首页
论坛
课程
招聘
[原创]QEMU/KVM虚拟机运行核心流程
2022-9-7 12:23 11602

[原创]QEMU/KVM虚拟机运行核心流程

2022-9-7 12:23
11602

1. 前言

这几天学习了虚拟机在创建和运行过程中,QEMU和KVM的核心执行流程。当然只是大概过程,并没有做到流程中的每个函数都分析。
很喜欢侯捷老师的一句话:源码之前,了无秘密。我阅读的源码是qemu-6.2.0和linux-5.15.39。

2. 用户层QEMU核心流程

编译安装qemu的过程很简单,参考官方文档就行。
可以直接用gdb命令行调试qemu,也可以vscode搭配gdb,调试属于基本能力,不多说。
动态调试qemu,并结合qemu源码分析流程。
启动参数如下:

1
2
3
4
5
6
7
8
9
$ ./qemu-system-x86_64 \
--enable-kvm \
-machine q35 \
-cpu host,+vmx \
-smp 1 \
-m 2048 \
-name ubuntu \
-hda /opt/vms/ubuntu.qcow2 \
-cdrom /opt/vms/ubuntu.iso

2.1 参数解析过程

qemu-6.2.0/softmmu/vl.c,line 2765
在vl.c文件2765行是入口,对运行程序传入的参数进行解析。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
void qemu_init(int argc, char **argv, char **envp)
{
    //...
    // 对参数进行解析
    for(;;) {
        if (optind >= argc)
            break;
        if (argv[optind][0] != '-') {
            loc_set_cmdline(argv, optind, 1);
            drive_add(IF_DEFAULT, 0, argv[optind++], HD_OPTS);
        } else {
            const QEMUOption *popt;
 
            popt = lookup_opt(argc, argv, &optarg, &optind);
            if (!(popt->arch_mask & arch_type)) {
                error_report("Option not supported for this target");
                exit(1);
            }
            switch(popt->index) {
            case QEMU_OPTION_cpu:
                /* hw initialization will check this */
                cpu_option = optarg;
                break;
            //...
            // 主要关注下面几个参数
            case QEMU_OPTION_m:
                opts = qemu_opts_parse_noisily(qemu_find_opts("memory"),
                                               optarg, true);
                if (!opts) {
                    exit(EXIT_FAILURE);
                }
                break;
            case QEMU_OPTION_enable_kvm:
                qdict_put_str(machine_opts_dict, "accel", "kvm");
                break;
            case QEMU_OPTION_M:
            case QEMU_OPTION_machine:
                {
                    bool help;
 
                    keyval_parse_into(machine_opts_dict, optarg, "type", &help, &error_fatal);
                    if (help) {
                        machine_help_func(machine_opts_dict);
                        exit(EXIT_SUCCESS);
                    }
                    break;
                }
            case QEMU_OPTION_smp:
                machine_parse_property_opt(qemu_find_opts("smp-opts"),
                                           "smp", optarg);
                break;
            }
        }
    }
    //...
    // 根据accel设置accelerators = kvm
    qemu_apply_legacy_machine_options(machine_opts_dict);
    qemu_apply_machine_options(machine_opts_dict);
 
    // 也会根据进程名判断可用的加速类型
    configure_accelerators(argv[0]);
    // 内部调用了do_configure_accelerator --> accel_init_machine
    // accel_init_machine --> kvm_init
    // 初始化具体的accel类(这里是kvm)
    // 在qemu-6.2.0/accel/kvm/kvm-all.c line 3629
    // 函数kvm_accel_class_init内部找到真正的初始化函数
    // ac->init_machine = kvm_init;
    //...
 
    // 在qmp_x_exit_preconfig与虚拟cpu创建有关
    if (!preconfig_requested) {
        qmp_x_exit_preconfig(&error_fatal);
    }
    qemu_init_displays();
    // 设置accel
    accel_setup_post(current_machine);
    os_setup_post();
    resume_mux_open();
}
参数 描述
QEMU_OPTION_m 虚拟机内存大小
QEMU_OPTION_enable_kvm 启用kvm加速
QEMU_OPTION_machine 虚拟机机器类型
QEMU_OPTION_smp 虚拟机cpu数量

2.2 虚拟机创建

qemu-6.2.0/accel/kvm/kvm-all.c line 2306
在kvm-all.c文件2306行是kvm_init
调用栈:

1
2
3
4
5
6
7
kvm_init(MachineState * ms) (qemu-6.2.0\accel\kvm\kvm-all.c:2308)
accel_init_machine(AccelState * accel, MachineState * ms) (qemu-6.2.0\accel\accel-softmmu.c:39)
do_configure_accelerator(void * opaque, QemuOpts * opts, Error ** errp) (qemu-6.2.0\softmmu\vl.c:2348)
qemu_opts_foreach(QemuOptsList * list, qemu_opts_loopfunc func, void * opaque, Error ** errp) (qemu-6.2.0\util\qemu-option.c:1135)
configure_accelerators(const char * progname) (qemu-6.2.0\softmmu\vl.c:2414)
qemu_init(int argc, char ** argv, char ** envp) (qemu-6.2.0\softmmu\vl.c:3724)
main(int argc, char ** argv, char ** envp) (qemu-6.2.0\softmmu\main.c:49)

主要函数kvm_init

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
static int kvm_init(MachineState *ms)
{
    MachineClass *mc = MACHINE_GET_CLASS(ms);
    static const char upgrade_note[] =
        "Please upgrade to at least kernel 2.6.29 or recent kvm-kmod\n"
        "(see http://sourceforge.net/projects/kvm).\n";
    //...
    QLIST_INIT(&s->kvm_parked_vcpus);
    // 开始使用kvm之前的标准流程
    // 打开设备/dev/kvm,检查kvm API版本
    // 保存了kvm设备描述符s->fd
    s->fd = qemu_open_old("/dev/kvm", O_RDWR);
    if (s->fd == -1) {
        fprintf(stderr, "Could not access KVM kernel module: %m\n");
        ret = -errno;
        goto err;
    }
 
    ret = kvm_ioctl(s, KVM_GET_API_VERSION, 0);
    if (ret < KVM_API_VERSION) {
        if (ret >= 0) {
            ret = -EINVAL;
        }
        fprintf(stderr, "kvm version too old\n");
        goto err;
    }
 
    if (ret > KVM_API_VERSION) {
        ret = -EINVAL;
        fprintf(stderr, "kvm version not supported\n");
        goto err;
    }
 
    //...
    // 创建虚拟机,保存虚拟机描述符s->vmfd
    do {
        ret = kvm_ioctl(s, KVM_CREATE_VM, type);
    } while (ret == -EINTR);
    //...
    s->vmfd = ret;
    //...
}

其主要功能是保存了kvm设备描述符s->fd,创建的虚拟机的描述符s->vmfd。

2.3 虚拟cpu创建

在2.1节中有提到,qmp_x_exit_preconfig函数与虚拟cpu的创建有关。
动态调试跟踪分析
qmp_x_exit_preconfig qemu-6.2.0\softmmu\vl.c:2740
--> qemu_init_board qemu-6.2.0\softmmu\vl.c:2652
--> machine_run_board_init qemu-6.2.0\hw\core\machine.c:1181
--> pc_q35_init qemu-6.2.0\hw\i386\pc_q35.c:182
--> x86_cpus_init qemu-6.2.0\hw\i386\x86.c:141
--> x86_cpu_new qemu-6.2.0\hw\i386\x86.c:114

 

在machine_run_board_init函数中根据参数中给的机器类型调用不同的pc_machine_init函数
machine_class->init(machine)----pc_q35_init

1
2
3
4
5
6
7
8
void x86_cpus_init(X86MachineState *x86ms, int default_cpu_version)
{
    //...
    // 根据参数smp的值,创建对应数量的虚拟cpu
    for (i = 0; i < ms->smp.cpus; i++) {
        x86_cpu_new(x86ms, possible_cpus->cpus[i].arch_id, &error_fatal);
    }
}

在x86_cpu_new中继续虚拟cpu的创建
x86_cpu_new
--> qdev_realize qemu-6.2.0\hw\core\qdev.c:333
--> device_set_realized qemu-6.2.0\hw\core\qdev.c:531
--> x86_cpu_realizefn qemu-6.2.0\target\i386\cpu.c:6447
--> qemu_init_vcpu qemu-6.2.0\softmmu\cpus.c:613
在x86_cpu_realizefn中调用qemu_init_vcpu对创建的虚拟cpu进行初始化

2.4 虚拟机运行

qemu-6.2.0/softmmu/cpus.c line 611
在cpus.c文件611行qemu_init_vcpu中初始化虚拟cpu,创建执行线程。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
void qemu_init_vcpu(CPUState *cpu)
{
    //...
    // 调用函数kvm_start_vcpu_thread创建虚拟cpu执行线程
    cpus_accel->create_vcpu_thread(cpu);
    //...
}
static void kvm_start_vcpu_thread(CPUState *cpu)
{
    //...
    // 线程函数kvm_vcpu_thread_fn
    qemu_thread_create(cpu->thread, thread_name, kvm_vcpu_thread_fn,
                       cpu, QEMU_THREAD_JOINABLE);
    //...
}
static void *kvm_vcpu_thread_fn(void *arg)
{
    //...
    // kvm_init_vcpu中通过kvm_vm_ioctl(s, KVM_CREATE_VCPU, (void *)vcpu_id)
    // 获取了vcpu描述符 cpu->kvm_fd = ret;
    r = kvm_init_vcpu(cpu, &error_fatal);
    kvm_init_cpu_signals(cpu);
 
    /* signal CPU creation */
    cpu_thread_signal_created(cpu);
    qemu_guest_random_seed_thread_part2(cpu->random_seed);
 
    // do while循环执行kvm_cpu_exec
    do {
        if (cpu_can_run(cpu)) {
            r = kvm_cpu_exec(cpu);
            if (r == EXCP_DEBUG) {
                cpu_handle_guest_debug(cpu);
            }
        }
        qemu_wait_io_event(cpu);
    } while (!cpu->unplug || cpu_can_run(cpu));
 
    kvm_destroy_vcpu(cpu);
    cpu_thread_signal_destroyed(cpu);
    qemu_mutex_unlock_iothread();
    rcu_unregister_thread();
    return NULL;
}
int kvm_cpu_exec(CPUState *cpu)
{
    //...
    do {
        // kvm_vcpu_ioctl(cpu, KVM_RUN, 0)
        // 从这里进入kvm内核阶段,开始运行虚拟机
        run_ret = kvm_vcpu_ioctl(cpu, KVM_RUN, 0);
        //...
        // 根据退出原因,分发处理
        switch (run->exit_reason) {
        case KVM_EXIT_IO:
            DPRINTF("handle_io\n");
            /* Called outside BQL */
            kvm_handle_io(run->io.port, attrs,
                          (uint8_t *)run + run->io.data_offset,
                          run->io.direction,
                          run->io.size,
                          run->io.count);
            ret = 0;
            break;
        default:
            DPRINTF("kvm_arch_handle_exit\n");
            ret = kvm_arch_handle_exit(cpu, run);
            break;
        }
    } while (ret == 0);
    cpu_exec_end(cpu);
    //...
    qatomic_set(&cpu->exit_request, 0);
    return ret;
}

虚拟机的运行就是kvm_cpu_exec中的do()while(ret == 0)的循环,该循环体中主要通过KVM_RUN启动虚拟机,进入了kvm的内核处理阶段,并等待返回结果。
当虚拟机退出,会根据返回的原因进行相应处理,最后将处理结果返回。
而kvm_cpu_exec自身也处于vcpu线程函数kvm_vcpu_thread_fn的循环当中,所以虚拟机的运行就是在这两个循环中不断进行。

2.5 用户层QEMU流程小结

解析参数,创建虚拟机,创建虚拟cpu,并获取三个最主要的描述符kvmfd、vmfd以及vcpufd。
根据vcpu数量创建具体的执行线程。
在线程中通过KVM_RUN启动虚拟机,进入内核KVM的处理流程。
重复循环KVM_RUN阶段。

3. 内核层KVM核心流程

在用户层QEMU阶段有提到通过函数kvm_vcpu_ioctl(cpu, KVM_RUN, 0)进入到内核KVM处理阶段。

3.1 运行guest的准备过程

linux-5.15.39/virt/kvm/kvm_main.c,line 3764
在kvm_main.c文件3764行找到内核中实际的kvm_vcpu_ioctl函数。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
static long kvm_vcpu_ioctl(struct file *filp,
               unsigned int ioctl, unsigned long arg)
{
    //...
    switch (ioctl) {
    case KVM_RUN: {
        //...
        // 根据KVM_RUN,调用kvm_arch_vcpu_ioctl_run
        r = kvm_arch_vcpu_ioctl_run(vcpu);
        trace_kvm_userspace_exit(vcpu->run->exit_reason, r);
        break;
    }
    //...
    }
out:
    mutex_unlock(&vcpu->mutex);
    kfree(fpu);
    kfree(kvm_sregs);
    return r;
}
// arch/x86/kvm/x86.c line 10103
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
{
    //...
    if (kvm_run->immediate_exit)
        r = -EINTR;
    else
        r = vcpu_run(vcpu);
    //...
    return r;
}
// arch/x86/kvm/x86.c line 9923
static int vcpu_run(struct kvm_vcpu *vcpu)
{
    int r;
    struct kvm *kvm = vcpu->kvm;
    //...
    for (;;) {
        if (kvm_vcpu_running(vcpu)) {
            // 进入guest模式的入口
            r = vcpu_enter_guest(vcpu);
        } else {
            r = vcpu_block(kvm, vcpu);
        }
        // 当返回值r<=0时,退出循环,一步步返回到用户层QEMU处理
        // 当返回值r>0时,继续guest运行循环
        if (r <= 0)
            break;
    }
    //...
    return r;
}
// arch/x86/kvm/x86.c line 9532
static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
{
    int r;
    // 一系列kvm_check_request函数调用
    // 检查guest请求
    //...
    // guest内存管理单元
    r = kvm_mmu_reload(vcpu);
 
    //...
    // 禁用内核抢占
    preempt_disable();
    //...
 
    vcpu->mode = IN_GUEST_MODE;
    // exit_fastpath = static_call(kvm_x86_run)(vcpu);
    // 调用架构相关的run函数进入guest模式运行
    for (;;) {
        exit_fastpath = static_call(kvm_x86_run)(vcpu);
        //...
        break;
    }
    // 能走到这里标志已退出guest模式
    vcpu->mode = OUTSIDE_GUEST_MODE;
    //...
    // 启用内核抢占
    preempt_enable();
 
    // 调用架构相关kvm_x86_handle_exit函数
    // 根据具体退出原因进行处理
    r = static_call(kvm_x86_handle_exit)(vcpu, exit_fastpath);
    return r;
}

那调用流程就是
kvm_vcpu_ioctl --> kvm_arch_vcpu_ioctl_run
--> vcpu_run --> vcpu_enter_guest
--> static_call(kvm_x86_run)(vcpu)

3.2 Guest的进入

在arch/x86/kvm/vmx/vmx.c line 7584
定义了一系列架构相关的操作函数
关注运行相关的
.run = vmx_vcpu_run,

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
// arch/x86/kvm/vmx/vmx.c line 6628
static fastpath_t vmx_vcpu_run(struct kvm_vcpu *vcpu)
{
    // 检查和准备工作
    //...
    vmx_vcpu_enter_exit(vcpu, vmx);
    //...
}
// arch/x86/kvm/vmx/vmx.c line 6606
static noinstr void vmx_vcpu_enter_exit(struct kvm_vcpu *vcpu,
                    struct vcpu_vmx *vmx)
{
    //...
    // arch/x86/kvm/vmx/vmenter.S汇编
    vmx->fail = __vmx_vcpu_run(vmx, (unsigned long *)&vcpu->arch.regs,
                   vmx->loaded_vmcs->launched);
    //...
}
1
2
3
4
5
6
7
arch/x86/kvm/vmx/vmenter.S
1.保存host状态
2.加载guest状态
3.进入guest模式: call vmx_vmenter
cpu从ROOT模式切换至NON-ROOT模式,进入到guest的世界运行
4.发生VM Exit时,保存guest状态,加载host状态
 cpu从NON-ROOT模式切换至ROOT模式,返回到host的世界

3.3 Guest的退出处理

在arch/x86/kvm/vmx/vmx.c line 7584
定义了一系列架构相关的操作函数
关注退出处理相关的
.handle_exit = vmx_handle_exit,

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
static int vmx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t exit_fastpath)
{
    int ret = __vmx_handle_exit(vcpu, exit_fastpath);
    //...
    return ret;
}
static int __vmx_handle_exit(struct kvm_vcpu *vcpu, fastpath_t exit_fastpath)
{
    //...
    exit_handler_index = array_index_nospec((u16)exit_reason.basic,
                        kvm_vmx_max_exit_handlers);
    return kvm_vmx_exit_handlers[exit_handler_index](vcpu);
 
}
 // 退出处理例程返回<=0,表示异常需要到用户层qemu进行进一步处理
 // 退出处理例程返回值>0,表示内核层已经处理完,可继续执行
static int (*kvm_vmx_exit_handlers[])(struct kvm_vcpu *vcpu) = {
    [EXIT_REASON_EXCEPTION_NMI]           = handle_exception_nmi,
    [EXIT_REASON_EXTERNAL_INTERRUPT]      = handle_external_interrupt,
    [EXIT_REASON_TRIPLE_FAULT]            = handle_triple_fault,
    [EXIT_REASON_NMI_WINDOW]          = handle_nmi_window,
    [EXIT_REASON_IO_INSTRUCTION]          = handle_io,
    [EXIT_REASON_CR_ACCESS]               = handle_cr,
    [EXIT_REASON_DR_ACCESS]               = handle_dr,
    [EXIT_REASON_CPUID]                   = kvm_emulate_cpuid,
    [EXIT_REASON_MSR_READ]                = kvm_emulate_rdmsr,
    [EXIT_REASON_MSR_WRITE]               = kvm_emulate_wrmsr,
    [EXIT_REASON_INTERRUPT_WINDOW]        = handle_interrupt_window,
    [EXIT_REASON_HLT]                     = kvm_emulate_halt,
    [EXIT_REASON_INVD]              = kvm_emulate_invd,
    [EXIT_REASON_INVLPG]              = handle_invlpg,
    [EXIT_REASON_RDPMC]                   = kvm_emulate_rdpmc,
    [EXIT_REASON_VMCALL]                  = kvm_emulate_hypercall,
    [EXIT_REASON_VMCLEAR]              = handle_vmx_instruction,
    [EXIT_REASON_VMLAUNCH]              = handle_vmx_instruction,
    [EXIT_REASON_VMPTRLD]              = handle_vmx_instruction,
    [EXIT_REASON_VMPTRST]              = handle_vmx_instruction,
    [EXIT_REASON_VMREAD]              = handle_vmx_instruction,
    [EXIT_REASON_VMRESUME]              = handle_vmx_instruction,
    [EXIT_REASON_VMWRITE]              = handle_vmx_instruction,
    [EXIT_REASON_VMOFF]              = handle_vmx_instruction,
    [EXIT_REASON_VMON]              = handle_vmx_instruction,
    [EXIT_REASON_TPR_BELOW_THRESHOLD]     = handle_tpr_below_threshold,
    [EXIT_REASON_APIC_ACCESS]             = handle_apic_access,
    [EXIT_REASON_APIC_WRITE]              = handle_apic_write,
    [EXIT_REASON_EOI_INDUCED]             = handle_apic_eoi_induced,
    [EXIT_REASON_WBINVD]                  = kvm_emulate_wbinvd,
    [EXIT_REASON_XSETBV]                  = kvm_emulate_xsetbv,
    [EXIT_REASON_TASK_SWITCH]             = handle_task_switch,
    [EXIT_REASON_MCE_DURING_VMENTRY]      = handle_machine_check,
    [EXIT_REASON_GDTR_IDTR]              = handle_desc,
    [EXIT_REASON_LDTR_TR]              = handle_desc,
    [EXIT_REASON_EPT_VIOLATION]          = handle_ept_violation,
    [EXIT_REASON_EPT_MISCONFIG]           = handle_ept_misconfig,
    [EXIT_REASON_PAUSE_INSTRUCTION]       = handle_pause,
    [EXIT_REASON_MWAIT_INSTRUCTION]          = kvm_emulate_mwait,
    [EXIT_REASON_MONITOR_TRAP_FLAG]       = handle_monitor_trap,
    [EXIT_REASON_MONITOR_INSTRUCTION]     = kvm_emulate_monitor,
    [EXIT_REASON_INVEPT]                  = handle_vmx_instruction,
    [EXIT_REASON_INVVPID]                 = handle_vmx_instruction,
    [EXIT_REASON_RDRAND]                  = kvm_handle_invalid_op,
    [EXIT_REASON_RDSEED]                  = kvm_handle_invalid_op,
    [EXIT_REASON_PML_FULL]              = handle_pml_full,
    [EXIT_REASON_INVPCID]                 = handle_invpcid,
    [EXIT_REASON_VMFUNC]              = handle_vmx_instruction,
    [EXIT_REASON_PREEMPTION_TIMER]          = handle_preemption_timer,
    [EXIT_REASON_ENCLS]              = handle_encls,
    [EXIT_REASON_BUS_LOCK]                = handle_bus_lock_vmexit,
};

3.4 内核层KVM流程小结

进入guest世界的准备工作。
正式进入guest执行。
根据guest退出原因进行处理,KVM先自行处理,
若kvm不能完全处理,则返回到用户层由QEMU处理。
QEMU处理后再次通过KVM_RUN进入到内核KVM流程。

4. 参考资料

.《Intel® Volume 3 System Programming Guide》
.《系统虚拟化:原理与实现》
.《处理器虚拟化技术》
. https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.15.39.tar.xz
. https://download.qemu.org/qemu-6.2.0.tar.xz


[2022冬季班]《安卓高级研修班(网课)》月薪两万班招生中~

收藏
点赞10
打赏
分享
最新回复 (8)
雪    币: 38
活跃值: 活跃值 (1075)
能力值: ( LV2,RANK:10 )
在线值:
发帖
回帖
粉丝
lukarl 活跃值 2022-9-9 14:44
2
0
感谢分享
雪    币: 3528
活跃值: 活跃值 (4365)
能力值: ( LV6,RANK:96 )
在线值:
发帖
回帖
粉丝
VirtualCC 活跃值 2022-9-9 15:10
3
0
路总牛逼!
雪    币: 252
活跃值: 活跃值 (1732)
能力值: ( LV3,RANK:20 )
在线值:
发帖
回帖
粉丝
杰克王 活跃值 2022-9-9 15:12
4
0
路总牛逼!
雪    币: 21373
活跃值: 活跃值 (2515)
能力值: ( LV4,RANK:41 )
在线值:
发帖
回帖
粉丝
_重黎 活跃值 2022-9-9 15:14
5
0
路总牛逼!
雪    币: 5
活跃值: 活跃值 (287)
能力值: ( LV2,RANK:10 )
在线值:
发帖
回帖
粉丝
wx_G-H-Z℡ 活跃值 2022-9-10 10:18
6
0
路总牛逼!
雪    币: 9877
活跃值: 活跃值 (2053)
能力值: ( LV6,RANK:87 )
在线值:
发帖
回帖
粉丝
Lixinist 活跃值 1 2022-9-11 16:43
7
0
路总牛逼!
雪    币: 58
活跃值: 活跃值 (192)
能力值: ( LV5,RANK:70 )
在线值:
发帖
回帖
粉丝
hhstudy 活跃值 1 2022-11-2 10:46
8
0
Mark
雪    币: 1804
活跃值: 活跃值 (1741)
能力值: ( LV4,RANK:50 )
在线值:
发帖
回帖
粉丝
fengyunabc 活跃值 1 2022-11-2 11:27
9
0
感谢分享!
游客
登录 | 注册 方可回帖
返回