宋文武 <iyzsong@gmail.com> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> 白い熊＠相撲道 <guix-devel_gnu.org@sumou.com> skribis:
>>
>>> On 2015-03-26 22:24, ludo@gnu.org wrote:
>>>> I have no idea, but I would recommend running:
>>>>
>>>>   sudo strace -o log qemu-nbd -c /dev/nbd0 disk.qcow2
>>>>
>>>> and then find out the socket name in ‘log’ above “Failed to bind
>>>> socket”.
>>>
>>> Yeah, was exploring this... I'm attaching the strace log, I don't
>>> understand what's causing it... Any ideas?
>>
>> [...]
>>
>>> --- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=2063, si_status=0, si_utime=0, si_stime=0} ---
>>> read(3, "Failed to bind socket: No such f"..., 1024) = 49
>>> write(2, "Failed to bind socket: No such f"..., 49) = 49
>>
>> The problem seems to happen in a child process.  Can you add the ‘-f’
>> flag to the ‘strace’ command line so we have details about child
>> processes?
> I tried 'strace -f', then find out that qemu-nbd need access to "/var/lock".
> After "mkdir /var/lock", it works.

Thanks, I’ve committed this: