ss.add(when: 'CONFIG_TCG', if_false: files('tcg-stub.c'))
+
+specific_ss.add_all(when: ['CONFIG_SOFTMMU'], if_true: sysemu_stubs_ss)
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index c68270f794..c997c2e8e0 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -422,7 +422,7 @@ static void cpu_exec_exit(CPUState *cpu)
 
 void cpu_exec_step_atomic(CPUState *cpu)
 {
-    CPUArchState *env = (CPUArchState *)cpu->env_ptr;
+    CPUArchState *env = cpu->env_ptr;
     TranslationBlock *tb;
     target_ulong cs_base, pc;
     uint32_t flags, cflags;
@@ -532,7 +532,7 @@ TranslationBlock *tb_htable_lookup(CPUState *cpu, target_ulong pc,
     struct tb_desc desc;
     uint32_t h;
 
-    desc.env = (CPUArchState *)cpu->env_ptr;
+    desc.env = cpu->env_ptr;
     desc.cs_base = cs_base;
     desc.flags = flags;
     desc.cflags = cflags;
diff --git a/accel/tcg/tcg-accel-ops-icount.c b/accel/tcg/tcg-accel-ops-icount.c
index ea42d1d51b..bdaf2c943b 100644
--- a/accel/tcg/tcg-accel-ops-icount.c
+++ b/accel/tcg/tcg-accel-ops-icount.c
@@ -27,6 +27,7 @@
 #include "qemu-common.h"
 #include "sysemu/tcg.h"
 #include "sysemu/replay.h"
+#include "sysemu/cpu-timers.h"
 #include "qemu/main-loop.h"
 #include "qemu/guest-random.h"
 #include "exec/exec-all.h"
diff --git a/accel/tcg/tcg-accel-ops-mttcg.c b/accel/tcg/tcg-accel-ops-mttcg.c
index 29632bd4c0..dc421c8fd7 100644
--- a/accel/tcg/tcg-accel-ops-mttcg.c
+++ b/accel/tcg/tcg-accel-ops-mttcg.c
@@ -27,6 +27,7 @@
 #include "qemu-common.h"
 #include "sysemu/tcg.h"
 #include "sysemu/replay.h"
+#include "sysemu/cpu-timers.h"
 #include "qemu/main-loop.h"
 #include "qemu/notify.h"
 #include "qemu/guest-random.h"
diff --git a/accel/tcg/tcg-accel-ops-rr.c b/accel/tcg/tcg-accel-ops-rr.c
index bf59f53dbc..a805fb6bdd 100644
--- a/accel/tcg/tcg-accel-ops-rr.c
+++ b/accel/tcg/tcg-accel-ops-rr.c
@@ -27,6 +27,7 @@
 #include "qemu-common.h"
 #include "sysemu/tcg.h"
 #include "sysemu/replay.h"
+#include "sysemu/cpu-timers.h"
 #include "qemu/main-loop.h"
 #include "qemu/notify.h"
 #include "qemu/guest-random.h"
diff --git a/accel/tcg/tcg-accel-ops.c b/accel/tcg/tcg-accel-ops.c
index 1a8e8390bd..ea7dcad674 100644
--- a/accel/tcg/tcg-accel-ops.c
+++ b/accel/tcg/tcg-accel-ops.c
@@ -29,6 +29,7 @@
 #include "qemu-common.h"
 #include "sysemu/tcg.h"
 #include "sysemu/replay.h"
+#include "sysemu/cpu-timers.h"
 #include "qemu/main-loop.h"
 #include "qemu/guest-random.h"
 #include "exec/exec-all.h"
diff --git a/cpu.c b/cpu.c
index 3ea38aea70..d564886149 100644
--- a/cpu.c
+++ b/cpu.c
@@ -35,10 +35,12 @@
 #include "sysemu/tcg.h"
 #include "sysemu/kvm.h"
 #include "sysemu/replay.h"
+#include "exec/exec-all.h"
 #include "exec/translate-all.h"
 #include "exec/log.h"
 #include "hw/core/accel-cpu.h"
 #include "trace/trace-root.h"
+#include "qemu/accel.h"
 
 uintptr_t qemu_host_page_size;
 intptr_t qemu_host_page_mask;
@@ -415,11 +417,11 @@ void cpu_abort(CPUState *cpu, const char *fmt, ...)
 
 /* physical memory access (slow version, mainly for debug) */
 #if defined(CONFIG_USER_ONLY)
-int cpu_memory_rw_debug(CPUState *cpu, target_ulong addr,
-                        void *ptr, target_ulong len, bool is_write)
+int cpu_memory_rw_debug(CPUState *cpu, vaddr addr,
+                        void *ptr, size_t len, bool is_write)
 {
     int flags;
-    target_ulong l, page;
+    vaddr l, page;
     void * p;
     uint8_t *buf = ptr;
 
diff --git a/include/exec/cpu-all.h b/include/exec/cpu-all.h
index 84caf5c3d9..c0f0fab28a 100644
--- a/include/exec/cpu-all.h
+++ b/include/exec/cpu-all.h
@@ -433,10 +433,6 @@ int cpu_exec(CPUState *cpu);
 void tcg_exec_realizefn(CPUState *cpu, Error **errp);
 void tcg_exec_unrealizefn(CPUState *cpu);
 
-/* 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);
-
 /**
  * cpu_set_cpustate_pointers(cpu)
  * @cpu: The cpu object
diff --git a/include/exec/cpu-common.h b/include/exec/cpu-common.h
index de5f444b19..7f7b5943c7 100644
--- a/include/exec/cpu-common.h
+++ b/include/exec/cpu-common.h
@@ -7,6 +7,18 @@
 #include "exec/hwaddr.h"
 #endif
 
+/**
+ * vaddr:
+ * Type wide enough to contain any #target_ulong virtual address.
+ */
+typedef uint64_t vaddr;
+#define VADDR_PRId PRId64
+#define VADDR_PRIu PRIu64
+#define VADDR_PRIo PRIo64
+#define VADDR_PRIx PRIx64
+#define VADDR_PRIX PRIX64
+#define VADDR_MAX UINT64_MAX
+
 /* Using intptr_t ensures that qemu_*_page_mask is sign-extended even
  * when intptr_t is 32-bit and we are aligning a long long.
  */
@@ -78,6 +90,28 @@ void qemu_ram_unset_migratable(RAMBlock *rb);
 size_t qemu_ram_pagesize(RAMBlock *block);
 size_t qemu_ram_pagesize_largest(void);
 
+/**
+ * cpu_address_space_init:
+ * @cpu: CPU to add this address space to
+ * @asidx: integer index of this address space
+ * @prefix: prefix to be used as name of address space
+ * @mr: the root memory region of address space
+ *
+ * Add the specified address space to the CPU's cpu_ases list.
+ * The address space added with @asidx 0 is the one used for the
+ * convenience pointer cpu->as.
+ * The target-specific code which registers ASes is responsible
+ * for defining what semantics address space 0, 1, 2, etc have.
+ *
+ * Before the first call to this function, the caller must set
+ * cpu->num_ases to the total number of address spaces it needs
+ * to support.
+ *
+ * Note that with KVM only one address space is supported.
+ */
+void cpu_address_space_init(CPUState *cpu, int asidx,
+                            const char *prefix, MemoryRegion *mr);
+
 void cpu_physical_memory_rw(hwaddr addr, void *buf,
                             hwaddr len, bool is_write);
 static inline void cpu_physical_memory_read(hwaddr addr,
@@ -90,6 +124,7 @@ static inline void cpu_physical_memory_write(hwaddr addr,
 {
     cpu_physical_memory_rw(addr, (void *)buf, len, true);
 }
+void cpu_reloading_memory_map(void);
 void *cpu_physical_memory_map(hwaddr addr,
                               hwaddr *plen,
                               bool is_write);
@@ -116,6 +151,10 @@ int ram_block_discard_range(RAMBlock *rb, uint64_t start, size_t length);
 
 #endif
 
+/* Returns: 0 on success, -1 on error */
+int cpu_memory_rw_debug(CPUState *cpu, vaddr addr,
+                        void *ptr, size_t len, bool is_write);
+
 /* vl.c */
 extern int singlestep;
 
diff --git a/include/exec/cpu_ldst.h b/include/exec/cpu_ldst.h
index da987fe8ad..6adacf8928 100644
--- a/include/exec/cpu_ldst.h
+++ b/include/exec/cpu_ldst.h
@@ -64,6 +64,7 @@
 
 #include "exec/memopidx.h"
 #include "qemu/int128.h"
+#include "cpu.h"
 
 #if defined(CONFIG_USER_ONLY)
 /* sparc32plus has 64bit long but 32bit space address
diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h
index 227e10ba56..d2cb0981f4 100644
--- a/include/exec/exec-all.h
+++ b/include/exec/exec-all.h
@@ -24,7 +24,6 @@
 #ifdef CONFIG_TCG
 #include "exec/cpu_ldst.h"
 #endif
-#include "sysemu/cpu-timers.h"
 
 /* allow to see translation results - the slowdown should be negligible, so we leave it */
 #define DEBUG_DISAS
@@ -81,31 +80,6 @@ static inline bool cpu_loop_exit_requested(CPUState *cpu)
     return (int32_t)qatomic_read(&cpu_neg(cpu)->icount_decr.u32) < 0;
 }
 
-#if !defined(CONFIG_USER_ONLY)
-void cpu_reloading_memory_map(void);
-/**
- * cpu_address_space_init:
- * @cpu: CPU to add this address space to
- * @asidx: integer index of this address space
- * @prefix: prefix to be used as name of address space
- * @mr: the root memory region of address space
- *
- * Add the specified address space to the CPU's cpu_ases list.
- * The address space added with @asidx 0 is the one used for the
- * convenience pointer cpu->as.
- * The target-specific code which registers ASes is responsible
- * for defining what semantics address space 0, 1, 2, etc have.
- *
- * Before the first call to this function, the caller must set
- * cpu->num_ases to the total number of address spaces it needs
- * to support.
- *
- * Note that with KVM only one address space is supported.
- */
-void cpu_address_space_init(CPUState *cpu, int asidx,
-                            const char *prefix, MemoryRegion *mr);
-#endif
-
 #if !defined(CONFIG_USER_ONLY) && defined(CONFIG_TCG)
 /* cputlb.c */
 /**
diff --git a/include/exec/gdbstub.h b/include/exec/gdbstub.h
index a024a0350d..89edf94d28 100644
--- a/include/exec/gdbstub.h
+++ b/include/exec/gdbstub.h
@@ -45,17 +45,6 @@ void gdb_do_syscall(gdb_syscall_complete_cb cb, const char *fmt, ...);
  */
 void gdb_do_syscallv(gdb_syscall_complete_cb cb, const char *fmt, va_list va);
 int use_gdb_syscalls(void);
-void gdb_set_stop_cpu(CPUState *cpu);
-
-/**
- * gdb_exit: exit gdb session, reporting inferior status
- * @code: exit code reported
- *
- * This closes the session and sends a final packet to GDB reporting
- * the exit status of the program. It also cleans up any connections
- * detritus before returning.
- */
-void gdb_exit(int code);
 
 #ifdef CONFIG_USER_ONLY
 /**
@@ -165,7 +154,7 @@ static inline uint8_t * gdb_get_reg_ptr(GByteArray *buf, int len)
 #define ldtul_p(addr) ldl_p(addr)
 #endif
 
-#endif
+#endif /* NEED_CPU_H */
 
 /**
  * gdbserver_start: start the gdb server
@@ -178,6 +167,18 @@ static inline uint8_t * gdb_get_reg_ptr(GByteArray *buf, int len)
 int gdbserver_start(const char *port_or_device);
 
 /**
+ * gdb_exit: exit gdb session, reporting inferior status
+ * @code: exit code reported
+ *
+ * This closes the session and sends a final packet to GDB reporting
+ * the exit status of the program. It also cleans up any connections
+ * detritus before returning.
+ */
+void gdb_exit(int code);
+
+void gdb_set_stop_cpu(CPUState *cpu);
+
+/**
  * gdb_has_xml:
  * This is an ugly hack to cope with both new and old gdb.
  * If gdb sends qXfer:features:read then assume we're talking to a newish
diff --git a/include/exec/poison.h b/include/exec/poison.h
index 7ad4ad18e8..7c5c02f03f 100644
--- a/include/exec/poison.h
+++ b/include/exec/poison.h
@@ -51,8 +51,6 @@
 #pragma GCC poison TARGET_PAGE_BITS
 #pragma GCC poison TARGET_PAGE_ALIGN
 
-#pragma GCC poison CPUArchState
-
 #pragma GCC poison CPU_INTERRUPT_HARD
 #pragma GCC poison CPU_INTERRUPT_EXITTB
 #pragma GCC poison CPU_INTERRUPT_HALT
diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h
index 76ab3b851c..0efc6153ed 100644
--- a/include/hw/core/cpu.h
+++ b/include/hw/core/cpu.h
@@ -22,6 +22,7 @@
 
 #include "hw/qdev-core.h"
 #include "disas/dis-asm.h"
+#include "exec/cpu-common.h"
 #include "exec/hwaddr.h"
 #include "exec/memattrs.h"
 #include "qapi/qapi-types-run-state.h"
@@ -36,18 +37,6 @@ typedef int (*WriteCoreDumpFunction)(const void *buf, size_t size,
                                      void *opaque);
 
 /**
- * vaddr:
- * Type wide enough to contain any #target_ulong virtual address.
- */
-typedef uint64_t vaddr;
-#define VADDR_PRId PRId64
-#define VADDR_PRIu PRIu64
-#define VADDR_PRIo PRIo64
-#define VADDR_PRIx PRIx64
-#define VADDR_PRIX PRIX64
-#define VADDR_MAX UINT64_MAX
-
-/**
  * SECTION:cpu
  * @section_id: QEMU-cpu
  * @title: CPU Class
@@ -66,6 +55,24 @@ typedef struct CPUClass CPUClass;
 DECLARE_CLASS_CHECKERS(CPUClass, CPU,
                        TYPE_CPU)
 
+/**
+ * OBJECT_DECLARE_CPU_TYPE:
+ * @CpuInstanceType: instance struct name
+ * @CpuClassType: class struct name
+ * @CPU_MODULE_OBJ_NAME: the CPU name in uppercase with underscore separators
+ *
+ * This macro is typically used in "cpu-qom.h" header file, and will:
+ *
+ *   - create the typedefs for the CPU object and class structs
+ *   - register the type for use with g_autoptr
+ *   - provide three standard type cast functions
+ *
+ * The object struct and class struct need to be declared manually.
+ */
+#define OBJECT_DECLARE_CPU_TYPE(CpuInstanceType, CpuClassType, CPU_MODULE_OBJ_NAME) \
+    typedef struct ArchCPU CpuInstanceType; \
+    OBJECT_DECLARE_TYPE(ArchCPU, CpuClassType, CPU_MODULE_OBJ_NAME);
+
 typedef enum MMUAccessType {
     MMU_DATA_LOAD  = 0,
     MMU_DATA_STORE = 1,
@@ -351,7 +358,7 @@ struct CPUState {
     AddressSpace *as;
     MemoryRegion *memory;
 
-    void *env_ptr; /* CPUArchState */
+    CPUArchState *env_ptr;
     IcountDecr *icount_decr_ptr;
 
     /* Accessed in parallel; all accesses must be atomic */
diff --git a/include/qemu/typedefs.h b/include/qemu/typedefs.h
index ee60eb3de4..c564f54c11 100644
--- a/include/qemu/typedefs.h
+++ b/include/qemu/typedefs.h
@@ -26,6 +26,7 @@ typedef struct AddressSpace AddressSpace;
 typedef struct AioContext AioContext;
 typedef struct Aml Aml;
 typedef struct AnnounceTimer AnnounceTimer;
+typedef struct ArchCPU ArchCPU;
 typedef struct BdrvDirtyBitmap BdrvDirtyBitmap;
 typedef struct BdrvDirtyBitmapIter BdrvDirtyBitmapIter;
 typedef struct BlockBackend BlockBackend;
@@ -39,6 +40,7 @@ typedef struct CompatProperty CompatProperty;
 typedef struct CoMutex CoMutex;
 typedef struct ConfidentialGuestSupport ConfidentialGuestSupport;
 typedef struct CPUAddressSpace CPUAddressSpace;
+typedef struct CPUArchState CPUArchState;
 typedef struct CPUState CPUState;
 typedef struct DeviceListener DeviceListener;
 typedef struct DeviceState DeviceState;
diff --git a/include/sysemu/accel-ops.h b/include/sysemu/accel-ops.h
index 032f6979d7..6013c9444c 100644
--- a/include/sysemu/accel-ops.h
+++ b/include/sysemu/accel-ops.h
@@ -28,8 +28,11 @@ struct AccelOpsClass {
     /* initialization function called when accel is chosen */
     void (*ops_init)(AccelOpsClass *ops);
 
+    bool (*cpus_are_resettable)(void);
+
     void (*create_vcpu_thread)(CPUState *cpu); /* MANDATORY NON-NULL */
     void (*kick_vcpu_thread)(CPUState *cpu);
+    bool (*cpu_thread_is_idle)(CPUState *cpu);
 
     void (*synchronize_post_reset)(CPUState *cpu);
     void (*synchronize_post_init)(CPUState *cpu);
diff --git a/include/sysemu/arch_init.h b/include/sysemu/arch_init.h
index 70c579560a..79c2591425 100644
--- a/include/sysemu/arch_init.h
+++ b/include/sysemu/arch_init.h
@@ -28,4 +28,6 @@ enum {
 
 extern const uint32_t arch_type;
 
+void qemu_init_arch_modules(void);
+
 #endif
diff --git a/include/sysemu/hax.h b/include/sysemu/hax.h
index 247f0661d1..bf8f99a824 100644
--- a/include/sysemu/hax.h
+++ b/include/sysemu/hax.h
@@ -25,17 +25,23 @@
 int hax_sync_vcpus(void);
 
 #ifdef NEED_CPU_H
+# ifdef CONFIG_HAX
+#  define CONFIG_HAX_IS_POSSIBLE
+# endif
+#else /* !NEED_CPU_H */
+# define CONFIG_HAX_IS_POSSIBLE
+#endif
 
-#ifdef CONFIG_HAX
+#ifdef CONFIG_HAX_IS_POSSIBLE
 
-int hax_enabled(void);
+extern bool hax_allowed;
 
-#else /* CONFIG_HAX */
+#define hax_enabled()               (hax_allowed)
 
-#define hax_enabled() (0)
+#else /* !CONFIG_HAX_IS_POSSIBLE */
 
-#endif /* CONFIG_HAX */
+#define hax_enabled()               (0)
 
-#endif /* NEED_CPU_H */
+#endif /* CONFIG_HAX_IS_POSSIBLE */
 
 #endif /* QEMU_HAX_H */
diff --git a/include/sysemu/hw_accel.h b/include/sysemu/hw_accel.h
index 01b5ebf442..22903a55f7 100644
--- a/include/sysemu/hw_accel.h
+++ b/include/sysemu/hw_accel.h
@@ -23,9 +23,4 @@ void cpu_synchronize_post_reset(CPUState *cpu);
 void cpu_synchronize_post_init(CPUState *cpu);
 void cpu_synchronize_pre_loadvm(CPUState *cpu);
 
-static inline bool cpu_check_are_resettable(void)
-{
-    return kvm_enabled() ? kvm_cpu_check_are_resettable() : true;
-}
-
 #endif /* QEMU_HW_ACCEL_H */
diff --git a/include/sysemu/kvm.h b/include/sysemu/kvm.h
index 6eb39a088b..a5bec96fb0 100644
--- a/include/sysemu/kvm.h
+++ b/include/sysemu/kvm.h
@@ -249,6 +249,9 @@ int kvm_has_intx_set_mask(void);
 bool kvm_arm_supports_user_irq(void);
 
 
+int kvm_on_sigbus_vcpu(CPUState *cpu, int code, void *addr);
+int kvm_on_sigbus(int code, void *addr);
+
 #ifdef NEED_CPU_H
 #include "cpu.h"
 
@@ -261,9 +264,6 @@ int kvm_remove_breakpoint(CPUState *cpu, target_ulong addr,
 void kvm_remove_all_breakpoints(CPUState *cpu);
 int kvm_update_guest_debug(CPUState *cpu, unsigned long reinject_trap);
 
-int kvm_on_sigbus_vcpu(CPUState *cpu, int code, void *addr);
-int kvm_on_sigbus(int code, void *addr);
-
 /* internal API */
 
 int kvm_ioctl(KVMState *s, int type, ...);
diff --git a/include/sysemu/memory_mapping.h b/include/sysemu/memory_mapping.h
index 4b20f1a639..3bbeb1bcb4 100644
--- a/include/sysemu/memory_mapping.h
+++ b/include/sysemu/memory_mapping.h
@@ -15,8 +15,7 @@
 #define MEMORY_MAPPING_H
 
 #include "qemu/queue.h"
-#include "exec/cpu-defs.h"
-#include "exec/memory.h"
+#include "exec/cpu-common.h"
 
 typedef struct GuestPhysBlock {
     /* visible to guest, reflects PCI hole, etc */
@@ -43,7 +42,7 @@ typedef struct GuestPhysBlockList {
 /* The physical and virtual address in the memory mapping are contiguous. */
 typedef struct MemoryMapping {
     hwaddr phys_addr;
-    target_ulong virt_addr;
+    vaddr virt_addr;
     ram_addr_t length;
     QTAILQ_ENTRY(MemoryMapping) next;
 } MemoryMapping;
diff --git a/meson.build b/meson.build
index 038502714a..746ce478e0 100644
--- a/meson.build
+++ b/meson.build
@@ -2432,8 +2432,8 @@ if get_option('cfi') and slirp_opt == 'system'
 endif
 
 fdt = not_found
-fdt_opt = get_option('fdt')
 if have_system
+  fdt_opt = get_option('fdt')
   if fdt_opt in ['enabled', 'auto', 'system']
     have_internal = fs.exists(meson.current_source_dir() / 'dtc/libfdt/Makefile.libfdt')
     fdt = cc.find_library('fdt', kwargs: static_kwargs,
@@ -2476,6 +2476,8 @@ if have_system
     fdt = declare_dependency(link_with: libfdt,
                              include_directories: fdt_inc)
   endif
+else
+  fdt_opt = 'disabled'
 endif
 if not fdt.found() and fdt_required.length() > 0
   error('fdt not available but required by targets ' + ', '.join(fdt_required))
diff --git a/softmmu/arch_init.c b/softmmu/arch_init.c
index 8919405c7b..79716f959b 100644
--- a/softmmu/arch_init.c
+++ b/softmmu/arch_init.c
@@ -22,6 +22,7 @@
  * THE SOFTWARE.
  */
 #include "qemu/osdep.h"
+#include "qemu/module.h"
 #include "sysemu/arch_init.h"
 
 #ifdef TARGET_SPARC
@@ -39,3 +40,11 @@ int graphic_depth = 32;
 #endif
 
 const uint32_t arch_type = QEMU_ARCH;
+
+void qemu_init_arch_modules(void)
+{
+#ifdef CONFIG_MODULES
+    module_init_info(qemu_modinfo);
+    module_allow_arch(TARGET_NAME);
+#endif
+}
diff --git a/softmmu/cpu-timers.c b/softmmu/cpu-timers.c
index 34ddfa02f1..204d946a17 100644
--- a/softmmu/cpu-timers.c
+++ b/softmmu/cpu-timers.c
@@ -28,7 +28,6 @@
 #include "migration/vmstate.h"
 #include "qapi/error.h"
 #include "qemu/error-report.h"
-#include "exec/exec-all.h"
 #include "sysemu/cpus.h"
 #include "qemu/main-loop.h"
 #include "qemu/option.h"
diff --git a/softmmu/cpus.c b/softmmu/cpus.c
index 1681844b61..e1d84c8ccb 100644
--- a/softmmu/cpus.c
+++ b/softmmu/cpus.c
@@ -33,7 +33,7 @@
 #include "qapi/qmp/qerror.h"
 #include "exec/gdbstub.h"
 #include "sysemu/hw_accel.h"
-#include "exec/exec-all.h"
+#include "exec/cpu-common.h"
 #include "qemu/thread.h"
 #include "qemu/plugin.h"
 #include "sysemu/cpus.h"
@@ -67,6 +67,11 @@
 
 static QemuMutex qemu_global_mutex;
 
+/*
+ * The chosen accelerator is supposed to register this.
+ */
+static const AccelOpsClass *cpus_accel;
+
 bool cpu_is_stopped(CPUState *cpu)
 {
     return cpu->stopped || !runstate_is_running();
@@ -85,10 +90,12 @@ bool cpu_thread_is_idle(CPUState *cpu)
     if (cpu_is_stopped(cpu)) {
         return true;
     }
-    if (!cpu->halted || cpu_has_work(cpu) ||
-        kvm_halt_in_kernel() || whpx_apic_in_platform()) {
+    if (!cpu->halted || cpu_has_work(cpu)) {
         return false;
     }
+    if (cpus_accel->cpu_thread_is_idle) {
+        return cpus_accel->cpu_thread_is_idle(cpu);
+    }
     return true;
 }
 
@@ -122,11 +129,6 @@ void hw_error(const char *fmt, ...)
     abort();
 }
 
-/*
- * The chosen accelerator is supposed to register this.
- */
-static const AccelOpsClass *cpus_accel;
-
 void cpu_synchronize_all_states(void)
 {
     CPUState *cpu;
@@ -193,7 +195,10 @@ void cpu_synchronize_pre_loadvm(CPUState *cpu)
 
 bool cpus_are_resettable(void)
 {
-    return cpu_check_are_resettable();
+    if (cpus_accel->cpus_are_resettable) {
+        return cpus_accel->cpus_are_resettable();
+    }
+    return true;
 }
 
 int64_t cpus_get_virtual_clock(void)
diff --git a/softmmu/globals.c b/softmmu/globals.c
index 7d0fc81183..3ebd718e35 100644
--- a/softmmu/globals.c
+++ b/softmmu/globals.c
@@ -25,8 +25,6 @@
 #include "qemu/osdep.h"
 #include "exec/cpu-common.h"
 #include "hw/display/vga.h"
-#include "hw/i386/pc.h"
-#include "hw/i386/x86.h"
 #include "hw/loader.h"
 #include "hw/xen/xen.h"
 #include "net/net.h"
diff --git a/softmmu/memory_mapping.c b/softmmu/memory_mapping.c
index a62eaa49cc..8320165ea2 100644
--- a/softmmu/memory_mapping.c
+++ b/softmmu/memory_mapping.c
@@ -17,6 +17,7 @@
 #include "sysemu/memory_mapping.h"
 #include "exec/memory.h"
 #include "exec/address-spaces.h"
+#include "hw/core/cpu.h"
 
 //#define DEBUG_GUEST_PHYS_REGION_ADD
 
diff --git a/softmmu/meson.build b/softmmu/meson.build
index 39f766ce7c..8138248661 100644
--- a/softmmu/meson.build
+++ b/softmmu/meson.build
@@ -1,20 +1,9 @@
 specific_ss.add(when: 'CONFIG_SOFTMMU', if_true: [files(
   'arch_init.c',
-  'balloon.c',
-  'cpus.c',
-  'cpu-throttle.c',
-  'datadir.c',
-  'globals.c',
-  'physmem.c',
   'ioport.c',
-  'rtc.c',
-  'runstate.c',
   'memory.c',
-  'memory_mapping.c',
+  'physmem.c',
   'qtest.c',
-  'vl.c',
-  'cpu-timers.c',
-  'runstate-action.c',
 )])
 
 specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: [files(
@@ -22,9 +11,20 @@ specific_ss.add(when: ['CONFIG_SOFTMMU', 'CONFIG_TCG'], if_true: [files(
 )])
 
 softmmu_ss.add(files(
+  'balloon.c',
   'bootdevice.c',
+  'cpus.c',
+  'cpu-throttle.c',
+  'cpu-timers.c',
+  'datadir.c',
   'dma-helpers.c',
+  'globals.c',
+  'memory_mapping.c',
   'qdev-monitor.c',
+  'rtc.c',
+  'runstate-action.c',
+  'runstate.c',
+  'vl.c',
 ), sdl, libpmem, libdaxctl)
 
 if have_tpm
diff --git a/softmmu/physmem.c b/softmmu/physmem.c
index a13289a594..364d563ee4 100644
--- a/softmmu/physmem.c
+++ b/softmmu/physmem.c
@@ -61,7 +61,6 @@
 
 #include "exec/memory-internal.h"
 #include "exec/ram_addr.h"
-#include "exec/log.h"
 
 #include "qemu/pmem.h"
 
@@ -3436,11 +3435,11 @@ address_space_write_cached_slow(MemoryRegionCache *cache, hwaddr addr,
 #include "memory_ldst.c.inc"
 
 /* virtual memory access for debug (includes writing to ROM) */
-int cpu_memory_rw_debug(CPUState *cpu, target_ulong addr,
-                        void *ptr, target_ulong len, bool is_write)
+int cpu_memory_rw_debug(CPUState *cpu, vaddr addr,
+                        void *ptr, size_t len, bool is_write)
 {
     hwaddr phys_addr;
-    target_ulong l, page;
+    vaddr l, page;
     uint8_t *buf = ptr;
 
     cpu_synchronize_state(cpu);
diff --git a/softmmu/vl.c b/softmmu/vl.c
index 1fe028800f..0b81f61535 100644
--- a/softmmu/vl.c
+++ b/softmmu/vl.c
@@ -2815,10 +2815,7 @@ void qemu_init(int argc, char **argv, char **envp)
     error_init(argv[0]);
     qemu_init_exec_dir(argv[0]);
 
-#ifdef CONFIG_MODULES
-    module_init_info(qemu_modinfo);
-    module_allow_arch(TARGET_NAME);
-#endif
+    qemu_init_arch_modules();
 
     qemu_init_subsystems();
 
diff --git a/target/alpha/cpu-qom.h b/target/alpha/cpu-qom.h
index 7bb9173c57..1f200724b6 100644
--- a/target/alpha/cpu-qom.h
+++ b/target/alpha/cpu-qom.h
@@ -25,8 +25,7 @@
 
 #define TYPE_ALPHA_CPU "alpha-cpu"
 
-OBJECT_DECLARE_TYPE(AlphaCPU, AlphaCPUClass,
-                    ALPHA_CPU)
+OBJECT_DECLARE_CPU_TYPE(AlphaCPU, AlphaCPUClass, ALPHA_CPU)
 
 /**
  * AlphaCPUClass:
diff --git a/target/alpha/cpu.h b/target/alpha/cpu.h
index e819211503..58f00b7814 100644
--- a/target/alpha/cpu.h
+++ b/target/alpha/cpu.h
@@ -197,9 +197,7 @@ enum {
 #define MMU_USER_IDX     1
 #define MMU_PHYS_IDX     2
 
-typedef struct CPUAlphaState CPUAlphaState;
-
-struct CPUAlphaState {
+typedef struct CPUArchState {
     uint64_t ir[31];
     float64 fir[31];
     uint64_t pc;
@@ -251,7 +249,7 @@ struct CPUAlphaState {
     uint32_t features;
     uint32_t amask;
     int implver;
-};
+} CPUAlphaState;
 
 /**
  * AlphaCPU:
@@ -259,7 +257,7 @@ struct CPUAlphaState {
  *
  * An Alpha CPU.
  */
-struct AlphaCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -285,9 +283,6 @@ int alpha_cpu_gdb_write_register(CPUState *cpu, uint8_t *buf, int reg);
 
 #define cpu_list alpha_cpu_list
 
-typedef CPUAlphaState CPUArchState;
-typedef AlphaCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 enum {
diff --git a/target/alpha/translate.c b/target/alpha/translate.c
index ca78a0faed..66768ab47a 100644
--- a/target/alpha/translate.c
+++ b/target/alpha/translate.c
@@ -20,7 +20,6 @@
 #include "qemu/osdep.h"
 #include "cpu.h"
 #include "sysemu/cpus.h"
-#include "sysemu/cpu-timers.h"
 #include "disas/disas.h"
 #include "qemu/host-utils.h"
 #include "exec/exec-all.h"
diff --git a/target/arm/cpu-qom.h b/target/arm/cpu-qom.h
index a22bd506d0..64c44cef2d 100644
--- a/target/arm/cpu-qom.h
+++ b/target/arm/cpu-qom.h
@@ -27,8 +27,7 @@ struct arm_boot_info;
 
 #define TYPE_ARM_CPU "arm-cpu"
 
-OBJECT_DECLARE_TYPE(ARMCPU, ARMCPUClass,
-                    ARM_CPU)
+OBJECT_DECLARE_CPU_TYPE(ARMCPU, ARMCPUClass, ARM_CPU)
 
 #define TYPE_ARM_MAX_CPU "max-" TYPE_ARM_CPU
 
diff --git a/target/arm/cpu.h b/target/arm/cpu.h
index 24d9fff170..0b4b5bbf54 100644
--- a/target/arm/cpu.h
+++ b/target/arm/cpu.h
@@ -232,7 +232,7 @@ typedef struct CPUARMTBFlags {
     target_ulong flags2;
 } CPUARMTBFlags;
 
-typedef struct CPUARMState {
+typedef struct CPUArchState {
     /* Regs for current mode.  */
     uint32_t regs[16];
 
@@ -774,7 +774,7 @@ typedef struct ARMISARegisters ARMISARegisters;
  *
  * An ARM CPU core.
  */
-struct ARMCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -3410,9 +3410,6 @@ static inline bool arm_cpu_data_is_big_endian(CPUARMState *env)
     }
 }
 
-typedef CPUARMState CPUArchState;
-typedef ARMCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 /*
diff --git a/target/arm/hvf_arm.h b/target/arm/hvf_arm.h
index ea238cff83..9a9d1a0bf5 100644
--- a/target/arm/hvf_arm.h
+++ b/target/arm/hvf_arm.h
@@ -13,6 +13,6 @@
 
 #include "cpu.h"
 
-void hvf_arm_set_cpu_features_from_host(struct ARMCPU *cpu);
+void hvf_arm_set_cpu_features_from_host(ARMCPU *cpu);
 
 #endif
diff --git a/target/avr/cpu-qom.h b/target/avr/cpu-qom.h
index 14e5b3ce72..32a1c762e6 100644
--- a/target/avr/cpu-qom.h
+++ b/target/avr/cpu-qom.h
@@ -26,8 +26,7 @@
 
 #define TYPE_AVR_CPU "avr-cpu"
 
-OBJECT_DECLARE_TYPE(AVRCPU, AVRCPUClass,
-                    AVR_CPU)
+OBJECT_DECLARE_CPU_TYPE(AVRCPU, AVRCPUClass, AVR_CPU)
 
 /**
  *  AVRCPUClass:
diff --git a/target/avr/cpu.h b/target/avr/cpu.h
index dceacf3cd7..55497f851d 100644
--- a/target/avr/cpu.h
+++ b/target/avr/cpu.h
@@ -108,9 +108,7 @@ typedef enum AVRFeature {
     AVR_FEATURE_RAMPZ,
 } AVRFeature;
 
-typedef struct CPUAVRState CPUAVRState;
-
-struct CPUAVRState {
+typedef struct CPUArchState {
     uint32_t pc_w; /* 0x003fffff up to 22 bits */
 
     uint32_t sregC; /* 0x00000001 1 bit */
@@ -137,7 +135,7 @@ struct CPUAVRState {
     bool fullacc; /* CPU/MEM if true MEM only otherwise */
 
     uint64_t features;
-};
+} CPUAVRState;
 
 /**
  *  AVRCPU:
@@ -145,14 +143,14 @@ struct CPUAVRState {
  *
  *  A AVR CPU.
  */
-typedef struct AVRCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
 
     CPUNegativeOffsetState neg;
     CPUAVRState env;
-} AVRCPU;
+};
 
 extern const struct VMStateDescription vms_avr_cpu;
 
@@ -247,9 +245,6 @@ bool avr_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
                       MMUAccessType access_type, int mmu_idx,
                       bool probe, uintptr_t retaddr);
 
-typedef CPUAVRState CPUArchState;
-typedef AVRCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 #endif /* !defined (QEMU_AVR_CPU_H) */
diff --git a/target/cris/cpu-qom.h b/target/cris/cpu-qom.h
index 2596edc7e3..71e8af0e70 100644
--- a/target/cris/cpu-qom.h
+++ b/target/cris/cpu-qom.h
@@ -25,8 +25,7 @@
 
 #define TYPE_CRIS_CPU "cris-cpu"
 
-OBJECT_DECLARE_TYPE(CRISCPU, CRISCPUClass,
-                    CRIS_CPU)
+OBJECT_DECLARE_CPU_TYPE(CRISCPU, CRISCPUClass, CRIS_CPU)
 
 /**
  * CRISCPUClass:
diff --git a/target/cris/cpu.h b/target/cris/cpu.h
index b445b194ea..e6776f25b1 100644
--- a/target/cris/cpu.h
+++ b/target/cris/cpu.h
@@ -105,7 +105,7 @@ typedef struct {
     uint32_t lo;
 } TLBSet;
 
-typedef struct CPUCRISState {
+typedef struct CPUArchState {
 	uint32_t regs[16];
 	/* P0 - P15 are referred to as special registers in the docs.  */
 	uint32_t pregs[16];
@@ -173,7 +173,7 @@ typedef struct CPUCRISState {
  *
  * A CRIS CPU.
  */
-struct CRISCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -265,9 +265,6 @@ static inline int cpu_mmu_index (CPUCRISState *env, bool ifetch)
 #define SFR_RW_MM_TLB_LO   env->pregs[PR_SRS]][5
 #define SFR_RW_MM_TLB_HI   env->pregs[PR_SRS]][6
 
-typedef CPUCRISState CPUArchState;
-typedef CRISCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 static inline void cpu_get_tb_cpu_state(CPUCRISState *env, target_ulong *pc,
diff --git a/target/hexagon/cpu.h b/target/hexagon/cpu.h
index 58a0d3870b..2a65a57bab 100644
--- a/target/hexagon/cpu.h
+++ b/target/hexagon/cpu.h
@@ -1,5 +1,5 @@
 /*
- *  Copyright(c) 2019-2021 Qualcomm Innovation Center, Inc. All Rights Reserved.
+ *  Copyright(c) 2019-2022 Qualcomm Innovation Center, Inc. All Rights Reserved.
  *
  *  This program is free software; you can redistribute it and/or modify
  *  it under the terms of the GNU General Public License as published by
@@ -18,14 +18,13 @@
 #ifndef HEXAGON_CPU_H
 #define HEXAGON_CPU_H
 
-/* Forward declaration needed by some of the header files */
-typedef struct CPUHexagonState CPUHexagonState;
-
 #include "fpu/softfloat-types.h"
 
 #include "exec/cpu-defs.h"
 #include "hex_regs.h"
 #include "mmvec/mmvec.h"
+#include "qom/object.h"
+#include "hw/core/cpu.h"
 
 #define NUM_PREGS 4
 #define TOTAL_PER_THREAD_REGS 64
@@ -75,7 +74,7 @@ typedef struct {
 /* Maximum number of vector temps in a packet */
 #define VECTOR_TEMPS_MAX            4
 
-struct CPUHexagonState {
+typedef struct CPUArchState {
     target_ulong gpr[TOTAL_PER_THREAD_REGS];
     target_ulong pred[NUM_PREGS];
     target_ulong branch_taken;
@@ -129,14 +128,9 @@ struct CPUHexagonState {
     target_ulong vstore_pending[VSTORES_MAX];
     bool vtcm_pending;
     VTCMStoreLog vtcm_log;
-};
+} CPUHexagonState;
 
-#define HEXAGON_CPU_CLASS(klass) \
-    OBJECT_CLASS_CHECK(HexagonCPUClass, (klass), TYPE_HEXAGON_CPU)
-#define HEXAGON_CPU(obj) \
-    OBJECT_CHECK(HexagonCPU, (obj), TYPE_HEXAGON_CPU)
-#define HEXAGON_CPU_GET_CLASS(obj) \
-    OBJECT_GET_CLASS(HexagonCPUClass, (obj), TYPE_HEXAGON_CPU)
+OBJECT_DECLARE_CPU_TYPE(HexagonCPU, HexagonCPUClass, HEXAGON_CPU)
 
 typedef struct HexagonCPUClass {
     /*< private >*/
@@ -146,7 +140,7 @@ typedef struct HexagonCPUClass {
     DeviceReset parent_reset;
 } HexagonCPUClass;
 
-typedef struct HexagonCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -155,7 +149,7 @@ typedef struct HexagonCPU {
 
     bool lldb_compat;
     target_ulong lldb_stack_adjust;
-} HexagonCPU;
+};
 
 #include "cpu_bits.h"
 
@@ -180,7 +174,6 @@ static inline int cpu_mmu_index(CPUHexagonState *env, bool ifetch)
 #endif
 }
 
-typedef struct CPUHexagonState CPUArchState;
 typedef HexagonCPU ArchCPU;
 
 void hexagon_translate_init(void);
diff --git a/target/hppa/cpu-qom.h b/target/hppa/cpu-qom.h
index d424f88370..b96e0318c7 100644
--- a/target/hppa/cpu-qom.h
+++ b/target/hppa/cpu-qom.h
@@ -25,8 +25,7 @@
 
 #define TYPE_HPPA_CPU "hppa-cpu"
 
-OBJECT_DECLARE_TYPE(HPPACPU, HPPACPUClass,
-                    HPPA_CPU)
+OBJECT_DECLARE_CPU_TYPE(HPPACPU, HPPACPUClass, HPPA_CPU)
 
 /**
  * HPPACPUClass:
diff --git a/target/hppa/cpu.h b/target/hppa/cpu.h
index 93c119532a..4cc936b6bf 100644
--- a/target/hppa/cpu.h
+++ b/target/hppa/cpu.h
@@ -138,8 +138,6 @@
 #define CR_IPSW          22
 #define CR_EIRR          23
 
-typedef struct CPUHPPAState CPUHPPAState;
-
 #if TARGET_REGISTER_BITS == 32
 typedef uint32_t target_ureg;
 typedef int32_t  target_sreg;
@@ -168,7 +166,7 @@ typedef struct {
     unsigned access_id : 16;
 } hppa_tlb_entry;
 
-struct CPUHPPAState {
+typedef struct CPUArchState {
     target_ureg gr[32];
     uint64_t fr[32];
     uint64_t sr[8];          /* stored shifted into place for gva */
@@ -207,7 +205,7 @@ struct CPUHPPAState {
     /* ??? We should use a more intelligent data structure.  */
     hppa_tlb_entry tlb[HPPA_TLB_ENTRIES];
     uint32_t tlb_last;
-};
+} CPUHPPAState;
 
 /**
  * HPPACPU:
@@ -215,7 +213,7 @@ struct CPUHPPAState {
  *
  * An HPPA CPU.
  */
-struct HPPACPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -225,10 +223,6 @@ struct HPPACPU {
     QEMUTimer *alarm_timer;
 };
 
-
-typedef CPUHPPAState CPUArchState;
-typedef HPPACPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 static inline int cpu_mmu_index(CPUHPPAState *env, bool ifetch)
diff --git a/target/i386/cpu-qom.h b/target/i386/cpu-qom.h
index f9923cee04..c557a522e1 100644
--- a/target/i386/cpu-qom.h
+++ b/target/i386/cpu-qom.h
@@ -30,8 +30,7 @@
 #define TYPE_X86_CPU "i386-cpu"
 #endif
 
-OBJECT_DECLARE_TYPE(X86CPU, X86CPUClass,
-                    X86_CPU)
+OBJECT_DECLARE_CPU_TYPE(X86CPU, X86CPUClass, X86_CPU)
 
 typedef struct X86CPUModel X86CPUModel;
 
diff --git a/target/i386/cpu.h b/target/i386/cpu.h
index e69ab5dd78..e11734ba86 100644
--- a/target/i386/cpu.h
+++ b/target/i386/cpu.h
@@ -1431,7 +1431,7 @@ typedef struct HVFX86LazyFlags {
     target_ulong auxbits;
 } HVFX86LazyFlags;
 
-typedef struct CPUX86State {
+typedef struct CPUArchState {
     /* standard registers */
     target_ulong regs[CPU_NB_REGS];
     target_ulong eip;
@@ -1707,7 +1707,7 @@ struct kvm_msrs;
  *
  * An x86 CPU.
  */
-struct X86CPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -2074,9 +2074,6 @@ static inline int cpu_mmu_index_kernel(CPUX86State *env)
 #define CC_SRC2 (env->cc_src2)
 #define CC_OP   (env->cc_op)
 
-typedef CPUX86State CPUArchState;
-typedef X86CPU ArchCPU;
-
 #include "exec/cpu-all.h"
 #include "svm.h"
 
diff --git a/target/i386/hax/hax-all.c b/target/i386/hax/hax-all.c
index bf65ed6fa9..81f665e212 100644
--- a/target/i386/hax/hax-all.c
+++ b/target/i386/hax/hax-all.c
@@ -49,18 +49,13 @@ const uint32_t hax_cur_version = 0x4; /* API v4: unmapping and MMIO moves */
 /* Minimum HAX kernel version */
 const uint32_t hax_min_version = 0x4; /* API v4: supports unmapping */
 
-static bool hax_allowed;
+bool hax_allowed;
 
 struct hax_state hax_global;
 
 static void hax_vcpu_sync_state(CPUArchState *env, int modified);
 static int hax_arch_get_registers(CPUArchState *env);
 
-int hax_enabled(void)
-{
-    return hax_allowed;
-}
-
 int valid_hax_tunnel_size(uint16_t size)
 {
     return size >= sizeof(struct hax_tunnel);
@@ -227,7 +222,7 @@ int hax_init_vcpu(CPUState *cpu)
 
     cpu->hax_vcpu = hax_global.vm->vcpus[cpu->cpu_index];
     cpu->vcpu_dirty = true;
-    qemu_register_reset(hax_reset_vcpu_state, (CPUArchState *) (cpu->env_ptr));
+    qemu_register_reset(hax_reset_vcpu_state, cpu->env_ptr);
 
     return ret;
 }
@@ -674,7 +669,7 @@ void hax_cpu_synchronize_pre_loadvm(CPUState *cpu)
 
 int hax_smp_cpu_exec(CPUState *cpu)
 {
-    CPUArchState *env = (CPUArchState *) (cpu->env_ptr);
+    CPUArchState *env = cpu->env_ptr;
     int fatal;
     int ret;
 
diff --git a/target/i386/hvf/x86_emu.c b/target/i386/hvf/x86_emu.c
index 7c8203b21f..050428795b 100644
--- a/target/i386/hvf/x86_emu.c
+++ b/target/i386/hvf/x86_emu.c
@@ -171,12 +171,12 @@ void write_val_to_reg(target_ulong reg_ptr, target_ulong val, int size)
     }
 }
 
-static bool is_host_reg(struct CPUX86State *env, target_ulong ptr)
+static bool is_host_reg(CPUX86State *env, target_ulong ptr)
 {
     return (ptr - (target_ulong)&env->regs[0]) < sizeof(env->regs);
 }
 
-void write_val_ext(struct CPUX86State *env, target_ulong ptr, target_ulong val, int size)
+void write_val_ext(CPUX86State *env, target_ulong ptr, target_ulong val, int size)
 {
     if (is_host_reg(env, ptr)) {
         write_val_to_reg(ptr, val, size);
@@ -185,14 +185,14 @@ void write_val_ext(struct CPUX86State *env, target_ulong ptr, target_ulong val,
     vmx_write_mem(env_cpu(env), ptr, &val, size);
 }
 
-uint8_t *read_mmio(struct CPUX86State *env, target_ulong ptr, int bytes)
+uint8_t *read_mmio(CPUX86State *env, target_ulong ptr, int bytes)
 {
     vmx_read_mem(env_cpu(env), env->hvf_mmio_buf, ptr, bytes);
     return env->hvf_mmio_buf;
 }
 
 
-target_ulong read_val_ext(struct CPUX86State *env, target_ulong ptr, int size)
+target_ulong read_val_ext(CPUX86State *env, target_ulong ptr, int size)
 {
     target_ulong val;
     uint8_t *mmio_ptr;
@@ -222,7 +222,7 @@ target_ulong read_val_ext(struct CPUX86State *env, target_ulong ptr, int size)
     return val;
 }
 
-static void fetch_operands(struct CPUX86State *env, struct x86_decode *decode,
+static void fetch_operands(CPUX86State *env, struct x86_decode *decode,
                            int n, bool val_op0, bool val_op1, bool val_op2)
 {
     int i;
@@ -261,7 +261,7 @@ static void fetch_operands(struct CPUX86State *env, struct x86_decode *decode,
     }
 }
 
-static void exec_mov(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_mov(CPUX86State *env, struct x86_decode *decode)
 {
     fetch_operands(env, decode, 2, false, true, false);
     write_val_ext(env, decode->op[0].ptr, decode->op[1].val,
@@ -270,49 +270,49 @@ static void exec_mov(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_add(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_add(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, +, SET_FLAGS_OSZAPC_ADD, true);
     env->eip += decode->len;
 }
 
-static void exec_or(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_or(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, |, SET_FLAGS_OSZAPC_LOGIC, true);
     env->eip += decode->len;
 }
 
-static void exec_adc(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_adc(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, +get_CF(env)+, SET_FLAGS_OSZAPC_ADD, true);
     env->eip += decode->len;
 }
 
-static void exec_sbb(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_sbb(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, -get_CF(env)-, SET_FLAGS_OSZAPC_SUB, true);
     env->eip += decode->len;
 }
 
-static void exec_and(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_and(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, &, SET_FLAGS_OSZAPC_LOGIC, true);
     env->eip += decode->len;
 }
 
-static void exec_sub(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_sub(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, -, SET_FLAGS_OSZAPC_SUB, true);
     env->eip += decode->len;
 }
 
-static void exec_xor(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_xor(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, ^, SET_FLAGS_OSZAPC_LOGIC, true);
     env->eip += decode->len;
 }
 
-static void exec_neg(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_neg(CPUX86State *env, struct x86_decode *decode)
 {
     /*EXEC_2OP_FLAGS_CMD(env, decode, -, SET_FLAGS_OSZAPC_SUB, false);*/
     int32_t val;
@@ -335,13 +335,13 @@ static void exec_neg(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_cmp(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_cmp(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, -, SET_FLAGS_OSZAPC_SUB, false);
     env->eip += decode->len;
 }
 
-static void exec_inc(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_inc(CPUX86State *env, struct x86_decode *decode)
 {
     decode->op[1].type = X86_VAR_IMMEDIATE;
     decode->op[1].val = 0;
@@ -351,7 +351,7 @@ static void exec_inc(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_dec(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_dec(CPUX86State *env, struct x86_decode *decode)
 {
     decode->op[1].type = X86_VAR_IMMEDIATE;
     decode->op[1].val = 0;
@@ -360,13 +360,13 @@ static void exec_dec(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_tst(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_tst(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, &, SET_FLAGS_OSZAPC_LOGIC, false);
     env->eip += decode->len;
 }
 
-static void exec_not(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_not(CPUX86State *env, struct x86_decode *decode)
 {
     fetch_operands(env, decode, 1, true, false, false);
 
@@ -375,7 +375,7 @@ static void exec_not(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-void exec_movzx(struct CPUX86State *env, struct x86_decode *decode)
+void exec_movzx(CPUX86State *env, struct x86_decode *decode)
 {
     int src_op_size;
     int op_size = decode->operand_size;
@@ -395,7 +395,7 @@ void exec_movzx(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_out(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_out(CPUX86State *env, struct x86_decode *decode)
 {
     switch (decode->opcode[0]) {
     case 0xe6:
@@ -419,7 +419,7 @@ static void exec_out(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_in(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_in(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong val = 0;
     switch (decode->opcode[0]) {
@@ -455,7 +455,7 @@ static void exec_in(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static inline void string_increment_reg(struct CPUX86State *env, int reg,
+static inline void string_increment_reg(CPUX86State *env, int reg,
                                         struct x86_decode *decode)
 {
     target_ulong val = read_reg(env, reg, decode->addressing_size);
@@ -467,8 +467,8 @@ static inline void string_increment_reg(struct CPUX86State *env, int reg,
     write_reg(env, reg, val, decode->addressing_size);
 }
 
-static inline void string_rep(struct CPUX86State *env, struct x86_decode *decode,
-                              void (*func)(struct CPUX86State *env,
+static inline void string_rep(CPUX86State *env, struct x86_decode *decode,
+                              void (*func)(CPUX86State *env,
                                            struct x86_decode *ins), int rep)
 {
     target_ulong rcx = read_reg(env, R_ECX, decode->addressing_size);
@@ -484,7 +484,7 @@ static inline void string_rep(struct CPUX86State *env, struct x86_decode *decode
     }
 }
 
-static void exec_ins_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_ins_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong addr = linear_addr_size(env_cpu(env), RDI(env),
                                          decode->addressing_size, R_ES);
@@ -497,7 +497,7 @@ static void exec_ins_single(struct CPUX86State *env, struct x86_decode *decode)
     string_increment_reg(env, R_EDI, decode);
 }
 
-static void exec_ins(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_ins(CPUX86State *env, struct x86_decode *decode)
 {
     if (decode->rep) {
         string_rep(env, decode, exec_ins_single, 0);
@@ -508,7 +508,7 @@ static void exec_ins(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_outs_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_outs_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong addr = decode_linear_addr(env, decode, RSI(env), R_DS);
 
@@ -520,7 +520,7 @@ static void exec_outs_single(struct CPUX86State *env, struct x86_decode *decode)
     string_increment_reg(env, R_ESI, decode);
 }
 
-static void exec_outs(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_outs(CPUX86State *env, struct x86_decode *decode)
 {
     if (decode->rep) {
         string_rep(env, decode, exec_outs_single, 0);
@@ -531,7 +531,7 @@ static void exec_outs(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_movs_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_movs_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong src_addr;
     target_ulong dst_addr;
@@ -548,7 +548,7 @@ static void exec_movs_single(struct CPUX86State *env, struct x86_decode *decode)
     string_increment_reg(env, R_EDI, decode);
 }
 
-static void exec_movs(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_movs(CPUX86State *env, struct x86_decode *decode)
 {
     if (decode->rep) {
         string_rep(env, decode, exec_movs_single, 0);
@@ -559,7 +559,7 @@ static void exec_movs(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_cmps_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_cmps_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong src_addr;
     target_ulong dst_addr;
@@ -579,7 +579,7 @@ static void exec_cmps_single(struct CPUX86State *env, struct x86_decode *decode)
     string_increment_reg(env, R_EDI, decode);
 }
 
-static void exec_cmps(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_cmps(CPUX86State *env, struct x86_decode *decode)
 {
     if (decode->rep) {
         string_rep(env, decode, exec_cmps_single, decode->rep);
@@ -590,7 +590,7 @@ static void exec_cmps(struct CPUX86State *env, struct x86_decode *decode)
 }
 
 
-static void exec_stos_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_stos_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong addr;
     target_ulong val;
@@ -604,7 +604,7 @@ static void exec_stos_single(struct CPUX86State *env, struct x86_decode *decode)
 }
 
 
-static void exec_stos(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_stos(CPUX86State *env, struct x86_decode *decode)
 {
     if (decode->rep) {
         string_rep(env, decode, exec_stos_single, 0);
@@ -615,7 +615,7 @@ static void exec_stos(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_scas_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_scas_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong addr;
 
@@ -628,7 +628,7 @@ static void exec_scas_single(struct CPUX86State *env, struct x86_decode *decode)
     string_increment_reg(env, R_EDI, decode);
 }
 
-static void exec_scas(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_scas(CPUX86State *env, struct x86_decode *decode)
 {
     decode->op[0].type = X86_VAR_REG;
     decode->op[0].reg = R_EAX;
@@ -641,7 +641,7 @@ static void exec_scas(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_lods_single(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_lods_single(CPUX86State *env, struct x86_decode *decode)
 {
     target_ulong addr;
     target_ulong val = 0;
@@ -653,7 +653,7 @@ static void exec_lods_single(struct CPUX86State *env, struct x86_decode *decode)
     string_increment_reg(env, R_ESI, decode);
 }
 
-static void exec_lods(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_lods(CPUX86State *env, struct x86_decode *decode)
 {
     if (decode->rep) {
         string_rep(env, decode, exec_lods_single, 0);
@@ -760,7 +760,7 @@ void simulate_rdmsr(struct CPUState *cpu)
     RDX(env) = (uint32_t)(val >> 32);
 }
 
-static void exec_rdmsr(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_rdmsr(CPUX86State *env, struct x86_decode *decode)
 {
     simulate_rdmsr(env_cpu(env));
     env->eip += decode->len;
@@ -855,7 +855,7 @@ void simulate_wrmsr(struct CPUState *cpu)
     printf("write msr %llx\n", RCX(cpu));*/
 }
 
-static void exec_wrmsr(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_wrmsr(CPUX86State *env, struct x86_decode *decode)
 {
     simulate_wrmsr(env_cpu(env));
     env->eip += decode->len;
@@ -865,7 +865,7 @@ static void exec_wrmsr(struct CPUX86State *env, struct x86_decode *decode)
  * flag:
  * 0 - bt, 1 - btc, 2 - bts, 3 - btr
  */
-static void do_bt(struct CPUX86State *env, struct x86_decode *decode, int flag)
+static void do_bt(CPUX86State *env, struct x86_decode *decode, int flag)
 {
     int32_t displacement;
     uint8_t index;
@@ -911,31 +911,31 @@ static void do_bt(struct CPUX86State *env, struct x86_decode *decode, int flag)
     set_CF(env, cf);
 }
 
-static void exec_bt(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_bt(CPUX86State *env, struct x86_decode *decode)
 {
     do_bt(env, decode, 0);
     env->eip += decode->len;
 }
 
-static void exec_btc(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_btc(CPUX86State *env, struct x86_decode *decode)
 {
     do_bt(env, decode, 1);
     env->eip += decode->len;
 }
 
-static void exec_btr(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_btr(CPUX86State *env, struct x86_decode *decode)
 {
     do_bt(env, decode, 3);
     env->eip += decode->len;
 }
 
-static void exec_bts(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_bts(CPUX86State *env, struct x86_decode *decode)
 {
     do_bt(env, decode, 2);
     env->eip += decode->len;
 }
 
-void exec_shl(struct CPUX86State *env, struct x86_decode *decode)
+void exec_shl(CPUX86State *env, struct x86_decode *decode)
 {
     uint8_t count;
     int of = 0, cf = 0;
@@ -1022,7 +1022,7 @@ void exec_movsx(CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-void exec_ror(struct CPUX86State *env, struct x86_decode *decode)
+void exec_ror(CPUX86State *env, struct x86_decode *decode)
 {
     uint8_t count;
 
@@ -1100,7 +1100,7 @@ void exec_ror(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-void exec_rol(struct CPUX86State *env, struct x86_decode *decode)
+void exec_rol(CPUX86State *env, struct x86_decode *decode)
 {
     uint8_t count;
 
@@ -1182,7 +1182,7 @@ void exec_rol(struct CPUX86State *env, struct x86_decode *decode)
 }
 
 
-void exec_rcl(struct CPUX86State *env, struct x86_decode *decode)
+void exec_rcl(CPUX86State *env, struct x86_decode *decode)
 {
     uint8_t count;
     int of = 0, cf = 0;
@@ -1267,7 +1267,7 @@ void exec_rcl(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-void exec_rcr(struct CPUX86State *env, struct x86_decode *decode)
+void exec_rcr(CPUX86State *env, struct x86_decode *decode)
 {
     uint8_t count;
     int of = 0, cf = 0;
@@ -1342,7 +1342,7 @@ void exec_rcr(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_xchg(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_xchg(CPUX86State *env, struct x86_decode *decode)
 {
     fetch_operands(env, decode, 2, true, true, false);
 
@@ -1354,7 +1354,7 @@ static void exec_xchg(struct CPUX86State *env, struct x86_decode *decode)
     env->eip += decode->len;
 }
 
-static void exec_xadd(struct CPUX86State *env, struct x86_decode *decode)
+static void exec_xadd(CPUX86State *env, struct x86_decode *decode)
 {
     EXEC_2OP_FLAGS_CMD(env, decode, +, SET_FLAGS_OSZAPC_ADD, true);
     write_val_ext(env, decode->op[1].ptr, decode->op[0].val,
@@ -1365,7 +1365,7 @@ static void exec_xadd(struct CPUX86State *env, struct x86_decode *decode)
 
 static struct cmd_handler {
     enum x86_decode_cmd cmd;
-    void (*handler)(struct CPUX86State *env, struct x86_decode *ins);
+    void (*handler)(CPUX86State *env, struct x86_decode *ins);
 } handlers[] = {
     {X86_DECODE_CMD_INVL, NULL,},
     {X86_DECODE_CMD_MOV, exec_mov},
@@ -1465,7 +1465,7 @@ void store_regs(struct CPUState *cpu)
     macvm_set_rip(cpu, env->eip);
 }
 
-bool exec_instruction(struct CPUX86State *env, struct x86_decode *ins)
+bool exec_instruction(CPUX86State *env, struct x86_decode *ins)
 {
     /*if (hvf_vcpu_id(cpu))
     printf("%d, %llx: exec_instruction %s\n", hvf_vcpu_id(cpu),  env->eip,
diff --git a/target/i386/hvf/x86_emu.h b/target/i386/hvf/x86_emu.h
index 233f7b8daa..640da90b30 100644
--- a/target/i386/hvf/x86_emu.h
+++ b/target/i386/hvf/x86_emu.h
@@ -24,7 +24,7 @@
 #include "cpu.h"
 
 void init_emu(void);
-bool exec_instruction(struct CPUX86State *env, struct x86_decode *ins);
+bool exec_instruction(CPUX86State *env, struct x86_decode *ins);
 
 void load_regs(struct CPUState *cpu);
 void store_regs(struct CPUState *cpu);
@@ -36,15 +36,15 @@ target_ulong read_reg(CPUX86State *env, int reg, int size);
 void write_reg(CPUX86State *env, int reg, target_ulong val, int size);
 target_ulong read_val_from_reg(target_ulong reg_ptr, int size);
 void write_val_to_reg(target_ulong reg_ptr, target_ulong val, int size);
-void write_val_ext(struct CPUX86State *env, target_ulong ptr, target_ulong val, int size);
-uint8_t *read_mmio(struct CPUX86State *env, target_ulong ptr, int bytes);
-target_ulong read_val_ext(struct CPUX86State *env, target_ulong ptr, int size);
+void write_val_ext(CPUX86State *env, target_ulong ptr, target_ulong val, int size);
+uint8_t *read_mmio(CPUX86State *env, target_ulong ptr, int bytes);
+target_ulong read_val_ext(CPUX86State *env, target_ulong ptr, int size);
 
-void exec_movzx(struct CPUX86State *env, struct x86_decode *decode);
-void exec_shl(struct CPUX86State *env, struct x86_decode *decode);
-void exec_movsx(struct CPUX86State *env, struct x86_decode *decode);
-void exec_ror(struct CPUX86State *env, struct x86_decode *decode);
-void exec_rol(struct CPUX86State *env, struct x86_decode *decode);
-void exec_rcl(struct CPUX86State *env, struct x86_decode *decode);
-void exec_rcr(struct CPUX86State *env, struct x86_decode *decode);
+void exec_movzx(CPUX86State *env, struct x86_decode *decode);
+void exec_shl(CPUX86State *env, struct x86_decode *decode);
+void exec_movsx(CPUX86State *env, struct x86_decode *decode);
+void exec_ror(CPUX86State *env, struct x86_decode *decode);
+void exec_rol(CPUX86State *env, struct x86_decode *decode);
+void exec_rcl(CPUX86State *env, struct x86_decode *decode);
+void exec_rcr(CPUX86State *env, struct x86_decode *decode);
 #endif
diff --git a/target/i386/nvmm/nvmm-all.c b/target/i386/nvmm/nvmm-all.c
index 9af261eea3..b97d091a50 100644
--- a/target/i386/nvmm/nvmm-all.c
+++ b/target/i386/nvmm/nvmm-all.c
@@ -85,7 +85,7 @@ nvmm_set_segment(struct nvmm_x64_state_seg *nseg, const SegmentCache *qseg)
 static void
 nvmm_set_registers(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     struct nvmm_machine *mach = get_nvmm_mach();
     struct qemu_vcpu *qcpu = get_qemu_vcpu(cpu);
     struct nvmm_vcpu *vcpu = &qcpu->vcpu;
@@ -222,7 +222,7 @@ nvmm_get_segment(SegmentCache *qseg, const struct nvmm_x64_state_seg *nseg)
 static void
 nvmm_get_registers(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     struct nvmm_machine *mach = get_nvmm_mach();
     struct qemu_vcpu *qcpu = get_qemu_vcpu(cpu);
     struct nvmm_vcpu *vcpu = &qcpu->vcpu;
@@ -347,7 +347,7 @@ nvmm_get_registers(CPUState *cpu)
 static bool
 nvmm_can_take_int(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     struct qemu_vcpu *qcpu = get_qemu_vcpu(cpu);
     struct nvmm_vcpu *vcpu = &qcpu->vcpu;
     struct nvmm_machine *mach = get_nvmm_mach();
@@ -394,7 +394,7 @@ nvmm_can_take_nmi(CPUState *cpu)
 static void
 nvmm_vcpu_pre_run(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     struct nvmm_machine *mach = get_nvmm_mach();
     struct qemu_vcpu *qcpu = get_qemu_vcpu(cpu);
     struct nvmm_vcpu *vcpu = &qcpu->vcpu;
@@ -480,7 +480,7 @@ static void
 nvmm_vcpu_post_run(CPUState *cpu, struct nvmm_vcpu_exit *exit)
 {
     struct qemu_vcpu *qcpu = get_qemu_vcpu(cpu);
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
     uint64_t tpr;
 
@@ -652,7 +652,7 @@ static int
 nvmm_handle_halted(struct nvmm_machine *mach, CPUState *cpu,
     struct nvmm_vcpu_exit *exit)
 {
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     int ret = 0;
 
     qemu_mutex_lock_iothread();
@@ -685,7 +685,7 @@ nvmm_inject_ud(struct nvmm_machine *mach, struct nvmm_vcpu *vcpu)
 static int
 nvmm_vcpu_loop(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)cpu->env_ptr;
+    CPUX86State *env = cpu->env_ptr;
     struct nvmm_machine *mach = get_nvmm_mach();
     struct qemu_vcpu *qcpu = get_qemu_vcpu(cpu);
     struct nvmm_vcpu *vcpu = &qcpu->vcpu;
diff --git a/target/i386/tcg/sysemu/excp_helper.c b/target/i386/tcg/sysemu/excp_helper.c
index 5ba739fbed..5627772e7c 100644
--- a/target/i386/tcg/sysemu/excp_helper.c
+++ b/target/i386/tcg/sysemu/excp_helper.c
@@ -19,6 +19,7 @@
 
 #include "qemu/osdep.h"
 #include "cpu.h"
+#include "exec/exec-all.h"
 #include "tcg/helper-tcg.h"
 
 int get_pg_mode(CPUX86State *env)
diff --git a/target/i386/tcg/sysemu/misc_helper.c b/target/i386/tcg/sysemu/misc_helper.c
index 9ccaa054c4..3715c1e262 100644
--- a/target/i386/tcg/sysemu/misc_helper.c
+++ b/target/i386/tcg/sysemu/misc_helper.c
@@ -23,6 +23,7 @@
 #include "exec/helper-proto.h"
 #include "exec/cpu_ldst.h"
 #include "exec/address-spaces.h"
+#include "exec/exec-all.h"
 #include "tcg/helper-tcg.h"
 
 void helper_outb(CPUX86State *env, uint32_t port, uint32_t data)
diff --git a/target/i386/whpx/whpx-accel-ops.c b/target/i386/whpx/whpx-accel-ops.c
index 6bc47c5309..1d30e4e2ed 100644
--- a/target/i386/whpx/whpx-accel-ops.c
+++ b/target/i386/whpx/whpx-accel-ops.c
@@ -83,12 +83,18 @@ static void whpx_kick_vcpu_thread(CPUState *cpu)
     }
 }
 
+static bool whpx_vcpu_thread_is_idle(CPUState *cpu)
+{
+    return !whpx_apic_in_platform();
+}
+
 static void whpx_accel_ops_class_init(ObjectClass *oc, void *data)
 {
     AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
     ops->create_vcpu_thread = whpx_start_vcpu_thread;
     ops->kick_vcpu_thread = whpx_kick_vcpu_thread;
+    ops->cpu_thread_is_idle = whpx_vcpu_thread_is_idle;
 
     ops->synchronize_post_reset = whpx_cpu_synchronize_post_reset;
     ops->synchronize_post_init = whpx_cpu_synchronize_post_init;
diff --git a/target/i386/whpx/whpx-all.c b/target/i386/whpx/whpx-all.c
index ef896da0a2..c7e25abf42 100644
--- a/target/i386/whpx/whpx-all.c
+++ b/target/i386/whpx/whpx-all.c
@@ -221,7 +221,7 @@ static SegmentCache whpx_seg_h2q(const WHV_X64_SEGMENT_REGISTER *hs)
 
 static int whpx_set_tsc(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     WHV_REGISTER_NAME tsc_reg = WHvX64RegisterTsc;
     WHV_REGISTER_VALUE tsc_val;
     HRESULT hr;
@@ -260,7 +260,7 @@ static void whpx_set_registers(CPUState *cpu, int level)
 {
     struct whpx_state *whpx = &whpx_global;
     struct whpx_vcpu *vcpu = get_whpx_vcpu(cpu);
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
     struct whpx_register_set vcxt;
     HRESULT hr;
@@ -428,7 +428,7 @@ static void whpx_set_registers(CPUState *cpu, int level)
 
 static int whpx_get_tsc(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     WHV_REGISTER_NAME tsc_reg = WHvX64RegisterTsc;
     WHV_REGISTER_VALUE tsc_val;
     HRESULT hr;
@@ -449,7 +449,7 @@ static void whpx_get_registers(CPUState *cpu)
 {
     struct whpx_state *whpx = &whpx_global;
     struct whpx_vcpu *vcpu = get_whpx_vcpu(cpu);
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
     struct whpx_register_set vcxt;
     uint64_t tpr, apic_base;
@@ -760,7 +760,7 @@ static int whpx_handle_portio(CPUState *cpu,
 
 static int whpx_handle_halt(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     int ret = 0;
 
     qemu_mutex_lock_iothread();
@@ -781,7 +781,7 @@ static void whpx_vcpu_pre_run(CPUState *cpu)
     HRESULT hr;
     struct whpx_state *whpx = &whpx_global;
     struct whpx_vcpu *vcpu = get_whpx_vcpu(cpu);
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
     int irq;
     uint8_t tpr;
@@ -903,7 +903,7 @@ static void whpx_vcpu_pre_run(CPUState *cpu)
 static void whpx_vcpu_post_run(CPUState *cpu)
 {
     struct whpx_vcpu *vcpu = get_whpx_vcpu(cpu);
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
 
     env->eflags = vcpu->exit_ctx.VpContext.Rflags;
@@ -927,7 +927,7 @@ static void whpx_vcpu_post_run(CPUState *cpu)
 
 static void whpx_vcpu_process_async_events(CPUState *cpu)
 {
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
     struct whpx_vcpu *vcpu = get_whpx_vcpu(cpu);
 
@@ -1333,7 +1333,7 @@ int whpx_init_vcpu(CPUState *cpu)
     struct whpx_state *whpx = &whpx_global;
     struct whpx_vcpu *vcpu = NULL;
     Error *local_error = NULL;
-    struct CPUX86State *env = (CPUArchState *)(cpu->env_ptr);
+    CPUX86State *env = cpu->env_ptr;
     X86CPU *x86_cpu = X86_CPU(cpu);
     UINT64 freq = 0;
     int ret;
diff --git a/target/m68k/cpu-qom.h b/target/m68k/cpu-qom.h
index 1ceb160ecb..cd9687192c 100644
--- a/target/m68k/cpu-qom.h
+++ b/target/m68k/cpu-qom.h
@@ -25,8 +25,7 @@
 
 #define TYPE_M68K_CPU "m68k-cpu"
 
-OBJECT_DECLARE_TYPE(M68kCPU, M68kCPUClass,
-                    M68K_CPU)
+OBJECT_DECLARE_CPU_TYPE(M68kCPU, M68kCPUClass, M68K_CPU)
 
 /*
  * M68kCPUClass:
diff --git a/target/m68k/cpu.h b/target/m68k/cpu.h
index a3423729ef..872e8ce637 100644
--- a/target/m68k/cpu.h
+++ b/target/m68k/cpu.h
@@ -79,7 +79,7 @@
 
 typedef CPU_LDoubleU FPReg;
 
-typedef struct CPUM68KState {
+typedef struct CPUArchState {
     uint32_t dregs[8];
     uint32_t aregs[8];
     uint32_t pc;
@@ -156,7 +156,7 @@ typedef struct CPUM68KState {
  *
  * A Motorola 68k CPU.
  */
-struct M68kCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -574,9 +574,6 @@ void m68k_cpu_transaction_failed(CPUState *cs, hwaddr physaddr, vaddr addr,
                                  int mmu_idx, MemTxAttrs attrs,
                                  MemTxResult response, uintptr_t retaddr);
 
-typedef CPUM68KState CPUArchState;
-typedef M68kCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 /* TB flags */
diff --git a/target/microblaze/cpu-qom.h b/target/microblaze/cpu-qom.h
index e520eefb12..255b39a45d 100644
--- a/target/microblaze/cpu-qom.h
+++ b/target/microblaze/cpu-qom.h
@@ -25,8 +25,7 @@
 
 #define TYPE_MICROBLAZE_CPU "microblaze-cpu"
 
-OBJECT_DECLARE_TYPE(MicroBlazeCPU, MicroBlazeCPUClass,
-                    MICROBLAZE_CPU)
+OBJECT_DECLARE_CPU_TYPE(MicroBlazeCPU, MicroBlazeCPUClass, MICROBLAZE_CPU)
 
 /**
  * MicroBlazeCPUClass:
diff --git a/target/microblaze/cpu.h b/target/microblaze/cpu.h
index e9cd0b88de..0a0ce71b6a 100644
--- a/target/microblaze/cpu.h
+++ b/target/microblaze/cpu.h
@@ -24,7 +24,7 @@
 #include "exec/cpu-defs.h"
 #include "fpu/softfloat-types.h"
 
-typedef struct CPUMBState CPUMBState;
+typedef struct CPUArchState CPUMBState;
 #if !defined(CONFIG_USER_ONLY)
 #include "mmu.h"
 #endif
@@ -239,7 +239,7 @@ typedef struct CPUMBState CPUMBState;
 #define USE_NON_SECURE_M_AXI_DC_MASK 0x4
 #define USE_NON_SECURE_M_AXI_IC_MASK 0x8
 
-struct CPUMBState {
+struct CPUArchState {
     uint32_t bvalue;   /* TCG temporary, only valid during a TB */
     uint32_t btarget;  /* Full resolved branch destination */
 
@@ -339,7 +339,7 @@ typedef struct {
  *
  * A MicroBlaze CPU.
  */
-struct MicroBlazeCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
 
@@ -394,9 +394,6 @@ void mb_tcg_init(void);
 #define MMU_USER_IDX    2
 /* See NB_MMU_MODES further up the file.  */
 
-typedef CPUMBState CPUArchState;
-typedef MicroBlazeCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 /* Ensure there is no overlap between the two masks. */
diff --git a/target/microblaze/mmu.h b/target/microblaze/mmu.h
index b6b4b9ad60..1068bd2d52 100644
--- a/target/microblaze/mmu.h
+++ b/target/microblaze/mmu.h
@@ -20,6 +20,8 @@
 #ifndef TARGET_MICROBLAZE_MMU_H
 #define TARGET_MICROBLAZE_MMU_H
 
+#include "cpu.h"
+
 #define MMU_R_PID    0
 #define MMU_R_ZPR    1
 #define MMU_R_TLBX   2
diff --git a/target/mips/cpu-qom.h b/target/mips/cpu-qom.h
index dda0c911fa..e28b529607 100644
--- a/target/mips/cpu-qom.h
+++ b/target/mips/cpu-qom.h
@@ -29,8 +29,7 @@
 #define TYPE_MIPS_CPU "mips-cpu"
 #endif
 
-OBJECT_DECLARE_TYPE(MIPSCPU, MIPSCPUClass,
-                    MIPS_CPU)
+OBJECT_DECLARE_CPU_TYPE(MIPSCPU, MIPSCPUClass, MIPS_CPU)
 
 /**
  * MIPSCPUClass:
diff --git a/target/mips/cpu.h b/target/mips/cpu.h
index 56b1cbd091..09e98f64de 100644
--- a/target/mips/cpu.h
+++ b/target/mips/cpu.h
@@ -524,8 +524,7 @@ struct TCState {
 };
 
 struct MIPSITUState;
-typedef struct CPUMIPSState CPUMIPSState;
-struct CPUMIPSState {
+typedef struct CPUArchState {
     TCState active_tc;
     CPUMIPSFPUContext active_fpu;
 
@@ -1161,7 +1160,7 @@ struct CPUMIPSState {
     QEMUTimer *timer; /* Internal timer */
     target_ulong exception_base; /* ExceptionBase input to the core */
     uint64_t cp0_count_ns; /* CP0_Count clock period (in nanoseconds) */
-};
+} CPUMIPSState;
 
 /**
  * MIPSCPU:
@@ -1172,7 +1171,7 @@ struct CPUMIPSState {
  *
  * A MIPS CPU.
  */
-struct MIPSCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -1218,9 +1217,6 @@ static inline int cpu_mmu_index(CPUMIPSState *env, bool ifetch)
     return hflags_mmu_index(env->hflags);
 }
 
-typedef CPUMIPSState CPUArchState;
-typedef MIPSCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 /* Exceptions */
diff --git a/target/mips/internal.h b/target/mips/internal.h
index daddb05fd4..ac6e03e2f2 100644
--- a/target/mips/internal.h
+++ b/target/mips/internal.h
@@ -12,6 +12,7 @@
 #ifdef CONFIG_TCG
 #include "tcg/tcg-internal.h"
 #endif
+#include "cpu.h"
 
 /*
  * MMU types, the first four entries have the same layout as the
@@ -133,14 +134,14 @@ struct r4k_tlb_t {
 struct CPUMIPSTLBContext {
     uint32_t nb_tlb;
     uint32_t tlb_in_use;
-    int (*map_address)(struct CPUMIPSState *env, hwaddr *physical, int *prot,
+    int (*map_address)(CPUMIPSState *env, hwaddr *physical, int *prot,
                        target_ulong address, MMUAccessType access_type);
-    void (*helper_tlbwi)(struct CPUMIPSState *env);
-    void (*helper_tlbwr)(struct CPUMIPSState *env);
-    void (*helper_tlbp)(struct CPUMIPSState *env);
-    void (*helper_tlbr)(struct CPUMIPSState *env);
-    void (*helper_tlbinv)(struct CPUMIPSState *env);
-    void (*helper_tlbinvf)(struct CPUMIPSState *env);
+    void (*helper_tlbwi)(CPUMIPSState *env);
+    void (*helper_tlbwr)(CPUMIPSState *env);
+    void (*helper_tlbp)(CPUMIPSState *env);
+    void (*helper_tlbr)(CPUMIPSState *env);
+    void (*helper_tlbinv)(CPUMIPSState *env);
+    void (*helper_tlbinvf)(CPUMIPSState *env);
     union {
         struct {
             r4k_tlb_t tlb[MIPS_TLB_MAX];
diff --git a/target/nios2/cpu.h b/target/nios2/cpu.h
index a00e4229ce..ca0f3420cd 100644
--- a/target/nios2/cpu.h
+++ b/target/nios2/cpu.h
@@ -25,15 +25,14 @@
 #include "hw/core/cpu.h"
 #include "qom/object.h"
 
-typedef struct CPUNios2State CPUNios2State;
+typedef struct CPUArchState CPUNios2State;
 #if !defined(CONFIG_USER_ONLY)
 #include "mmu.h"
 #endif
 
 #define TYPE_NIOS2_CPU "nios2-cpu"
 
-OBJECT_DECLARE_TYPE(Nios2CPU, Nios2CPUClass,
-                    NIOS2_CPU)
+OBJECT_DECLARE_CPU_TYPE(Nios2CPU, Nios2CPUClass, NIOS2_CPU)
 
 /**
  * Nios2CPUClass:
@@ -155,7 +154,7 @@ struct Nios2CPUClass {
 
 #define CPU_INTERRUPT_NMI       CPU_INTERRUPT_TGT_EXT_3
 
-struct CPUNios2State {
+struct CPUArchState {
     uint32_t regs[NUM_CORE_REGS];
 
 #if !defined(CONFIG_USER_ONLY)
@@ -170,7 +169,7 @@ struct CPUNios2State {
  *
  * A Nios2 CPU.
  */
-struct Nios2CPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
diff --git a/target/nios2/mmu.h b/target/nios2/mmu.h
index b7785b46c0..5b085900fb 100644
--- a/target/nios2/mmu.h
+++ b/target/nios2/mmu.h
@@ -21,6 +21,8 @@
 #ifndef NIOS2_MMU_H
 #define NIOS2_MMU_H
 
+#include "cpu.h"
+
 typedef struct Nios2TLBEntry {
     target_ulong tag;
     target_ulong data;
diff --git a/target/openrisc/cpu.h b/target/openrisc/cpu.h
index ee069b080c..bdf29d2dc4 100644
--- a/target/openrisc/cpu.h
+++ b/target/openrisc/cpu.h
@@ -24,13 +24,9 @@
 #include "hw/core/cpu.h"
 #include "qom/object.h"
 
-/* cpu_openrisc_map_address_* in CPUOpenRISCTLBContext need this decl.  */
-struct OpenRISCCPU;
-
 #define TYPE_OPENRISC_CPU "or1k-cpu"
 
-OBJECT_DECLARE_TYPE(OpenRISCCPU, OpenRISCCPUClass,
-                    OPENRISC_CPU)
+OBJECT_DECLARE_CPU_TYPE(OpenRISCCPU, OpenRISCCPUClass, OPENRISC_CPU)
 
 /**
  * OpenRISCCPUClass:
@@ -231,18 +227,18 @@ typedef struct CPUOpenRISCTLBContext {
     OpenRISCTLBEntry itlb[TLB_SIZE];
     OpenRISCTLBEntry dtlb[TLB_SIZE];
 
-    int (*cpu_openrisc_map_address_code)(struct OpenRISCCPU *cpu,
+    int (*cpu_openrisc_map_address_code)(OpenRISCCPU *cpu,
                                          hwaddr *physical,
                                          int *prot,
                                          target_ulong address, int rw);
-    int (*cpu_openrisc_map_address_data)(struct OpenRISCCPU *cpu,
+    int (*cpu_openrisc_map_address_data)(OpenRISCCPU *cpu,
                                          hwaddr *physical,
                                          int *prot,
                                          target_ulong address, int rw);
 } CPUOpenRISCTLBContext;
 #endif
 
-typedef struct CPUOpenRISCState {
+typedef struct CPUArchState {
     target_ulong shadow_gpr[16][32]; /* Shadow registers */
 
     target_ulong pc;          /* Program counter */
@@ -301,7 +297,7 @@ typedef struct CPUOpenRISCState {
  *
  * A OpenRISC CPU.
  */
-struct OpenRISCCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -348,9 +344,6 @@ void cpu_openrisc_count_stop(OpenRISCCPU *cpu);
 #define OPENRISC_CPU_TYPE_NAME(model) model OPENRISC_CPU_TYPE_SUFFIX
 #define CPU_RESOLVING_TYPE TYPE_OPENRISC_CPU
 
-typedef CPUOpenRISCState CPUArchState;
-typedef OpenRISCCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 #define TB_FLAGS_SM    SR_SM
diff --git a/target/ppc/cpu-qom.h b/target/ppc/cpu-qom.h
index 98facee9fa..ad7e3c3db9 100644
--- a/target/ppc/cpu-qom.h
+++ b/target/ppc/cpu-qom.h
@@ -29,10 +29,9 @@
 #define TYPE_POWERPC_CPU "powerpc-cpu"
 #endif
 
-OBJECT_DECLARE_TYPE(PowerPCCPU, PowerPCCPUClass,
-                    POWERPC_CPU)
+OBJECT_DECLARE_CPU_TYPE(PowerPCCPU, PowerPCCPUClass, POWERPC_CPU)
 
-typedef struct CPUPPCState CPUPPCState;
+typedef struct CPUArchState CPUPPCState;
 typedef struct ppc_tb_t ppc_tb_t;
 typedef struct ppc_dcr_t ppc_dcr_t;
 
diff --git a/target/ppc/cpu.h b/target/ppc/cpu.h
index 1b687521c7..047b24ba50 100644
--- a/target/ppc/cpu.h
+++ b/target/ppc/cpu.h
@@ -1077,7 +1077,7 @@ struct ppc_radix_page_info {
 #define PPC_CPU_OPCODES_LEN          0x40
 #define PPC_CPU_INDIRECT_OPCODES_LEN 0x20
 
-struct CPUPPCState {
+struct CPUArchState {
     /* Most commonly used resources during translated code execution first */
     target_ulong gpr[32];  /* general purpose registers */
     target_ulong gprh[32]; /* storage for GPR MSB, used by the SPE extension */
@@ -1275,7 +1275,7 @@ typedef struct PPCVirtualHypervisorClass PPCVirtualHypervisorClass;
  *
  * A PowerPC CPU.
  */
-struct PowerPCCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -1477,9 +1477,6 @@ void ppc_compat_add_property(Object *obj, const char *name,
                              uint32_t *compat_pvr, const char *basedesc);
 #endif /* defined(TARGET_PPC64) */
 
-typedef CPUPPCState CPUArchState;
-typedef PowerPCCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 /*****************************************************************************/
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
index 9ba05042ed..c069fe85fa 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -98,7 +98,7 @@ enum {
 
 #define MAX_RISCV_PMPS (16)
 
-typedef struct CPURISCVState CPURISCVState;
+typedef struct CPUArchState CPURISCVState;
 
 #if !defined(CONFIG_USER_ONLY)
 #include "pmp.h"
@@ -113,7 +113,7 @@ FIELD(VTYPE, VMA, 7, 1)
 FIELD(VTYPE, VEDIV, 8, 2)
 FIELD(VTYPE, RESERVED, 10, sizeof(target_ulong) * 8 - 11)
 
-struct CPURISCVState {
+struct CPUArchState {
     target_ulong gpr[32];
     target_ulong gprh[32]; /* 64 top bits of the 128-bit registers */
     uint64_t fpr[32]; /* assume both F and D extensions */
@@ -320,8 +320,7 @@ struct CPURISCVState {
     uint64_t kvm_timer_frequency;
 };
 
-OBJECT_DECLARE_TYPE(RISCVCPU, RISCVCPUClass,
-                    RISCV_CPU)
+OBJECT_DECLARE_CPU_TYPE(RISCVCPU, RISCVCPUClass, RISCV_CPU)
 
 /**
  * RISCVCPUClass:
@@ -395,7 +394,7 @@ typedef struct RISCVCPUConfig RISCVCPUConfig;
  *
  * A RISCV CPU.
  */
-struct RISCVCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -499,8 +498,6 @@ void riscv_cpu_set_fflags(CPURISCVState *env, target_ulong);
 #define TB_FLAGS_MSTATUS_FS MSTATUS_FS
 #define TB_FLAGS_MSTATUS_VS MSTATUS_VS
 
-typedef CPURISCVState CPUArchState;
-typedef RISCVCPU ArchCPU;
 #include "exec/cpu-all.h"
 
 FIELD(TB_FLAGS, MEM_IDX, 0, 3)
diff --git a/target/riscv/csr.c b/target/riscv/csr.c
index aea82dff4a..0606cd0ea8 100644
--- a/target/riscv/csr.c
+++ b/target/riscv/csr.c
@@ -23,6 +23,7 @@
 #include "cpu.h"
 #include "qemu/main-loop.h"
 #include "exec/exec-all.h"
+#include "sysemu/cpu-timers.h"
 
 /* CSR function table public API */
 void riscv_get_csr_ops(int csrno, riscv_csr_operations *ops)
diff --git a/target/riscv/pmp.h b/target/riscv/pmp.h
index a9a0b363a7..fcb6b7c467 100644
--- a/target/riscv/pmp.h
+++ b/target/riscv/pmp.h
@@ -22,6 +22,8 @@
 #ifndef RISCV_PMP_H
 #define RISCV_PMP_H
 
+#include "cpu.h"
+
 typedef enum {
     PMP_READ  = 1 << 0,
     PMP_WRITE = 1 << 1,
diff --git a/target/rx/cpu-qom.h b/target/rx/cpu-qom.h
index 7310558e0c..4533759d96 100644
--- a/target/rx/cpu-qom.h
+++ b/target/rx/cpu-qom.h
@@ -26,8 +26,7 @@
 
 #define TYPE_RX62N_CPU RX_CPU_TYPE_NAME("rx62n")
 
-OBJECT_DECLARE_TYPE(RXCPU, RXCPUClass,
-                    RX_CPU)
+OBJECT_DECLARE_CPU_TYPE(RXCPU, RXCPUClass, RX_CPU)
 
 /*
  * RXCPUClass:
@@ -45,6 +44,4 @@ struct RXCPUClass {
     DeviceReset parent_reset;
 };
 
-#define CPUArchState struct CPURXState
-
 #endif
diff --git a/target/rx/cpu.h b/target/rx/cpu.h
index 58adf9edf6..b4abd90ccd 100644
--- a/target/rx/cpu.h
+++ b/target/rx/cpu.h
@@ -65,7 +65,7 @@ enum {
     NUM_REGS = 16,
 };
 
-typedef struct CPURXState {
+typedef struct CPUArchState {
     /* CPU registers */
     uint32_t regs[NUM_REGS];    /* general registers */
     uint32_t psw_o;             /* O bit of status register */
@@ -105,7 +105,7 @@ typedef struct CPURXState {
  *
  * A RX CPU
  */
-struct RXCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -114,8 +114,6 @@ struct RXCPU {
     CPURXState env;
 };
 
-typedef RXCPU ArchCPU;
-
 #define RX_CPU_TYPE_SUFFIX "-" TYPE_RX_CPU
 #define RX_CPU_TYPE_NAME(model) model RX_CPU_TYPE_SUFFIX
 #define CPU_RESOLVING_TYPE TYPE_RX_CPU
diff --git a/target/s390x/cpu-qom.h b/target/s390x/cpu-qom.h
index 9f3a0d86c5..00cae2b131 100644
--- a/target/s390x/cpu-qom.h
+++ b/target/s390x/cpu-qom.h
@@ -25,12 +25,13 @@
 
 #define TYPE_S390_CPU "s390x-cpu"
 
-OBJECT_DECLARE_TYPE(S390CPU, S390CPUClass,
-                    S390_CPU)
+OBJECT_DECLARE_CPU_TYPE(S390CPU, S390CPUClass, S390_CPU)
 
 typedef struct S390CPUModel S390CPUModel;
 typedef struct S390CPUDef S390CPUDef;
 
+typedef struct CPUArchState CPUS390XState;
+
 typedef enum cpu_reset_type {
     S390_CPU_RESET_NORMAL,
     S390_CPU_RESET_INITIAL,
@@ -63,6 +64,4 @@ struct S390CPUClass {
     void (*reset)(CPUState *cpu, cpu_reset_type type);
 };
 
-typedef struct CPUS390XState CPUS390XState;
-
 #endif
diff --git a/target/s390x/cpu.h b/target/s390x/cpu.h
index a75e559134..c49c8466e7 100644
--- a/target/s390x/cpu.h
+++ b/target/s390x/cpu.h
@@ -51,7 +51,7 @@ typedef struct PSW {
     uint64_t addr;
 } PSW;
 
-struct CPUS390XState {
+struct CPUArchState {
     uint64_t regs[16];     /* GP registers */
     /*
      * The floating point registers are part of the vector registers.
@@ -163,7 +163,7 @@ static inline uint64_t *get_freg(CPUS390XState *cs, int nr)
  *
  * An S/390 CPU.
  */
-struct S390CPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -840,9 +840,6 @@ uint64_t s390_cpu_get_psw_mask(CPUS390XState *env);
 /* outside of target/s390x/ */
 S390CPU *s390_cpu_addr2state(uint16_t cpu_addr);
 
-typedef CPUS390XState CPUArchState;
-typedef S390CPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 #endif
diff --git a/target/sh4/cpu-qom.h b/target/sh4/cpu-qom.h
index 8903b4b9c7..d4192d1090 100644
--- a/target/sh4/cpu-qom.h
+++ b/target/sh4/cpu-qom.h
@@ -29,8 +29,7 @@
 #define TYPE_SH7751R_CPU SUPERH_CPU_TYPE_NAME("sh7751r")
 #define TYPE_SH7785_CPU  SUPERH_CPU_TYPE_NAME("sh7785")
 
-OBJECT_DECLARE_TYPE(SuperHCPU, SuperHCPUClass,
-                    SUPERH_CPU)
+OBJECT_DECLARE_CPU_TYPE(SuperHCPU, SuperHCPUClass, SUPERH_CPU)
 
 /**
  * SuperHCPUClass:
diff --git a/target/sh4/cpu.h b/target/sh4/cpu.h
index fb9dd9db2f..c72a30edfd 100644
--- a/target/sh4/cpu.h
+++ b/target/sh4/cpu.h
@@ -130,7 +130,7 @@ typedef struct memory_content {
     struct memory_content *next;
 } memory_content;
 
-typedef struct CPUSH4State {
+typedef struct CPUArchState {
     uint32_t flags;		/* general execution flags */
     uint32_t gregs[24];		/* general registers */
     float32 fregs[32];		/* floating point registers */
@@ -195,7 +195,7 @@ typedef struct CPUSH4State {
  *
  * A SuperH CPU.
  */
-struct SuperHCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -264,9 +264,6 @@ static inline int cpu_mmu_index (CPUSH4State *env, bool ifetch)
     }
 }
 
-typedef CPUSH4State CPUArchState;
-typedef SuperHCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 /* MMU control register */
diff --git a/target/sparc/cpu-qom.h b/target/sparc/cpu-qom.h
index f33949aaee..86ed37d933 100644
--- a/target/sparc/cpu-qom.h
+++ b/target/sparc/cpu-qom.h
@@ -29,8 +29,7 @@
 #define TYPE_SPARC_CPU "sparc-cpu"
 #endif
 
-OBJECT_DECLARE_TYPE(SPARCCPU, SPARCCPUClass,
-                    SPARC_CPU)
+OBJECT_DECLARE_CPU_TYPE(SPARCCPU, SPARCCPUClass, SPARC_CPU)
 
 typedef struct sparc_def_t sparc_def_t;
 /**
diff --git a/target/sparc/cpu.h b/target/sparc/cpu.h
index 5a7f1ed5d6..abb38db674 100644
--- a/target/sparc/cpu.h
+++ b/target/sparc/cpu.h
@@ -420,7 +420,7 @@ struct CPUTimer
 
 typedef struct CPUTimer CPUTimer;
 
-typedef struct CPUSPARCState CPUSPARCState;
+typedef struct CPUArchState CPUSPARCState;
 #if defined(TARGET_SPARC64)
 typedef union {
    uint64_t mmuregs[16];
@@ -439,7 +439,7 @@ typedef union {
    };
 } SparcV9MMU;
 #endif
-struct CPUSPARCState {
+struct CPUArchState {
     target_ulong gregs[8]; /* general registers */
     target_ulong *regwptr; /* pointer to current register window */
     target_ulong pc;       /* program counter */
@@ -556,7 +556,7 @@ struct CPUSPARCState {
  *
  * A SPARC CPU.
  */
-struct SPARCCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -743,9 +743,6 @@ static inline int cpu_pil_allowed(CPUSPARCState *env1, int pil)
 #endif
 }
 
-typedef CPUSPARCState CPUArchState;
-typedef SPARCCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 #ifdef TARGET_SPARC64
diff --git a/target/tricore/cpu-qom.h b/target/tricore/cpu-qom.h
index 59bfd01bbc..ee24e9fa76 100644
--- a/target/tricore/cpu-qom.h
+++ b/target/tricore/cpu-qom.h
@@ -24,8 +24,7 @@
 
 #define TYPE_TRICORE_CPU "tricore-cpu"
 
-OBJECT_DECLARE_TYPE(TriCoreCPU, TriCoreCPUClass,
-                    TRICORE_CPU)
+OBJECT_DECLARE_CPU_TYPE(TriCoreCPU, TriCoreCPUClass, TRICORE_CPU)
 
 struct TriCoreCPUClass {
     /*< private >*/
diff --git a/target/tricore/cpu.h b/target/tricore/cpu.h
index c461387e71..108d6b8288 100644
--- a/target/tricore/cpu.h
+++ b/target/tricore/cpu.h
@@ -28,8 +28,7 @@ struct tricore_boot_info;
 
 typedef struct tricore_def_t tricore_def_t;
 
-typedef struct CPUTriCoreState CPUTriCoreState;
-struct CPUTriCoreState {
+typedef struct CPUArchState {
     /* GPR Register */
     uint32_t gpr_a[16];
     uint32_t gpr_d[16];
@@ -189,7 +188,7 @@ struct CPUTriCoreState {
     const tricore_def_t *cpu_model;
     void *irq[8];
     struct QEMUTimer *timer; /* Internal timer */
-};
+} CPUTriCoreState;
 
 /**
  * TriCoreCPU:
@@ -197,7 +196,7 @@ struct CPUTriCoreState {
  *
  * A TriCore CPU.
  */
-struct TriCoreCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -369,9 +368,6 @@ static inline int cpu_mmu_index(CPUTriCoreState *env, bool ifetch)
     return 0;
 }
 
-typedef CPUTriCoreState CPUArchState;
-typedef TriCoreCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 void cpu_state_reset(CPUTriCoreState *s);
diff --git a/target/xtensa/cpu-qom.h b/target/xtensa/cpu-qom.h
index 41d9859673..4fc35ee49b 100644
--- a/target/xtensa/cpu-qom.h
+++ b/target/xtensa/cpu-qom.h
@@ -34,8 +34,7 @@
 
 #define TYPE_XTENSA_CPU "xtensa-cpu"
 
-OBJECT_DECLARE_TYPE(XtensaCPU, XtensaCPUClass,
-                    XTENSA_CPU)
+OBJECT_DECLARE_CPU_TYPE(XtensaCPU, XtensaCPUClass, XTENSA_CPU)
 
 typedef struct XtensaConfig XtensaConfig;
 
diff --git a/target/xtensa/cpu.h b/target/xtensa/cpu.h
index 02143f2f77..4515f682aa 100644
--- a/target/xtensa/cpu.h
+++ b/target/xtensa/cpu.h
@@ -306,7 +306,7 @@ typedef enum {
     INTTYPE_MAX
 } interrupt_type;
 
-struct CPUXtensaState;
+typedef struct CPUArchState CPUXtensaState;
 
 typedef struct xtensa_tlb_entry {
     uint32_t vaddr;
@@ -344,7 +344,7 @@ typedef struct XtensaGdbRegmap {
 } XtensaGdbRegmap;
 
 typedef struct XtensaCcompareTimer {
-    struct CPUXtensaState *env;
+    CPUXtensaState *env;
     QEMUTimer *timer;
 } XtensaCcompareTimer;
 
@@ -506,7 +506,7 @@ enum {
 };
 #endif
 
-typedef struct CPUXtensaState {
+struct CPUArchState {
     const XtensaConfig *config;
     uint32_t regs[16];
     uint32_t pc;
@@ -545,7 +545,7 @@ typedef struct CPUXtensaState {
 
     /* Watchpoints for DBREAK registers */
     struct CPUWatchpoint *cpu_watchpoint[MAX_NDBREAK];
-} CPUXtensaState;
+};
 
 /**
  * XtensaCPU:
@@ -553,7 +553,7 @@ typedef struct CPUXtensaState {
  *
  * An Xtensa CPU.
  */
-struct XtensaCPU {
+struct ArchCPU {
     /*< private >*/
     CPUState parent_obj;
     /*< public >*/
@@ -722,9 +722,6 @@ static inline int cpu_mmu_index(CPUXtensaState *env, bool ifetch)
 #define XTENSA_CSBASE_LBEG_OFF_MASK 0x00ff0000
 #define XTENSA_CSBASE_LBEG_OFF_SHIFT 16
 
-typedef CPUXtensaState CPUArchState;
-typedef XtensaCPU ArchCPU;
-
 #include "exec/cpu-all.h"
 
 static inline void cpu_get_tb_cpu_state(CPUXtensaState *env, target_ulong *pc,
diff --git a/tests/unit/ptimer-test-stubs.c b/tests/unit/ptimer-test-stubs.c
index 2a3ef58799..f5e75a96b6 100644
--- a/tests/unit/ptimer-test-stubs.c
+++ b/tests/unit/ptimer-test-stubs.c
@@ -12,7 +12,6 @@
 #include "qemu/main-loop.h"
 #include "sysemu/replay.h"
 #include "migration/vmstate.h"
-#include "sysemu/cpu-timers.h"
 
 #include "ptimer-test.h"