26 files changed, 94 insertions, 112 deletions

diff --git a/cpus.c b/cpus.c
index d74cc117b3..363d392cd9 100644
--- a/cpus.c
+++ b/cpus.c
@@ -86,7 +86,7 @@ static bool all_cpu_threads_idle(void)
 {
     CPUState *cpu;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         if (!cpu_thread_is_idle(cpu)) {
             return false;
         }
@@ -416,7 +416,7 @@ void hw_error(const char *fmt, ...)
     fprintf(stderr, "qemu: hardware error: ");
     vfprintf(stderr, fmt, ap);
     fprintf(stderr, "\n");
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         fprintf(stderr, "CPU #%d:\n", cpu->cpu_index);
         cpu_dump_state(cpu, stderr, fprintf, CPU_DUMP_FPU);
     }
@@ -428,7 +428,7 @@ void cpu_synchronize_all_states(void)
 {
     CPUState *cpu;
 
-    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         cpu_synchronize_state(cpu);
     }
 }
@@ -437,7 +437,7 @@ void cpu_synchronize_all_post_reset(void)
 {
     CPUState *cpu;
 
-    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         cpu_synchronize_post_reset(cpu);
     }
 }
@@ -446,7 +446,7 @@ void cpu_synchronize_all_post_init(void)
 {
     CPUState *cpu;
 
-    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         cpu_synchronize_post_init(cpu);
     }
 }
@@ -760,7 +760,7 @@ static void qemu_tcg_wait_io_event(void)
         qemu_cond_wait(&qemu_io_proceeded_cond, &qemu_global_mutex);
     }
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         qemu_wait_io_event_common(cpu);
     }
 }
@@ -872,11 +872,11 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
     qemu_cond_signal(&qemu_cpu_cond);
 
     /* wait for initial kick-off after machine start */
-    while (first_cpu->stopped) {
+    while (QTAILQ_FIRST(&cpus)->stopped) {
         qemu_cond_wait(tcg_halt_cond, &qemu_global_mutex);
 
         /* process any pending work */
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             qemu_wait_io_event_common(cpu);
         }
     }
@@ -991,13 +991,12 @@ void qemu_mutex_unlock_iothread(void)
 
 static int all_vcpus_paused(void)
 {
-    CPUState *cpu = first_cpu;
+    CPUState *cpu;
 
-    while (cpu) {
+    CPU_FOREACH(cpu) {
         if (!cpu->stopped) {
             return 0;
         }
-        cpu = cpu->next_cpu;
     }
 
     return 1;
@@ -1005,23 +1004,20 @@ static int all_vcpus_paused(void)
 
 void pause_all_vcpus(void)
 {
-    CPUState *cpu = first_cpu;
+    CPUState *cpu;
 
     qemu_clock_enable(QEMU_CLOCK_VIRTUAL, false);
-    while (cpu) {
+    CPU_FOREACH(cpu) {
         cpu->stop = true;
         qemu_cpu_kick(cpu);
-        cpu = cpu->next_cpu;
     }
 
     if (qemu_in_vcpu_thread()) {
         cpu_stop_current();
         if (!kvm_enabled()) {
-            cpu = first_cpu;
-            while (cpu) {
+            CPU_FOREACH(cpu) {
                 cpu->stop = false;
                 cpu->stopped = true;
-                cpu = cpu->next_cpu;
             }
             return;
         }
@@ -1029,10 +1025,8 @@ void pause_all_vcpus(void)
 
     while (!all_vcpus_paused()) {
         qemu_cond_wait(&qemu_pause_cond, &qemu_global_mutex);
-        cpu = first_cpu;
-        while (cpu) {
+        CPU_FOREACH(cpu) {
             qemu_cpu_kick(cpu);
-            cpu = cpu->next_cpu;
         }
     }
 }
@@ -1046,12 +1040,11 @@ void cpu_resume(CPUState *cpu)
 
 void resume_all_vcpus(void)
 {
-    CPUState *cpu = first_cpu;
+    CPUState *cpu;
 
     qemu_clock_enable(QEMU_CLOCK_VIRTUAL, true);
-    while (cpu) {
+    CPU_FOREACH(cpu) {
         cpu_resume(cpu);
-        cpu = cpu->next_cpu;
     }
 }
 
@@ -1215,7 +1208,7 @@ static void tcg_exec_all(void)
     if (next_cpu == NULL) {
         next_cpu = first_cpu;
     }
-    for (; next_cpu != NULL && !exit_request; next_cpu = next_cpu->next_cpu) {
+    for (; next_cpu != NULL && !exit_request; next_cpu = CPU_NEXT(next_cpu)) {
         CPUState *cpu = next_cpu;
         CPUArchState *env = cpu->env_ptr;
 
@@ -1240,7 +1233,7 @@ void set_numa_modes(void)
     CPUState *cpu;
     int i;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         for (i = 0; i < nb_numa_nodes; i++) {
             if (test_bit(cpu->cpu_index, node_cpumask[i])) {
                 cpu->numa_node = i;
@@ -1262,7 +1255,7 @@ CpuInfoList *qmp_query_cpus(Error **errp)
     CpuInfoList *head = NULL, *cur_item = NULL;
     CPUState *cpu;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         CpuInfoList *info;
 #if defined(TARGET_I386)
         X86CPU *x86_cpu = X86_CPU(cpu);
@@ -1391,7 +1384,7 @@ void qmp_inject_nmi(Error **errp)
 #if defined(TARGET_I386)
     CPUState *cs;
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         X86CPU *cpu = X86_CPU(cs);
         CPUX86State *env = &cpu->env;
 
@@ -1405,7 +1398,7 @@ void qmp_inject_nmi(Error **errp)
     CPUState *cs;
     S390CPU *cpu;
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         cpu = S390_CPU(cs);
         if (cpu->env.cpu_num == monitor_get_cpu_index()) {
             if (s390_cpu_restart(S390_CPU(cs)) == -1) {
diff --git a/cputlb.c b/cputlb.c
index 977c0ca59d..19ecf60983 100644
--- a/cputlb.c
+++ b/cputlb.c
@@ -189,7 +189,7 @@ void cpu_tlb_reset_dirty_all(ram_addr_t start1, ram_addr_t length)
     CPUState *cpu;
     CPUArchState *env;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         int mmu_idx;
 
         env = cpu->env_ptr;
diff --git a/dump.c b/dump.c
index c0dae2c3ff..846155cbc9 100644
--- a/dump.c
+++ b/dump.c
@@ -277,7 +277,7 @@ static int write_elf64_notes(DumpState *s)
     int ret;
     int id;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         id = cpu_index(cpu);
         ret = cpu_write_elf64_note(fd_write_vmcore, cpu, id, s);
         if (ret < 0) {
@@ -286,7 +286,7 @@ static int write_elf64_notes(DumpState *s)
         }
     }
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         ret = cpu_write_elf64_qemunote(fd_write_vmcore, cpu, s);
         if (ret < 0) {
             dump_error(s, "dump: failed to write CPU status.\n");
@@ -327,7 +327,7 @@ static int write_elf32_notes(DumpState *s)
     int ret;
     int id;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         id = cpu_index(cpu);
         ret = cpu_write_elf32_note(fd_write_vmcore, cpu, id, s);
         if (ret < 0) {
@@ -336,7 +336,7 @@ static int write_elf32_notes(DumpState *s)
         }
     }
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         ret = cpu_write_elf32_qemunote(fd_write_vmcore, cpu, s);
         if (ret < 0) {
             dump_error(s, "dump: failed to write CPU status.\n");
@@ -734,7 +734,7 @@ static int dump_init(DumpState *s, int fd, bool paging, bool has_filter,
      */
     cpu_synchronize_all_states();
     nr_cpus = 0;
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         nr_cpus++;
     }
 
diff --git a/exec.c b/exec.c
index 3ca9381214..ca2a504f16 100644
--- a/exec.c
+++ b/exec.c
@@ -69,7 +69,7 @@ static MemoryRegion io_mem_unassigned;
 
 #endif
 
-CPUState *first_cpu;
+struct CPUTailQ cpus = QTAILQ_HEAD_INITIALIZER(cpus);
 /* current CPU in the current thread. It is only valid inside
    cpu_exec() */
 DEFINE_TLS(CPUState *, current_cpu);
@@ -351,26 +351,23 @@ const VMStateDescription vmstate_cpu_common = {
 
 CPUState *qemu_get_cpu(int index)
 {
-    CPUState *cpu = first_cpu;
+    CPUState *cpu;
 
-    while (cpu) {
+    CPU_FOREACH(cpu) {
         if (cpu->cpu_index == index) {
-            break;
+            return cpu;
         }
-        cpu = cpu->next_cpu;
     }
 
-    return cpu;
+    return NULL;
 }
 
 void qemu_for_each_cpu(void (*func)(CPUState *cpu, void *data), void *data)
 {
     CPUState *cpu;
 
-    cpu = first_cpu;
-    while (cpu) {
+    CPU_FOREACH(cpu) {
         func(cpu, data);
-        cpu = cpu->next_cpu;
     }
 }
 
@@ -378,17 +375,14 @@ void cpu_exec_init(CPUArchState *env)
 {
     CPUState *cpu = ENV_GET_CPU(env);
     CPUClass *cc = CPU_GET_CLASS(cpu);
-    CPUState **pcpu;
+    CPUState *some_cpu;
     int cpu_index;
 
 #if defined(CONFIG_USER_ONLY)
     cpu_list_lock();
 #endif
-    cpu->next_cpu = NULL;
-    pcpu = &first_cpu;
     cpu_index = 0;
-    while (*pcpu != NULL) {
-        pcpu = &(*pcpu)->next_cpu;
+    CPU_FOREACH(some_cpu) {
         cpu_index++;
     }
     cpu->cpu_index = cpu_index;
@@ -398,7 +392,7 @@ void cpu_exec_init(CPUArchState *env)
 #ifndef CONFIG_USER_ONLY
     cpu->thread_id = qemu_get_thread_id();
 #endif
-    *pcpu = cpu;
+    QTAILQ_INSERT_TAIL(&cpus, cpu, node);
 #if defined(CONFIG_USER_ONLY)
     cpu_list_unlock();
 #endif
@@ -1762,7 +1756,7 @@ static void tcg_commit(MemoryListener *listener)
     /* since each CPU stores ram addresses in its TLB cache, we must
        reset the modified entries */
     /* XXX: slow ! */
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         CPUArchState *env = cpu->env_ptr;
 
         tlb_flush(env, 1);
diff --git a/gdbstub.c b/gdbstub.c
index 9d067d6b80..2b7f22b2d2 100644
--- a/gdbstub.c
+++ b/gdbstub.c
@@ -648,7 +648,7 @@ static int gdb_breakpoint_insert(target_ulong addr, target_ulong len, int type)
     switch (type) {
     case GDB_BREAKPOINT_SW:
     case GDB_BREAKPOINT_HW:
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             env = cpu->env_ptr;
             err = cpu_breakpoint_insert(env, addr, BP_GDB, NULL);
             if (err)
@@ -659,7 +659,7 @@ static int gdb_breakpoint_insert(target_ulong addr, target_ulong len, int type)
     case GDB_WATCHPOINT_WRITE:
     case GDB_WATCHPOINT_READ:
     case GDB_WATCHPOINT_ACCESS:
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             env = cpu->env_ptr;
             err = cpu_watchpoint_insert(env, addr, len, xlat_gdb_type[type],
                                         NULL);
@@ -686,7 +686,7 @@ static int gdb_breakpoint_remove(target_ulong addr, target_ulong len, int type)
     switch (type) {
     case GDB_BREAKPOINT_SW:
     case GDB_BREAKPOINT_HW:
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             env = cpu->env_ptr;
             err = cpu_breakpoint_remove(env, addr, BP_GDB);
             if (err)
@@ -697,7 +697,7 @@ static int gdb_breakpoint_remove(target_ulong addr, target_ulong len, int type)
     case GDB_WATCHPOINT_WRITE:
     case GDB_WATCHPOINT_READ:
     case GDB_WATCHPOINT_ACCESS:
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             env = cpu->env_ptr;
             err = cpu_watchpoint_remove(env, addr, len, xlat_gdb_type[type]);
             if (err)
@@ -720,7 +720,7 @@ static void gdb_breakpoint_remove_all(void)
         return;
     }
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         env = cpu->env_ptr;
         cpu_breakpoint_remove_all(env, BP_GDB);
 #ifndef CONFIG_USER_ONLY
@@ -744,7 +744,7 @@ static CPUState *find_cpu(uint32_t thread_id)
 {
     CPUState *cpu;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         if (cpu_index(cpu) == thread_id) {
             return cpu;
         }
@@ -1070,7 +1070,7 @@ static int gdb_handle_packet(GDBState *s, const char *line_buf)
             if (s->query_cpu) {
                 snprintf(buf, sizeof(buf), "m%x", cpu_index(s->query_cpu));
                 put_packet(s, buf);
-                s->query_cpu = s->query_cpu->next_cpu;
+                s->query_cpu = CPU_NEXT(s->query_cpu);
             } else
                 put_packet(s, "l");
             break;
diff --git a/hw/arm/boot.c b/hw/arm/boot.c
index 2cbeefdcba..1e313afe8d 100644
--- a/hw/arm/boot.c
+++ b/hw/arm/boot.c
@@ -468,7 +468,7 @@ void arm_load_kernel(ARMCPU *cpu, struct arm_boot_info *info)
     }
     info->is_linux = is_linux;
 
-    for (; cs; cs = cs->next_cpu) {
+    for (; cs; cs = CPU_NEXT(cs)) {
         cpu = ARM_CPU(cs);
         cpu->env.boot_info = info;
         qemu_register_reset(do_cpu_reset, cpu);
diff --git a/hw/i386/kvm/clock.c b/hw/i386/kvm/clock.c
index e89e2f768e..92aabb83b5 100644
--- a/hw/i386/kvm/clock.c
+++ b/hw/i386/kvm/clock.c
@@ -59,7 +59,7 @@ static void kvmclock_vm_state_change(void *opaque, int running,
         if (!cap_clock_ctrl) {
             return;
         }
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             ret = kvm_vcpu_ioctl(cpu, KVM_KVMCLOCK_CTRL, 0);
             if (ret) {
                 if (ret != -EINVAL) {
diff --git a/hw/i386/kvmvapic.c b/hw/i386/kvmvapic.c
index 15beb8044e..d3a6fbe1f9 100644
--- a/hw/i386/kvmvapic.c
+++ b/hw/i386/kvmvapic.c
@@ -498,7 +498,7 @@ static void vapic_enable_tpr_reporting(bool enable)
     X86CPU *cpu;
     CPUX86State *env;
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         cpu = X86_CPU(cs);
         env = &cpu->env;
         info.apic = env->apic_state;
diff --git a/hw/i386/pc.c b/hw/i386/pc.c
index 3a620a1856..0c313feb0b 100644
--- a/hw/i386/pc.c
+++ b/hw/i386/pc.c
@@ -191,13 +191,12 @@ static void pic_irq_request(void *opaque, int irq, int level)
 
     DPRINTF("pic_irqs: %s irq %d\n", level? "raise" : "lower", irq);
     if (env->apic_state) {
-        while (cs) {
+        CPU_FOREACH(cs) {
             cpu = X86_CPU(cs);
             env = &cpu->env;
             if (apic_accept_pic_intr(env->apic_state)) {
                 apic_deliver_pic_intr(env->apic_state, level);
             }
-            cs = cs->next_cpu;
         }
     } else {
         if (level) {
diff --git a/hw/ppc/e500.c b/hw/ppc/e500.c
index 9059ff9bc7..cfdd84b969 100644
--- a/hw/ppc/e500.c
+++ b/hw/ppc/e500.c
@@ -540,7 +540,7 @@ static DeviceState *ppce500_init_mpic_kvm(PPCE500Params *params,
         return NULL;
     }
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         if (kvm_openpic_connect_vcpu(dev, cs)) {
             fprintf(stderr, "%s: failed to connect vcpu to irqchip\n",
                     __func__);
diff --git a/hw/ppc/ppc.c b/hw/ppc/ppc.c
index 59b41cbc6f..bf2d3d4b35 100644
--- a/hw/ppc/ppc.c
+++ b/hw/ppc/ppc.c
@@ -443,7 +443,7 @@ void ppce500_set_mpic_proxy(bool enabled)
 {
     CPUState *cs;
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         PowerPCCPU *cpu = POWERPC_CPU(cs);
 
         cpu->env.mpic_proxy = enabled;
diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
index 04f0ee3da1..8c6e296ff0 100644
--- a/hw/ppc/spapr.c
+++ b/hw/ppc/spapr.c
@@ -187,7 +187,7 @@ static int spapr_fixup_cpu_dt(void *fdt, sPAPREnvironment *spapr)
 
     assert(spapr->cpu_model);
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         uint32_t associativity[] = {cpu_to_be32(0x5),
                                     cpu_to_be32(0x0),
                                     cpu_to_be32(0x0),
@@ -351,7 +351,7 @@ static void *spapr_create_fdt_skel(const char *cpu_model,
     /* This is needed during FDT finalization */
     spapr->cpu_model = g_strdup(modelname);
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         PowerPCCPU *cpu = POWERPC_CPU(cs);
         CPUPPCState *env = &cpu->env;
         PowerPCCPUClass *pcc = POWERPC_CPU_GET_CLASS(cs);
diff --git a/hw/ppc/spapr_hcall.c b/hw/ppc/spapr_hcall.c
index 89e6a00dd9..f10ba8a932 100644
--- a/hw/ppc/spapr_hcall.c
+++ b/hw/ppc/spapr_hcall.c
@@ -679,7 +679,7 @@ static target_ulong h_set_mode(PowerPCCPU *cpu, sPAPREnvironment *spapr,
 
         switch (mflags) {
         case H_SET_MODE_ENDIAN_BIG:
-            for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+            CPU_FOREACH(cs) {
                 PowerPCCPU *cp = POWERPC_CPU(cs);
                 CPUPPCState *env = &cp->env;
                 env->spr[SPR_LPCR] &= ~LPCR_ILE;
@@ -688,7 +688,7 @@ static target_ulong h_set_mode(PowerPCCPU *cpu, sPAPREnvironment *spapr,
             break;
 
         case H_SET_MODE_ENDIAN_LITTLE:
-            for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+            CPU_FOREACH(cs) {
                 PowerPCCPU *cp = POWERPC_CPU(cs);
                 CPUPPCState *env = &cp->env;
                 env->spr[SPR_LPCR] |= LPCR_ILE;
diff --git a/include/qom/cpu.h b/include/qom/cpu.h
index 3e4993661a..79f7c8709d 100644
--- a/include/qom/cpu.h
+++ b/include/qom/cpu.h
@@ -23,6 +23,7 @@
 #include <signal.h>
 #include "hw/qdev-core.h"
 #include "exec/hwaddr.h"
+#include "qemu/queue.h"
 #include "qemu/thread.h"
 #include "qemu/tls.h"
 #include "qemu/typedefs.h"
@@ -190,7 +191,7 @@ struct CPUState {
     struct GDBRegisterState *gdb_regs;
     int gdb_num_regs;
     int gdb_num_g_regs;
-    CPUState *next_cpu;
+    QTAILQ_ENTRY(CPUState) node;
 
     int kvm_fd;
     bool kvm_vcpu_dirty;
@@ -202,7 +203,13 @@ struct CPUState {
     uint32_t halted; /* used by alpha, cris, ppc TCG */
 };
 
-extern CPUState *first_cpu;
+QTAILQ_HEAD(CPUTailQ, CPUState);
+extern struct CPUTailQ cpus;
+#define CPU_NEXT(cpu) QTAILQ_NEXT(cpu, node)
+#define CPU_FOREACH(cpu) QTAILQ_FOREACH(cpu, &cpus, node)
+#define CPU_FOREACH_SAFE(cpu, next_cpu) \
+    QTAILQ_FOREACH_SAFE(cpu, &cpus, node, next_cpu)
+#define first_cpu QTAILQ_FIRST(&cpus)
 
 DECLARE_TLS(CPUState *, current_cpu);
 #define current_cpu tls_var(current_cpu)
diff --git a/kvm-all.c b/kvm-all.c
index 875e32ec87..c29a015cca 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -1925,7 +1925,7 @@ int kvm_insert_breakpoint(CPUState *cpu, target_ulong addr,
         }
     }
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         err = kvm_update_guest_debug(cpu, 0);
         if (err) {
             return err;
@@ -1965,7 +1965,7 @@ int kvm_remove_breakpoint(CPUState *cpu, target_ulong addr,
         }
     }
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         err = kvm_update_guest_debug(cpu, 0);
         if (err) {
             return err;
@@ -1982,7 +1982,7 @@ void kvm_remove_all_breakpoints(CPUState *cpu)
     QTAILQ_FOREACH_SAFE(bp, &s->kvm_sw_breakpoints, entry, next) {
         if (kvm_arch_remove_sw_breakpoint(cpu, bp) != 0) {
             /* Try harder to find a CPU that currently sees the breakpoint. */
-            for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+            CPU_FOREACH(cpu) {
                 if (kvm_arch_remove_sw_breakpoint(cpu, bp) == 0) {
                     break;
                 }
@@ -1993,7 +1993,7 @@ void kvm_remove_all_breakpoints(CPUState *cpu)
     }
     kvm_arch_remove_all_hw_breakpoints();
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         kvm_update_guest_debug(cpu, 0);
     }
 }
diff --git a/linux-user/elfload.c b/linux-user/elfload.c
index 7ce2eab1bb..72d92707c6 100644
--- a/linux-user/elfload.c
+++ b/linux-user/elfload.c
@@ -2668,7 +2668,7 @@ static int fill_note_info(struct elf_note_info *info,
 
     /* read and fill status of all threads */
     cpu_list_lock();
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         if (cpu == thread_cpu) {
             continue;
         }
diff --git a/linux-user/main.c b/linux-user/main.c
index 03859bcc23..5c2f7b26b4 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -117,10 +117,14 @@ void fork_end(int child)
 {
     mmap_fork_end(child);
     if (child) {
+        CPUState *cpu, *next_cpu;
         /* Child processes created by fork() only have a single thread.
            Discard information about the parent threads.  */
-        first_cpu = thread_cpu;
-        first_cpu->next_cpu = NULL;
+        CPU_FOREACH_SAFE(cpu, next_cpu) {
+            if (cpu != thread_cpu) {
+                QTAILQ_REMOVE(&cpus, thread_cpu, node);
+            }
+        }
         pending_cpus = 0;
         pthread_mutex_init(&exclusive_lock, NULL);
         pthread_mutex_init(&cpu_list_mutex, NULL);
@@ -154,7 +158,7 @@ static inline void start_exclusive(void)
 
     pending_cpus = 1;
     /* Make all other cpus stop executing.  */
-    for (other_cpu = first_cpu; other_cpu; other_cpu = other_cpu->next_cpu) {
+    CPU_FOREACH(other_cpu) {
         if (other_cpu->running) {
             pending_cpus++;
             cpu_exit(other_cpu);
diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index f986548c2d..ecead512a0 100644
--- a/linux-user/syscall.c
+++ b/linux-user/syscall.c
@@ -5113,25 +5113,12 @@ abi_long do_syscall(void *cpu_env, int num, abi_long arg1,
            Do thread termination if we have more then one thread.  */
         /* FIXME: This probably breaks if a signal arrives.  We should probably
            be disabling signals.  */
-        if (first_cpu->next_cpu) {
+        if (CPU_NEXT(first_cpu)) {
             TaskState *ts;
-            CPUState **lastp;
-            CPUState *p;
 
             cpu_list_lock();
-            lastp = &first_cpu;
-            p = first_cpu;
-            while (p && p != cpu) {
-                lastp = &p->next_cpu;
-                p = p->next_cpu;
-            }
-            /* If we didn't find the CPU for this thread then something is
-               horribly wrong.  */
-            if (!p) {
-                abort();
-            }
             /* Remove the CPU from the list.  */
-            *lastp = p->next_cpu;
+            QTAILQ_REMOVE(&cpus, cpu, node);
             cpu_list_unlock();
             ts = ((CPUArchState *)cpu_env)->opaque;
             if (ts->child_tidptr) {
diff --git a/memory_mapping.c b/memory_mapping.c
index eeeeb44026..87a6ed5c8e 100644
--- a/memory_mapping.c
+++ b/memory_mapping.c
@@ -270,7 +270,7 @@ static CPUState *find_paging_enabled_cpu(CPUState *start_cpu)
 {
     CPUState *cpu;
 
-    for (cpu = start_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         if (cpu_paging_enabled(cpu)) {
             return cpu;
         }
@@ -289,7 +289,8 @@ void qemu_get_guest_memory_mapping(MemoryMappingList *list,
 
     first_paging_enabled_cpu = find_paging_enabled_cpu(first_cpu);
     if (first_paging_enabled_cpu) {
-        for (cpu = first_paging_enabled_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        for (cpu = first_paging_enabled_cpu; cpu != NULL;
+             cpu = CPU_NEXT(cpu)) {
             Error *err = NULL;
             cpu_get_memory_mapping(cpu, list, &err);
             if (err) {
diff --git a/monitor.c b/monitor.c
index 0aeaf6c56b..683babf24a 100644
--- a/monitor.c
+++ b/monitor.c
@@ -2002,7 +2002,7 @@ static void do_info_numa(Monitor *mon, const QDict *qdict)
     monitor_printf(mon, "%d nodes\n", nb_numa_nodes);
     for (i = 0; i < nb_numa_nodes; i++) {
         monitor_printf(mon, "node %d cpus:", i);
-        for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+        CPU_FOREACH(cpu) {
             if (cpu->numa_node == i) {
                 monitor_printf(mon, " %d", cpu->cpu_index);
             }
diff --git a/target-i386/helper.c b/target-i386/helper.c
index bf3e2ac73d..7c58e274d9 100644
--- a/target-i386/helper.c
+++ b/target-i386/helper.c
@@ -1231,8 +1231,7 @@ void cpu_x86_inject_mce(Monitor *mon, X86CPU *cpu, int bank,
         params.mcg_status = MCG_STATUS_MCIP | MCG_STATUS_RIPV;
         params.addr = 0;
         params.misc = 0;
-        for (other_cs = first_cpu; other_cs != NULL;
-             other_cs = other_cs->next_cpu) {
+        CPU_FOREACH(other_cs) {
             if (other_cs == cs) {
                 continue;
             }
diff --git a/target-i386/misc_helper.c b/target-i386/misc_helper.c
index 957926ced7..93933fd162 100644
--- a/target-i386/misc_helper.c
+++ b/target-i386/misc_helper.c
@@ -610,7 +610,7 @@ void helper_mwait(CPUX86State *env, int next_eip_addend)
     cpu = x86_env_get_cpu(env);
     cs = CPU(cpu);
     /* XXX: not complete but not completely erroneous */
-    if (cs->cpu_index != 0 || cs->next_cpu != NULL) {
+    if (cs->cpu_index != 0 || CPU_NEXT(cs) != NULL) {
         /* more than one CPU: do not sleep because another CPU may
            wake this one */
     } else {
diff --git a/target-mips/op_helper.c b/target-mips/op_helper.c
index b828375714..8e3a6d7da6 100644
--- a/target-mips/op_helper.c
+++ b/target-mips/op_helper.c
@@ -1699,15 +1699,14 @@ target_ulong helper_dvpe(CPUMIPSState *env)
     CPUState *other_cs = first_cpu;
     target_ulong prev = env->mvp->CP0_MVPControl;
 
-    do {
+    CPU_FOREACH(other_cs) {
         MIPSCPU *other_cpu = MIPS_CPU(other_cs);
         /* Turn off all VPEs except the one executing the dvpe.  */
         if (&other_cpu->env != env) {
             other_cpu->env.mvp->CP0_MVPControl &= ~(1 << CP0MVPCo_EVP);
             mips_vpe_sleep(other_cpu);
         }
-        other_cs = other_cs->next_cpu;
-    } while (other_cs);
+    }
     return prev;
 }
 
@@ -1716,7 +1715,7 @@ target_ulong helper_evpe(CPUMIPSState *env)
     CPUState *other_cs = first_cpu;
     target_ulong prev = env->mvp->CP0_MVPControl;
 
-    do {
+    CPU_FOREACH(other_cs) {
         MIPSCPU *other_cpu = MIPS_CPU(other_cs);
 
         if (&other_cpu->env != env
@@ -1726,8 +1725,7 @@ target_ulong helper_evpe(CPUMIPSState *env)
             other_cpu->env.mvp->CP0_MVPControl |= (1 << CP0MVPCo_EVP);
             mips_vpe_wake(other_cpu); /* And wake it up.  */
         }
-        other_cs = other_cs->next_cpu;
-    } while (other_cs);
+    }
     return prev;
 }
 #endif /* !CONFIG_USER_ONLY */
diff --git a/target-ppc/excp_helper.c b/target-ppc/excp_helper.c
index e957761109..c959460f70 100644
--- a/target-ppc/excp_helper.c
+++ b/target-ppc/excp_helper.c
@@ -1002,7 +1002,7 @@ void helper_msgsnd(target_ulong rb)
         return;
     }
 
-    for (cs = first_cpu; cs != NULL; cs = cs->next_cpu) {
+    CPU_FOREACH(cs) {
         PowerPCCPU *cpu = POWERPC_CPU(cs);
         CPUPPCState *cenv = &cpu->env;
 
diff --git a/target-s390x/misc_helper.c b/target-s390x/misc_helper.c
index 4afd7dab1c..1690907169 100644
--- a/target-s390x/misc_helper.c
+++ b/target-s390x/misc_helper.c
@@ -183,12 +183,12 @@ uint32_t HELPER(servc)(CPUS390XState *env, uint64_t r1, uint64_t r2)
 #ifndef CONFIG_USER_ONLY
 static void cpu_reset_all(void)
 {
-    CPUState *cpu;
+    CPUState *cs;
     S390CPUClass *scc;
 
-    for (cpu = first_cpu; cpu; cpu = cpu->next_cpu) {
-        scc = S390_CPU_GET_CLASS(CPU(cpu));
-        scc->cpu_reset(CPU(cpu));
+    CPU_FOREACH(cs) {
+        scc = S390_CPU_GET_CLASS(cs);
+        scc->cpu_reset(cs);
     }
 }
 
diff --git a/translate-all.c b/translate-all.c
index 3b5fc7c901..2c923c644b 100644
--- a/translate-all.c
+++ b/translate-all.c
@@ -696,7 +696,7 @@ void tb_flush(CPUArchState *env1)
     }
     tcg_ctx.tb_ctx.nb_tbs = 0;
 
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         CPUArchState *env = cpu->env_ptr;
 
         memset(env->tb_jmp_cache, 0, TB_JMP_CACHE_SIZE * sizeof(void *));
@@ -850,7 +850,7 @@ void tb_phys_invalidate(TranslationBlock *tb, tb_page_addr_t page_addr)
 
     /* remove the TB from the hash list */
     h = tb_jmp_cache_hash_func(tb->pc);
-    for (cpu = first_cpu; cpu != NULL; cpu = cpu->next_cpu) {
+    CPU_FOREACH(cpu) {
         CPUArchState *env = cpu->env_ptr;
 
         if (env->tb_jmp_cache[h] == tb) {