closure regarding udev

Jeremy Utley jeremy at jutley.org
Sun May 23 02:17:06 PDT 2004


Alexander E. Patrakov wrote:

> Nick Fotopoulos wrote:
>
>> On Sun, 2004-05-23 at 01:09 -0700, Jeremy Utley wrote:
>>
>>> Alexander E. Patrakov wrote:
>>>
>>>
>>>> There is one more argument to consider: the combination "udev " + 
>>>> "no hotplug" + "modular kernel" is inherently broken, and that's 
>>>> what we have in b6_0. I vote for either (1) or (3) or "(2)+ removal 
>>>> of module-ilit-tools".
>>>>
>>>
>>> I have to respectfully disagree.  When the module is loaded, udev 
>>> *should* still get called to create the device node.
>>>
>>> *Note* I have not tried this myself, however, but I hope this is the 
>>> case, otherwise my lpt zip drive is dead :)
>>>
>>> -J-
>>> -- 
>>
>>
>> I'm not sure about that one either, but in any case, the new bootscripts
>> can handle this situation with the createfiles script...so the 3 above
>> options should stand still as the available choices...
>
>
> How is the createfiles solution different from make_devices.sh?
>
createfiles can create any individual file needed at the time of boot.  
Right now, I use it to create the nvidia devices, because they don't 
export their info to sysfs yet, and vmware devices if I'm using it.

In response to your previous question, for my zip drive, I always 
modprobe the device by hand - I rarely use it, so I don't build it into 
the kernel.  I assume that when I load the module, and it detects the 
drive, it would send a hotplug event to udev for the device to be 
created, but I haven't tested that as of yet.

-J-




More information about the lfs-dev mailing list