Re: [Xen-devel] [PATCH] [RFC] sysfs support for xen linux

[Keir Fraser <Keir.Fraser@xxxxxxxxxxxx>]

On 10 Jan 2006, at 12:58, Mike D. Day wrote:

> No, they don't. Drivers get sysfs attributes by using the driver core 
> (registering a subsystem, defining attributes, etc.).
> However, if all you want to do is create and remove files under /sys 
> (without using all the driver core), there is no simple way to do so 
> (at least not that I could find). I am assuming that folks will want 
> something that works like /proc does now.
> I wrote xen_sysfs.c to provide simple sysfs interfaces for 
> non-drivers. It seems as though it would be a crime against nature to 
> write a device driver just to create a file under /sys.
Well, here's a question, and I really don't know the answer: It of 
course makes sense that the kernel maintainers want drivers to install 
themselves under /sys, and fit in with the whole kobject and hotplug 
infrastructure. But, for a few odds-and-ends special files that don't 
really relate to a device, is /proc also out of bounds these days? 
Seems to me that the kernel proc interfaces were designed to have a few 
random files thrown at them, in a way that the sysfs interfaces aren't. 
If the argument is really that random special files are a bad idea, 
that would continue to hold regardless of whether we move e.g., 
/proc/xen/privcmd to /sys.
 -- Keir


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel