cvs commit: LFS/bootscripts/lfs/init.d udev

Nathan Coulson conathan at conet.dyndns.org
Thu Jun 3 12:53:44 PDT 2004


> Nathan Coulson wrote:
>>> nathan at linuxfromscratch.org wrote:
>>>
>>>> nathan      04/06/02 19:59:32
>>>>
>>>> Modified:    bootscripts CHANGELOG bootscripts/contrib/init.d
>>>> hotplug bootscripts/lfs/init.d udev Log: Just reorganized the
>>>> error checking in udev and hotplug
>>>
>>> Now the initscripts are ready for the no-default-hotplug kernel
>>> patch. It will enter the book today after I visit a doctor (sorry
>>> for the delay).
>>
>> Can I Ignore your previous email then?
>
> Our mails state the same, so if you mean the following words by me,
> ignore them:
>
>> This "if" breaks creation of devices from the "modules" script. Udev
>> should unconditionally become a temporary handler for hotplug events
>> at this point and that should be (and is) overridden in the hotplug
>> initscript later.
>
> The reason is that the problem is already fixed by you. If you mean
> something else, please explain.

nope, that's what I meant.  I was looking over the changes Zack did, and
made a few more of my own.  Then went to email.

> However, I found one more problem with userspace readiness to handle
> hotplug events, and it's at shutdown. Please add the following line at
> the beginning of the "stop" section of the sendsignals script:
>
> echo >/proc/sys/kernel/hotplug
>
> Rationale: we are going to kill all processes before unmounting
> filesystems, but without this line we don't prevent them from starting
> again due to unexpected hotplug events.
>
> --
> Alexander E. Patrakov
> To get my address: echo '0!42!+/6 at 5-3.535.25' | tr [!-:] [a-z] | tr n .
> --

good line, but not sure if I want it on sendsignals...  Perhaps a stop
section on udev or something.

BTW, can you give me a quick listing on what daemons udev, and hotplug
start up?



More information about the lfs-book mailing list