linux系统除了能ping通,啥也不行，看到内核挂在函数top_machine_cpu_stop上了

mordorwww

#0  stop_machine_cpu_stop (data=0xdb7afefc) at kernel/stop_machine.c:427
#1  0xc10a4518 in cpu_stopper_thread (cpu=<value optimized out> at kernel/stop_machine.c:285
#2  0xc106053a in smpboot_thread_fn (data=0xf64026a0) at kernel/smpboot.c:160
#3  0xc105882c in kthread (_create=0xf649de7 at kernel/kthread.c:200
#4  0xc16e6797 in ?? () at arch/x86/kernel/entry_32.S:311
#5  0xc1058790 in ?? () at kernel/kthread.c:117


除了能ping通,啥也不行

/* This is the cpu_stop function which stops the CPU. */
static int stop_machine_cpu_stop(void *data)

mordorwww

427                     if (smdata->state != curstate) {
(gdb) bt
#0  0xc10a47c2 in stop_machine_cpu_stop (data=0xdb7afefc) at kernel/stop_machine.c:427
#1  0xc10a4518 in cpu_stopper_thread (cpu=<value optimized out> at kernel/stop_machine.c:285
#2  0xc106053a in smpboot_thread_fn (data=0xf64026a0) at kernel/smpboot.c:160
#3  0xc105882c in kthread (_create=0xf649de7 at kernel/kthread.c:200
#4  0xc16e6797 in ?? () at arch/x86/kernel/entry_32.S:311
#5  0xc1058790 in ?? () at kernel/kthread.c:117
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb) n
443             } while (curstate != STOPMACHINE_EXIT);
(gdb) n
426                     cpu_relax();
(gdb) n
427                     if (smdata->state != curstate) {
(gdb) n
443             } while (curstate != STOPMACHINE_EXIT);
(gdb) n
426                     cpu_relax();
(gdb) n
427                     if (smdata->state != curstate) {
(gdb) n
443             } while (curstate != STOPMACHINE_EXIT);
(gdb) n
426                     cpu_relax();
(gdb) n
427                     if (smdata->state != curstate) {
(gdb) n
443             } while (curstate != STOPMACHINE_EXIT);
(gdb) n
426                     cpu_relax();
(gdb) c
Continuing.

mordorwww

(gdb) c
Continuing.
^C
Program received signal SIGINT, Interrupt.
0xc16e0430 in ?? () at /root/li/linux-3.12.8/arch/x86/include/asm/entry_arch.h:24
24      BUILD_INTERRUPT3(kvm_posted_intr_ipi, POSTED_INTR_VECTOR,
(gdb) bt
#0  0xc16e0430 in ?? () at /root/li/linux-3.12.8/arch/x86/include/asm/entry_arch.h:24
#1  0xc10a47bf in rep_nop (data=0xdb7afefc) at /root/li/linux-3.12.8/arch/x86/include/asm/processor.h:656
#2  cpu_relax (data=0xdb7afefc) at /root/li/linux-3.12.8/arch/x86/include/asm/processor.h:661
#3  stop_machine_cpu_stop (data=0xdb7afefc) at kernel/stop_machine.c:426
#4  0xc10a4518 in cpu_stopper_thread (cpu=<value optimized out> at kernel/stop_machine.c:285
#5  0xc106053a in smpboot_thread_fn (data=0xf64026a0) at kernel/smpboot.c:160
#6  0xc105882c in kthread (_create=0xf649de7 at kernel/kthread.c:200
#7  0xc16e6797 in ?? () at arch/x86/kernel/entry_32.S:311
#8  0xc1058790 in ?? () at kernel/kthread.c:117
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
(gdb) n
37      BUILD_INTERRUPT(apic_timer_interrupt,LOCAL_TIMER_VECTOR)
(gdb) n
[New Thread 3]
[New Thread 2]
334             GET_THREAD_INFO(%ebp)
(gdb) n
336             movl PT_EFLAGS(%esp), %eax      # mix EFLAGS and CS
(gdb) n
337             movb PT_CS(%esp), %al
(gdb) n
338             andl $(X86_EFLAGS_VM | SEGMENT_RPL_MASK), %eax
(gdb) n
346             cmpl $USER_RPL, %eax
(gdb) n
347             jb resume_kernel                # not returning to v8086 or userspace
(gdb) n
533             movl PT_EFLAGS(%esp), %eax      # mix EFLAGS, SS and CS
(gdb) n
537             movb PT_OLDSS(%esp), %ah
(gdb) n
538             movb PT_CS(%esp), %al
(gdb) n
539             andl $(X86_EFLAGS_VM | (SEGMENT_TI_MASK << | SEGMENT_RPL_MASK), %eax
(gdb) n
540             cmpl $((SEGMENT_LDT << | USER_RPL), %eax
(gdb) n
542             je ldt_ss                       # returning to user-space with LDT SS
(gdb) n
544             RESTORE_REGS 4                  # skip orig_eax/error_code
(gdb) n
546             INTERRUPT_RETURN
(gdb) n
stop_machine_cpu_stop (data=0xdb7afefc) at kernel/stop_machine.c:427
427                     if (smdata->state != curstate) {
(gdb) n
443             } while (curstate != STOPMACHINE_EXIT);
(gdb) n
426                     cpu_relax();
(gdb) n
427                     if (smdata->state != curstate) {
(gdb) n
443             } while (curstate != STOPMACHINE_EXIT);
(gdb) c
Continuing.

humjb_1983

mordorwww 发表于 2014-01-23 09:37
#0  stop_machine_cpu_stop (data=0xdb7afefc) at kernel/stop_machine.c:427
#1  0xc10a4518 in cpu_stop ...

你这是启动过程中的报错么？具体的问题现象是啥？

mordorwww

humjb_1983 发表于 2014-01-23 10:51
你这是启动过程中的报错么？具体的问题现象是啥？

多核下单内核线程死循环

humjb_1983

mordorwww 发表于 2014-01-23 10:21
427                     if (smdata->state != curstate) {
(gdb) bt
#0  0xc10a47c2 in stop_machine_c ...

哪个版本的代码呢？我的几个版本中都没有smpboot_thread_fn 函数呢~

mordorwww

最新的内核 3.12.8

humjb_1983

mordorwww 发表于 2014-01-23 11:20
最新的内核 3.12.8

能否把kernel/stop_machine.c:427这行代码贴出来？
能否把smdata数据结构内容打出来。
理论上stop_machine是不会卡住的，除非smdata->fn函数卡住了。

humjb_1983

mordorwww 发表于 2014-01-23 11:20
最新的内核 3.12.8

另外，stop_machine的过程中应该是关中断的，不太可能ping通哦，这个还能确认下？

mordorwww

humjb_1983 发表于 2014-01-23 16:25
另外，stop_machine的过程中应该是关中断的，不太可能ping通哦，这个还能确认下？

确认确实如此。非抢占内核多核单内核线程死循环都是这种表现