SMP support

git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@1640 c046a42c-6fe2-441c-8c8c-71466251a162

9 files changed, 296 insertions, 138 deletions

diff --git a/cpu-defs.h b/cpu-defs.h
index fb4f8e8a29..c308253813 100644
--- a/cpu-defs.h
+++ b/cpu-defs.h
@@ -96,6 +96,7 @@ typedef struct CPUTLBEntry {
 
 #define CPU_COMMON                                                      \
     struct TranslationBlock *current_tb; /* currently executing TB  */  \
+    int cpu_halted; /* TRUE if cpu is halted (sleep mode) */            \
     /* soft mmu support */                                              \
     /* in order to avoid passing too many arguments to the memory       \
        write helpers, we store some rarely used information in the CPU  \
@@ -115,9 +116,9 @@ typedef struct CPUTLBEntry {
     int nb_breakpoints;                                                 \
     int singlestep_enabled;                                             \
                                                                         \
+    void *next_cpu; /* next CPU sharing TB cache */                     \
+    int cpu_index; /* CPU index (informative) */                        \
     /* user data */                                                     \
     void *opaque;
 
-
-
 #endif
diff --git a/cpu-exec.c b/cpu-exec.c
index 9f6487adff..72e32684a1 100644
--- a/cpu-exec.c
+++ b/cpu-exec.c
@@ -251,6 +251,8 @@ int cpu_exec(CPUState *env1)
     TranslationBlock *tb;
     uint8_t *tc_ptr;
 
+    cpu_single_env = env1; 
+
     /* first we save global registers */
     saved_env = env;
     env = env1;
@@ -755,6 +757,8 @@ int cpu_exec(CPUState *env1)
     T2 = saved_T2;
 #endif
     env = saved_env;
+    /* fail safe : never use cpu_single_env outside cpu_exec() */
+    cpu_single_env = NULL; 
     return ret;
 }
 
diff --git a/disas.c b/disas.c
index feb8a93caa..78a72675df 100644
--- a/disas.c
+++ b/disas.c
@@ -138,6 +138,7 @@ print_insn_thumb1(bfd_vma pc, disassemble_info *info)
    values:
     i386 - nonzero means 16 bit code
     arm  - nonzero means thumb code 
+    ppc  - nonzero means little endian
     other targets - unused
  */
 void target_disas(FILE *out, target_ulong code, target_ulong size, int flags)
@@ -177,7 +178,7 @@ void target_disas(FILE *out, target_ulong code, target_ulong size, int flags)
     disasm_info.mach = bfd_mach_sparc_v9b;
 #endif    
 #elif defined(TARGET_PPC)
-    if (cpu_single_env->msr[MSR_LE])
+    if (flags)
         disasm_info.endian = BFD_ENDIAN_LITTLE;
 #ifdef TARGET_PPC64
     disasm_info.mach = bfd_mach_ppc64;
@@ -314,6 +315,7 @@ void term_vprintf(const char *fmt, va_list ap);
 void term_printf(const char *fmt, ...);
 
 static int monitor_disas_is_physical;
+static CPUState *monitor_disas_env;
 
 static int
 monitor_read_memory (memaddr, myaddr, length, info)
@@ -325,7 +327,7 @@ monitor_read_memory (memaddr, myaddr, length, info)
     if (monitor_disas_is_physical) {
         cpu_physical_memory_rw(memaddr, myaddr, length, 0);
     } else {
-        cpu_memory_rw_debug(cpu_single_env, memaddr,myaddr, length, 0);
+        cpu_memory_rw_debug(monitor_disas_env, memaddr,myaddr, length, 0);
     }
     return 0;
 }
@@ -339,7 +341,8 @@ static int monitor_fprintf(FILE *stream, const char *fmt, ...)
     return 0;
 }
 
-void monitor_disas(target_ulong pc, int nb_insn, int is_physical, int flags)
+void monitor_disas(CPUState *env,
+                   target_ulong pc, int nb_insn, int is_physical, int flags)
 {
     int count, i;
     struct disassemble_info disasm_info;
@@ -347,6 +350,7 @@ void monitor_disas(target_ulong pc, int nb_insn, int is_physical, int flags)
 
     INIT_DISASSEMBLE_INFO(disasm_info, NULL, monitor_fprintf);
 
+    monitor_disas_env = env;
     monitor_disas_is_physical = is_physical;
     disasm_info.read_memory_func = monitor_read_memory;
 
diff --git a/exec-all.h b/exec-all.h
index 874e9d9a35..67aa227ad4 100644
--- a/exec-all.h
+++ b/exec-all.h
@@ -91,7 +91,7 @@ int cpu_restore_state_copy(struct TranslationBlock *tb,
                            CPUState *env, unsigned long searched_pc,
                            void *puc);
 void cpu_resume_from_signal(CPUState *env1, void *puc);
-void cpu_exec_init(void);
+void cpu_exec_init(CPUState *env);
 int page_unprotect(unsigned long address, unsigned long pc, void *puc);
 void tb_invalidate_phys_page_range(target_ulong start, target_ulong end, 
                                    int is_cpu_write_access);
diff --git a/exec.c b/exec.c
index 8aed3d00c0..3dbd3eb075 100644
--- a/exec.c
+++ b/exec.c
@@ -74,6 +74,11 @@ int phys_ram_fd;
 uint8_t *phys_ram_base;
 uint8_t *phys_ram_dirty;
 
+CPUState *first_cpu;
+/* current CPU in the current thread. It is only valid inside
+   cpu_exec() */
+CPUState *cpu_single_env; 
+
 typedef struct PageDesc {
     /* list of TBs intersecting this ram page */
     TranslationBlock *first_tb;
@@ -233,19 +238,30 @@ static inline PhysPageDesc *phys_page_find(target_phys_addr_t index)
 }
 
 #if !defined(CONFIG_USER_ONLY)
-static void tlb_protect_code(CPUState *env, ram_addr_t ram_addr, 
-                             target_ulong vaddr);
+static void tlb_protect_code(ram_addr_t ram_addr);
 static void tlb_unprotect_code_phys(CPUState *env, ram_addr_t ram_addr, 
                                     target_ulong vaddr);
 #endif
 
-void cpu_exec_init(void)
+void cpu_exec_init(CPUState *env)
 {
+    CPUState **penv;
+    int cpu_index;
+
     if (!code_gen_ptr) {
         code_gen_ptr = code_gen_buffer;
         page_init();
         io_mem_init();
     }
+    env->next_cpu = NULL;
+    penv = &first_cpu;
+    cpu_index = 0;
+    while (*penv != NULL) {
+        penv = (CPUState **)&(*penv)->next_cpu;
+        cpu_index++;
+    }
+    env->cpu_index = cpu_index;
+    *penv = env;
 }
 
 static inline void invalidate_page_bitmap(PageDesc *p)
@@ -277,8 +293,9 @@ static void page_flush_tb(void)
 
 /* flush all the translation blocks */
 /* XXX: tb_flush is currently not thread safe */
-void tb_flush(CPUState *env)
+void tb_flush(CPUState *env1)
 {
+    CPUState *env;
 #if defined(DEBUG_FLUSH)
     printf("qemu: flush code_size=%d nb_tbs=%d avg_tb_size=%d\n", 
            code_gen_ptr - code_gen_buffer, 
@@ -286,7 +303,10 @@ void tb_flush(CPUState *env)
            nb_tbs > 0 ? (code_gen_ptr - code_gen_buffer) / nb_tbs : 0);
 #endif
     nb_tbs = 0;
-    memset (env->tb_jmp_cache, 0, TB_JMP_CACHE_SIZE * sizeof (void *));
+    
+    for(env = first_cpu; env != NULL; env = env->next_cpu) {
+        memset (env->tb_jmp_cache, 0, TB_JMP_CACHE_SIZE * sizeof (void *));
+    }
 
     memset (tb_phys_hash, 0, CODE_GEN_PHYS_HASH_SIZE * sizeof (void *));
     page_flush_tb();
@@ -424,6 +444,7 @@ static inline void tb_reset_jump(TranslationBlock *tb, int n)
 
 static inline void tb_phys_invalidate(TranslationBlock *tb, unsigned int page_addr)
 {
+    CPUState *env;
     PageDesc *p;
     unsigned int h, n1;
     target_ulong phys_pc;
@@ -451,7 +472,10 @@ static inline void tb_phys_invalidate(TranslationBlock *tb, unsigned int page_ad
 
     /* remove the TB from the hash list */
     h = tb_jmp_cache_hash_func(tb->pc);
-    cpu_single_env->tb_jmp_cache[h] = NULL;
+    for(env = first_cpu; env != NULL; env = env->next_cpu) {
+        if (env->tb_jmp_cache[h] == tb)
+            env->tb_jmp_cache[h] = NULL;
+    }
 
     /* suppress this TB from the two jump lists */
     tb_jmp_remove(tb, 0);
@@ -818,10 +842,7 @@ static inline void tb_alloc_page(TranslationBlock *tb,
        protected. So we handle the case where only the first TB is
        allocated in a physical page */
     if (!last_first_tb) {
-        target_ulong virt_addr;
-
-        virt_addr = (tb->pc & TARGET_PAGE_MASK) + (n << TARGET_PAGE_BITS);
-        tlb_protect_code(cpu_single_env, page_addr, virt_addr);
+        tlb_protect_code(page_addr);
     }
 #endif
 
@@ -1246,40 +1267,13 @@ void tlb_flush_page(CPUState *env, target_ulong addr)
 #endif
 }
 
-static inline void tlb_protect_code1(CPUTLBEntry *tlb_entry, target_ulong addr)
-{
-    if (addr == (tlb_entry->address & 
-                 (TARGET_PAGE_MASK | TLB_INVALID_MASK)) &&
-        (tlb_entry->address & ~TARGET_PAGE_MASK) == IO_MEM_RAM) {
-        tlb_entry->address = (tlb_entry->address & TARGET_PAGE_MASK) | IO_MEM_NOTDIRTY;
-    }
-}
-
 /* update the TLBs so that writes to code in the virtual page 'addr'
    can be detected */
-static void tlb_protect_code(CPUState *env, ram_addr_t ram_addr, 
-                             target_ulong vaddr)
+static void tlb_protect_code(ram_addr_t ram_addr)
 {
-    int i;
-
-    vaddr &= TARGET_PAGE_MASK;
-    i = (vaddr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1);
-    tlb_protect_code1(&env->tlb_write[0][i], vaddr);
-    tlb_protect_code1(&env->tlb_write[1][i], vaddr);
-
-#ifdef USE_KQEMU
-    if (env->kqemu_enabled) {
-        kqemu_set_notdirty(env, ram_addr);
-    }
-#endif
-    phys_ram_dirty[ram_addr >> TARGET_PAGE_BITS] &= ~CODE_DIRTY_FLAG;
-    
-#if !defined(CONFIG_SOFTMMU)
-    /* NOTE: as we generated the code for this page, it is already at
-       least readable */
-    if (vaddr < MMAP_AREA_END)
-        mprotect((void *)vaddr, TARGET_PAGE_SIZE, PROT_READ);
-#endif
+    cpu_physical_memory_reset_dirty(ram_addr, 
+                                    ram_addr + TARGET_PAGE_SIZE,
+                                    CODE_DIRTY_FLAG);
 }
 
 /* update the TLB so that writes in physical page 'phys_addr' are no longer
@@ -1317,8 +1311,9 @@ void cpu_physical_memory_reset_dirty(ram_addr_t start, ram_addr_t end,
     if (length == 0)
         return;
     len = length >> TARGET_PAGE_BITS;
-    env = cpu_single_env;
 #ifdef USE_KQEMU
+    /* XXX: should not depend on cpu context */
+    env = first_cpu;
     if (env->kqemu_enabled) {
         ram_addr_t addr;
         addr = start;
@@ -1336,10 +1331,12 @@ void cpu_physical_memory_reset_dirty(ram_addr_t start, ram_addr_t end,
     /* we modify the TLB cache so that the dirty bit will be set again
        when accessing the range */
     start1 = start + (unsigned long)phys_ram_base;
-    for(i = 0; i < CPU_TLB_SIZE; i++)
-        tlb_reset_dirty_range(&env->tlb_write[0][i], start1, length);
-    for(i = 0; i < CPU_TLB_SIZE; i++)
-        tlb_reset_dirty_range(&env->tlb_write[1][i], start1, length);
+    for(env = first_cpu; env != NULL; env = env->next_cpu) {
+        for(i = 0; i < CPU_TLB_SIZE; i++)
+            tlb_reset_dirty_range(&env->tlb_write[0][i], start1, length);
+        for(i = 0; i < CPU_TLB_SIZE; i++)
+            tlb_reset_dirty_range(&env->tlb_write[1][i], start1, length);
+    }
 
 #if !defined(CONFIG_SOFTMMU)
     /* XXX: this is expensive */
@@ -1407,9 +1404,9 @@ static inline void tlb_set_dirty1(CPUTLBEntry *tlb_entry,
 
 /* update the TLB corresponding to virtual page vaddr and phys addr
    addr so that it is no longer dirty */
-static inline void tlb_set_dirty(unsigned long addr, target_ulong vaddr)
+static inline void tlb_set_dirty(CPUState *env,
+                                 unsigned long addr, target_ulong vaddr)
 {
-    CPUState *env = cpu_single_env;
     int i;
 
     addr &= TARGET_PAGE_MASK;
@@ -1723,7 +1720,8 @@ void page_unprotect_range(uint8_t *data, unsigned long data_size)
     }
 }
 
-static inline void tlb_set_dirty(unsigned long addr, target_ulong vaddr)
+static inline void tlb_set_dirty(CPUState *env,
+                                 unsigned long addr, target_ulong vaddr)
 {
 }
 #endif /* defined(CONFIG_USER_ONLY) */
@@ -1787,7 +1785,7 @@ static void notdirty_mem_writeb(void *opaque, target_phys_addr_t addr, uint32_t
     /* we remove the notdirty callback only if the code has been
        flushed */
     if (dirty_flags == 0xff)
-        tlb_set_dirty(addr, cpu_single_env->mem_write_vaddr);
+        tlb_set_dirty(cpu_single_env, addr, cpu_single_env->mem_write_vaddr);
 }
 
 static void notdirty_mem_writew(void *opaque, target_phys_addr_t addr, uint32_t val)
@@ -1808,7 +1806,7 @@ static void notdirty_mem_writew(void *opaque, target_phys_addr_t addr, uint32_t
     /* we remove the notdirty callback only if the code has been
        flushed */
     if (dirty_flags == 0xff)
-        tlb_set_dirty(addr, cpu_single_env->mem_write_vaddr);
+        tlb_set_dirty(cpu_single_env, addr, cpu_single_env->mem_write_vaddr);
 }
 
 static void notdirty_mem_writel(void *opaque, target_phys_addr_t addr, uint32_t val)
@@ -1829,7 +1827,7 @@ static void notdirty_mem_writel(void *opaque, target_phys_addr_t addr, uint32_t
     /* we remove the notdirty callback only if the code has been
        flushed */
     if (dirty_flags == 0xff)
-        tlb_set_dirty(addr, cpu_single_env->mem_write_vaddr);
+        tlb_set_dirty(cpu_single_env, addr, cpu_single_env->mem_write_vaddr);
 }
 
 static CPUReadMemoryFunc *error_mem_read[3] = {
@@ -1953,6 +1951,8 @@ void cpu_physical_memory_rw(target_phys_addr_t addr, uint8_t *buf,
         if (is_write) {
             if ((pd & ~TARGET_PAGE_MASK) != IO_MEM_RAM) {
                 io_index = (pd >> IO_MEM_SHIFT) & (IO_MEM_NB_ENTRIES - 1);
+                /* XXX: could force cpu_single_env to NULL to avoid
+                   potential bugs */
                 if (l >= 4 && ((addr & 3) == 0)) {
                     /* 32 bit write access */
                     val = ldl_p(buf);
diff --git a/gdbstub.c b/gdbstub.c
index 4ddf021e68..ef76fb06a4 100644
--- a/gdbstub.c
+++ b/gdbstub.c
@@ -47,6 +47,7 @@ enum RSState {
 static int gdbserver_fd = -1;
 
 typedef struct GDBState {
+    CPUState *env; /* current CPU */
     enum RSState state; /* parsing state */
     int fd;
     char line_buf[4096];
@@ -576,10 +577,10 @@ static void gdb_vm_stopped(void *opaque, int reason)
     int ret;
 
     /* disable single step if it was enable */
-    cpu_single_step(cpu_single_env, 0);
+    cpu_single_step(s->env, 0);
 
     if (reason == EXCP_DEBUG) {
-	tb_flush(cpu_single_env);
+	tb_flush(s->env);
         ret = SIGTRAP;
     }
     else
@@ -589,8 +590,9 @@ static void gdb_vm_stopped(void *opaque, int reason)
 }
 #endif
 
-static void gdb_read_byte(GDBState *s, CPUState *env, int ch)
+static void gdb_read_byte(GDBState *s, int ch)
 {
+    CPUState *env = s->env;
     int i, csum;
     char reply[1];
 
@@ -676,7 +678,7 @@ gdb_handlesig (CPUState *env, int sig)
           int i;
 
           for (i = 0; i < n; i++)
-            gdb_read_byte (s, env, buf[i]);
+            gdb_read_byte (s, buf[i]);
         }
       else if (n == 0 || errno != EAGAIN)
         {
@@ -721,7 +723,7 @@ static void gdb_read(void *opaque)
         vm_start();
     } else {
         for(i = 0; i < size; i++)
-            gdb_read_byte(s, cpu_single_env, buf[i]);
+            gdb_read_byte(s, buf[i]);
     }
 }
 
@@ -759,6 +761,7 @@ static void gdb_accept(void *opaque)
         return;
     }
 #endif
+    s->env = first_cpu; /* XXX: allow to change CPU */
     s->fd = fd;
 
     fcntl(fd, F_SETFL, O_NONBLOCK);
diff --git a/monitor.c b/monitor.c
index 18f0c89910..85a997db86 100644
--- a/monitor.c
+++ b/monitor.c
@@ -64,6 +64,8 @@ static int term_outbuf_index;
 
 static void monitor_start_input(void);
 
+CPUState *mon_cpu = NULL;
+
 void term_flush(void)
 {
     if (term_outbuf_index > 0) {
@@ -201,17 +203,69 @@ static void do_info_block(void)
     bdrv_info();
 }
 
+/* get the current CPU defined by the user */
+int mon_set_cpu(int cpu_index)
+{
+    CPUState *env;
+
+    for(env = first_cpu; env != NULL; env = env->next_cpu) {
+        if (env->cpu_index == cpu_index) {
+            mon_cpu = env;
+            return 0;
+        }
+    }
+    return -1;
+}
+
+CPUState *mon_get_cpu(void)
+{
+    if (!mon_cpu) {
+        mon_set_cpu(0);
+    }
+    return mon_cpu;
+}
+
 static void do_info_registers(void)
 {
+    CPUState *env;
+    env = mon_get_cpu();
+    if (!env)
+        return;
 #ifdef TARGET_I386
-    cpu_dump_state(cpu_single_env, NULL, monitor_fprintf,
+    cpu_dump_state(env, NULL, monitor_fprintf,
                    X86_DUMP_FPU);
 #else
-    cpu_dump_state(cpu_single_env, NULL, monitor_fprintf, 
+    cpu_dump_state(env, NULL, monitor_fprintf, 
                    0);
 #endif
 }
 
+static void do_info_cpus(void)
+{
+    CPUState *env;
+
+    /* just to set the default cpu if not already done */
+    mon_get_cpu();
+
+    for(env = first_cpu; env != NULL; env = env->next_cpu) {
+        term_printf("%c CPU #%d:", 
+                    (env == mon_cpu) ? '*' : ' ',
+                    env->cpu_index);
+#if defined(TARGET_I386)
+        term_printf(" pc=0x" TARGET_FMT_lx, env->eip + env->segs[R_CS].base);
+        if (env->cpu_halted)
+            term_printf(" (halted)");
+#endif
+        term_printf("\n");
+    }
+}
+
+static void do_cpu_set(int index)
+{
+    if (mon_set_cpu(index) < 0)
+        term_printf("Invalid CPU index\n");
+}
+
 static void do_info_jit(void)
 {
     dump_exec_info(NULL, monitor_fprintf);
@@ -381,6 +435,7 @@ static void term_printc(int c)
 static void memory_dump(int count, int format, int wsize, 
                         target_ulong addr, int is_physical)
 {
+    CPUState *env;
     int nb_per_line, l, line_size, i, max_digits, len;
     uint8_t buf[16];
     uint64_t v;
@@ -388,19 +443,22 @@ static void memory_dump(int count, int format, int wsize,
     if (format == 'i') {
         int flags;
         flags = 0;
+        env = mon_get_cpu();
+        if (!env && !is_physical)
+            return;
 #ifdef TARGET_I386
         if (wsize == 2) {
             flags = 1;
         } else if (wsize == 4) {
             flags = 0;
         } else {
-            /* as default we use the current CS size */
+                /* as default we use the current CS size */
             flags = 0;
-            if (!(cpu_single_env->segs[R_CS].flags & DESC_B_MASK))
+            if (env && !(env->segs[R_CS].flags & DESC_B_MASK))
                 flags = 1;
         }
 #endif
-        monitor_disas(addr, count, is_physical, flags);
+        monitor_disas(env, addr, count, is_physical, flags);
         return;
     }
 
@@ -437,7 +495,10 @@ static void memory_dump(int count, int format, int wsize,
         if (is_physical) {
             cpu_physical_memory_rw(addr, buf, l, 0);
         } else {
-            cpu_memory_rw_debug(cpu_single_env, addr, buf, l, 0);
+            env = mon_get_cpu();
+            if (!env)
+                break;
+            cpu_memory_rw_debug(env, addr, buf, l, 0);
         }
         i = 0; 
         while (i < l) {
@@ -776,10 +837,14 @@ static void print_pte(uint32_t addr, uint32_t pte, uint32_t mask)
 
 static void tlb_info(void)
 {
-    CPUState *env = cpu_single_env;
+    CPUState *env;
     int l1, l2;
     uint32_t pgd, pde, pte;
 
+    env = mon_get_cpu();
+    if (!env)
+        return;
+
     if (!(env->cr[0] & CR0_PG_MASK)) {
         term_printf("PG disabled\n");
         return;
@@ -830,10 +895,14 @@ static void mem_print(uint32_t *pstart, int *plast_prot,
 
 static void mem_info(void)
 {
-    CPUState *env = cpu_single_env;
+    CPUState *env;
     int l1, l2, prot, last_prot;
     uint32_t pgd, pde, pte, start, end;
 
+    env = mon_get_cpu();
+    if (!env)
+        return;
+
     if (!(env->cr[0] & CR0_PG_MASK)) {
         term_printf("PG disabled\n");
         return;
@@ -874,10 +943,15 @@ static void mem_info(void)
 static void do_info_kqemu(void)
 {
 #ifdef USE_KQEMU
+    CPUState *env;
     int val;
     val = 0;
-    if (cpu_single_env)
-        val = cpu_single_env->kqemu_enabled;
+    env = mon_get_cpu();
+    if (!env) {
+        term_printf("No cpu initialized yet");
+        return;
+    }
+    val = env->kqemu_enabled;
     term_printf("kqemu is %s\n", val ? "enabled" : "disabled");
 #else
     term_printf("kqemu support is not compiled\n");
@@ -934,6 +1008,8 @@ static term_cmd_t term_cmds[] = {
       "device", "add USB device (e.g. 'host:bus.addr' or 'host:vendor_id:product_id')" },
     { "usb_del", "s", do_usb_del,
       "device", "remove USB device 'bus.addr'" },
+    { "cpu", "i", do_cpu_set, 
+      "index", "set the default CPU" },
     { NULL, NULL, }, 
 };
 
@@ -946,6 +1022,8 @@ static term_cmd_t info_cmds[] = {
       "", "show the block devices" },
     { "registers", "", do_info_registers,
       "", "show the cpu registers" },
+    { "cpus", "", do_info_cpus,
+      "", "show infos for each CPU" },
     { "history", "", do_info_history,
       "", "show the command line history", },
     { "irq", "", irq_info,
@@ -989,63 +1067,85 @@ typedef struct MonitorDef {
 #if defined(TARGET_I386)
 static target_long monitor_get_pc (struct MonitorDef *md, int val)
 {
-    return cpu_single_env->eip + cpu_single_env->segs[R_CS].base;
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return env->eip + env->segs[R_CS].base;
 }
 #endif
 
 #if defined(TARGET_PPC)
 static target_long monitor_get_ccr (struct MonitorDef *md, int val)
 {
+    CPUState *env = mon_get_cpu();
     unsigned int u;
     int i;
 
+    if (!env)
+        return 0;
+
     u = 0;
     for (i = 0; i < 8; i++)
-	u |= cpu_single_env->crf[i] << (32 - (4 * i));
+	u |= env->crf[i] << (32 - (4 * i));
 
     return u;
 }
 
 static target_long monitor_get_msr (struct MonitorDef *md, int val)
 {
-    return (cpu_single_env->msr[MSR_POW] << MSR_POW) |
-        (cpu_single_env->msr[MSR_ILE] << MSR_ILE) |
-        (cpu_single_env->msr[MSR_EE] << MSR_EE) |
-        (cpu_single_env->msr[MSR_PR] << MSR_PR) |
-        (cpu_single_env->msr[MSR_FP] << MSR_FP) |
-        (cpu_single_env->msr[MSR_ME] << MSR_ME) |
-        (cpu_single_env->msr[MSR_FE0] << MSR_FE0) |
-        (cpu_single_env->msr[MSR_SE] << MSR_SE) |
-        (cpu_single_env->msr[MSR_BE] << MSR_BE) |
-        (cpu_single_env->msr[MSR_FE1] << MSR_FE1) |
-        (cpu_single_env->msr[MSR_IP] << MSR_IP) |
-        (cpu_single_env->msr[MSR_IR] << MSR_IR) |
-        (cpu_single_env->msr[MSR_DR] << MSR_DR) |
-        (cpu_single_env->msr[MSR_RI] << MSR_RI) |
-        (cpu_single_env->msr[MSR_LE] << MSR_LE);
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return (env->msr[MSR_POW] << MSR_POW) |
+        (env->msr[MSR_ILE] << MSR_ILE) |
+        (env->msr[MSR_EE] << MSR_EE) |
+        (env->msr[MSR_PR] << MSR_PR) |
+        (env->msr[MSR_FP] << MSR_FP) |
+        (env->msr[MSR_ME] << MSR_ME) |
+        (env->msr[MSR_FE0] << MSR_FE0) |
+        (env->msr[MSR_SE] << MSR_SE) |
+        (env->msr[MSR_BE] << MSR_BE) |
+        (env->msr[MSR_FE1] << MSR_FE1) |
+        (env->msr[MSR_IP] << MSR_IP) |
+        (env->msr[MSR_IR] << MSR_IR) |
+        (env->msr[MSR_DR] << MSR_DR) |
+        (env->msr[MSR_RI] << MSR_RI) |
+        (env->msr[MSR_LE] << MSR_LE);
 }
 
 static target_long monitor_get_xer (struct MonitorDef *md, int val)
 {
-    return (cpu_single_env->xer[XER_SO] << XER_SO) |
-        (cpu_single_env->xer[XER_OV] << XER_OV) |
-        (cpu_single_env->xer[XER_CA] << XER_CA) |
-        (cpu_single_env->xer[XER_BC] << XER_BC);
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return (env->xer[XER_SO] << XER_SO) |
+        (env->xer[XER_OV] << XER_OV) |
+        (env->xer[XER_CA] << XER_CA) |
+        (env->xer[XER_BC] << XER_BC);
 }
 
 static target_long monitor_get_decr (struct MonitorDef *md, int val)
 {
-    return cpu_ppc_load_decr(cpu_single_env);
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return cpu_ppc_load_decr(env);
 }
 
 static target_long monitor_get_tbu (struct MonitorDef *md, int val)
 {
-    return cpu_ppc_load_tbu(cpu_single_env);
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return cpu_ppc_load_tbu(env);
 }
 
 static target_long monitor_get_tbl (struct MonitorDef *md, int val)
 {
-    return cpu_ppc_load_tbl(cpu_single_env);
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return cpu_ppc_load_tbl(env);
 }
 #endif
 
@@ -1053,13 +1153,19 @@ static target_long monitor_get_tbl (struct MonitorDef *md, int val)
 #ifndef TARGET_SPARC64
 static target_long monitor_get_psr (struct MonitorDef *md, int val)
 {
-    return GET_PSR(cpu_single_env);
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return GET_PSR(env);
 }
 #endif
 
 static target_long monitor_get_reg(struct MonitorDef *md, int val)
 {
-    return cpu_single_env->regwptr[val];
+    CPUState *env = mon_get_cpu();
+    if (!env)
+        return 0;
+    return env->regwptr[val];
 }
 #endif
 
@@ -1269,6 +1375,7 @@ static void expr_error(const char *fmt)
     longjmp(expr_env, 1);
 }
 
+/* return 0 if OK, -1 if not found, -2 if no CPU defined */
 static int get_monitor_def(target_long *pval, const char *name)
 {
     MonitorDef *md;
@@ -1279,7 +1386,10 @@ static int get_monitor_def(target_long *pval, const char *name)
             if (md->get_value) {
                 *pval = md->get_value(md, md->offset);
             } else {
-                ptr = (uint8_t *)cpu_single_env + md->offset;
+                CPUState *env = mon_get_cpu();
+                if (!env)
+                    return -2;
+                ptr = (uint8_t *)env + md->offset;
                 switch(md->type) {
                 case MD_I32:
                     *pval = *(int32_t *)ptr;
@@ -1313,6 +1423,7 @@ static target_long expr_unary(void)
 {
     target_long n;
     char *p;
+    int ret;
 
     switch(*pch) {
     case '+':
@@ -1362,8 +1473,11 @@ static target_long expr_unary(void)
             while (isspace(*pch))
                 pch++;
             *q = 0;
-            if (get_monitor_def(&n, buf))
+            ret = get_monitor_def(&n, buf);
+            if (ret == -1)
                 expr_error("unknown register");
+            else if (ret == -2) 
+                expr_error("no cpu defined");
         }
         break;
     case '\0':
diff --git a/vl.c b/vl.c
index 725e954d3e..b946390a13 100644
--- a/vl.c
+++ b/vl.c
@@ -83,8 +83,6 @@
 
 #include "exec-all.h"
 
-//#define DO_TB_FLUSH
-
 #define DEFAULT_NETWORK_SCRIPT "/etc/qemu-ifup"
 
 //#define DEBUG_UNUSED_IOPORT
@@ -109,8 +107,6 @@
 
 const char *bios_dir = CONFIG_QEMU_SHAREDIR;
 char phys_ram_file[1024];
-CPUState *global_env;
-CPUState *cpu_single_env;
 void *ioport_opaque[MAX_IOPORTS];
 IOPortReadFunc *ioport_read_table[3][MAX_IOPORTS];
 IOPortWriteFunc *ioport_write_table[3][MAX_IOPORTS];
@@ -156,6 +152,7 @@ int usb_enabled = 0;
 USBPort *vm_usb_ports[MAX_VM_USB_PORTS];
 USBDevice *vm_usb_hub;
 static VLANState *first_vlan;
+int smp_cpus = 1;
 
 /***********************************************************/
 /* x86 ISA bus support */
@@ -427,16 +424,20 @@ int cpu_inl(CPUState *env, int addr)
 void hw_error(const char *fmt, ...)
 {
     va_list ap;
+    CPUState *env;
 
     va_start(ap, fmt);
     fprintf(stderr, "qemu: hardware error: ");
     vfprintf(stderr, fmt, ap);
     fprintf(stderr, "\n");
+    for(env = first_cpu; env != NULL; env = env->next_cpu) {
+        fprintf(stderr, "CPU #%d:\n", env->cpu_index);
 #ifdef TARGET_I386
-    cpu_dump_state(global_env, stderr, fprintf, X86_DUMP_FPU | X86_DUMP_CCOP);
+        cpu_dump_state(env, stderr, fprintf, X86_DUMP_FPU);
 #else
-    cpu_dump_state(global_env, stderr, fprintf, 0);
+        cpu_dump_state(env, stderr, fprintf, 0);
 #endif
+    }
     va_end(ap);
     abort();
 }
@@ -879,13 +880,16 @@ static void host_alarm_handler(int host_signum)
                            qemu_get_clock(vm_clock)) ||
         qemu_timer_expired(active_timers[QEMU_TIMER_REALTIME],
                            qemu_get_clock(rt_clock))) {
-        /* stop the cpu because a timer occured */
-        cpu_interrupt(global_env, CPU_INTERRUPT_EXIT);
+        CPUState *env = cpu_single_env;
+        if (env) {
+            /* stop the currently executing cpu because a timer occured */
+            cpu_interrupt(env, CPU_INTERRUPT_EXIT);
 #ifdef USE_KQEMU
-        if (global_env->kqemu_enabled) {
-            kqemu_cpu_interrupt(global_env);
-        }
+            if (env->kqemu_enabled) {
+                kqemu_cpu_interrupt(env);
+            }
 #endif
+        }
     }
 }
 
@@ -2970,9 +2974,6 @@ int qemu_loadvm(const char *filename)
         goto the_end;
     }
     for(;;) {
-#if defined (DO_TB_FLUSH)
-        tb_flush(global_env);
-#endif
         len = qemu_get_byte(f);
         if (feof(f))
             break;
@@ -3583,27 +3584,22 @@ void qemu_system_reset(void)
 void qemu_system_reset_request(void)
 {
     reset_requested = 1;
-    cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
+    if (cpu_single_env)
+        cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
 }
 
 void qemu_system_shutdown_request(void)
 {
     shutdown_requested = 1;
-    cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
+    if (cpu_single_env)
+        cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
 }
 
 void qemu_system_powerdown_request(void)
 {
     powerdown_requested = 1;
-    cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
-}
-
-static void main_cpu_reset(void *opaque)
-{
-#if defined(TARGET_I386) || defined(TARGET_SPARC)
-    CPUState *env = opaque;
-    cpu_reset(env);
-#endif
+    if (cpu_single_env)
+        cpu_interrupt(cpu_single_env, CPU_INTERRUPT_EXIT);
 }
 
 void main_loop_wait(int timeout)
@@ -3684,14 +3680,42 @@ void main_loop_wait(int timeout)
                         qemu_get_clock(rt_clock));
 }
 
+static CPUState *cur_cpu;
+
+static CPUState *find_next_cpu(void)
+{
+    CPUState *env;
+    env = cur_cpu;
+    for(;;) {
+        /* get next cpu */
+        env = env->next_cpu;
+        if (!env)
+            env = first_cpu;
+        if (!env->cpu_halted) 
+            break;
+        /* all CPUs are halted ? */
+        if (env == cur_cpu)
+            return NULL;
+    }
+    cur_cpu = env;
+    return env;
+}
+
 int main_loop(void)
 {
     int ret, timeout;
-    CPUState *env = global_env;
+    CPUState *env;
 
+    cur_cpu = first_cpu;
     for(;;) {
         if (vm_running) {
-            ret = cpu_exec(env);
+            /* find next cpu to run */
+            /* XXX: handle HLT correctly */
+            env = find_next_cpu();
+            if (!env)
+                ret = EXCP_HLT;
+            else
+                ret = cpu_exec(env);
             if (shutdown_requested) {
                 ret = EXCP_INTERRUPT;
                 break;
@@ -3774,7 +3798,7 @@ void help(void)
            "                connect the host TAP network interface to VLAN 'n' and use\n"
            "                the network script 'file' (default=%s);\n"
            "                use 'fd=h' to connect to an already opened TAP interface\n"
-           "-net socket[,vlan=n][,fd=x][,listen=[host]:port][,connect=host:port]\n"
+           "-net socket[,vlan=n][,fd=h][,listen=[host]:port][,connect=host:port]\n"
            "                connect the vlan 'n' to another VLAN using a socket connection\n"
 #endif
            "-net none       use it alone to have zero network devices; if no -net option\n"
@@ -3899,6 +3923,7 @@ enum {
     QEMU_OPTION_win2k_hack,
     QEMU_OPTION_usb,
     QEMU_OPTION_usbdevice,
+    QEMU_OPTION_smp,
 };
 
 typedef struct QEMUOption {
@@ -3965,6 +3990,7 @@ const QEMUOption qemu_options[] = {
     { "pidfile", HAS_ARG, QEMU_OPTION_pidfile },
     { "win2k-hack", 0, QEMU_OPTION_win2k_hack },
     { "usbdevice", HAS_ARG, QEMU_OPTION_usbdevice },
+    { "smp", HAS_ARG, QEMU_OPTION_smp },
     
     /* temporary options */
     { "usb", 0, QEMU_OPTION_usb },
@@ -4120,7 +4146,6 @@ int main(int argc, char **argv)
 #endif
     int i, cdrom_index;
     int snapshot, linux_boot;
-    CPUState *env;
     const char *initrd_filename;
     const char *hd_filename[MAX_DISKS], *fd_filename[MAX_FD];
     const char *kernel_filename, *kernel_cmdline;
@@ -4511,6 +4536,13 @@ int main(int argc, char **argv)
                         optarg);
                 usb_devices_index++;
                 break;
+            case QEMU_OPTION_smp:
+                smp_cpus = atoi(optarg);
+                if (smp_cpus < 1 || smp_cpus > 8) {
+                    fprintf(stderr, "Invalid number of CPUs\n");
+                    exit(1);
+                }
+                break;
             }
         }
     }
@@ -4659,15 +4691,8 @@ int main(int argc, char **argv)
         }
     }
 
-    /* init CPU state */
-    env = cpu_init();
-    global_env = env;
-    cpu_single_env = env;
-
-    register_savevm("timer", 0, 1, timer_save, timer_load, env);
-    register_savevm("cpu", 0, 3, cpu_save, cpu_load, env);
+    register_savevm("timer", 0, 1, timer_save, timer_load, NULL);
     register_savevm("ram", 0, 1, ram_save, ram_load, NULL);
-    qemu_register_reset(main_cpu_reset, global_env);
 
     init_ioports();
     cpu_calibrate_ticks();
diff --git a/vl.h b/vl.h
index d5efcd2e7d..1daae5042e 100644
--- a/vl.h
+++ b/vl.h
@@ -142,6 +142,7 @@ extern const char *keyboard_layout;
 extern int kqemu_allowed;
 extern int win2k_install_hack;
 extern int usb_enabled;
+extern int smp_cpus;
 
 /* XXX: make it dynamic */
 #if defined (TARGET_PPC)
@@ -429,6 +430,9 @@ int register_savevm(const char *idstr,
 void qemu_get_timer(QEMUFile *f, QEMUTimer *ts);
 void qemu_put_timer(QEMUFile *f, QEMUTimer *ts);
 
+void cpu_save(QEMUFile *f, void *opaque);
+int cpu_load(QEMUFile *f, void *opaque, int version_id);
+
 /* block.c */
 typedef struct BlockDriverState BlockDriverState;
 typedef struct BlockDriver BlockDriver;
@@ -774,6 +778,9 @@ int pit_get_out(PITState *pit, int channel, int64_t current_time);
 extern QEMUMachine pc_machine;
 extern QEMUMachine isapc_machine;
 
+void ioport_set_a20(int enable);
+int ioport_get_a20(void);
+
 /* ppc.c */
 extern QEMUMachine prep_machine;
 extern QEMUMachine core99_machine;