[kbd] [systemd-devel] systemd-vconsole-setup fails very slowly

[Kay Sievers]

On Thu, Apr 2, 2015 at 12:00 AM, Andy Lutomirski <luto at amacapital.net> wrote:
> On Wed, Apr 1, 2015 at 2:47 PM, Kay Sievers <kay at vrfy.org> wrote:
>> On Wed, Apr 1, 2015 at 11:38 PM, Andy Lutomirski <luto at amacapital.net> wrote:
>>> On Wed, Apr 1, 2015 at 2:36 PM, Kay Sievers <kay at vrfy.org> wrote:
>>
>>>> They should only get created when something accesses the corresponding
>>>> tty. deallocvt(1) can kill unused ones and the device nodes should
>>>> disappear.
>>>>
>>>
>>> deallocvt doesn't seem to kill those device nodes for me.
>>
>> Seems to work here:
>>
>> # ls -l /dev/vcs[6789]
>> crw-rw---- 1 root tty 7, 6 Apr  1 22:21 /dev/vcs6
>> # cat /dev/tty7
>> ^C
>> # cat /dev/tty9
>> ^C
>> # ls -l /dev/vcs[6789]
>> crw-rw---- 1 root tty 7, 6 Apr  1 22:21 /dev/vcs6
>> crw-rw---- 1 root tty 7, 7 Apr  1 23:42 /dev/vcs7
>> crw-rw---- 1 root tty 7, 9 Apr  1 23:42 /dev/vcs9
>> # deallocvt 7
>> # ls -l /dev/vcs[6789]
>> crw-rw---- 1 root tty 7, 6 Apr  1 22:21 /dev/vcs6
>> crw-rw---- 1 root tty 7, 9 Apr  1 23:42 /dev/vcs9
>> # deallocvt 9
>> # ls -l /dev/vcs[6789]
>> crw-rw---- 1 root tty 7, 6 Apr  1 22:21 /dev/vcs6
>
> Aha.  It seems that I have something holding tty1-tty4 open.  I'll fix
> that on my end.  Will that make vconsole-setup stop calling setfont?

Oh, no. That was just in reply to the "vc_screen.c code seems to create those
devices unconditionally". These devices should all be fully dynamic,
handled inside the kernel. It will not influence the setfont behavior
of vconsole-setup.

To fix your issue, setfont should be changed, or we find and add some
dummy font-related ioctl to vconsole-setup, to check if setfont would
fail anyway on the current VT and we skip it.

Kay