Re: [RFC 1/6] sysfs-kobject.patch

[Dipankar Sarma <dipankar@in.ibm.com>]

On Mon, Oct 06, 2003 at 09:16:40AM -0700, Greg KH wrote:
> On Mon, Oct 06, 2003 at 02:30:03PM +0530, Maneesh Soni wrote:
> > diff -puN include/linux/kobject.h~sysfs-kobject include/linux/kobject.h
> > --- linux-2.6.0-test6/include/linux/kobject.h~sysfs-kobject	2003-10-06 11:48:37.000000000 +0530
> > +++ linux-2.6.0-test6-maneesh/include/linux/kobject.h	2003-10-06 11:48:51.000000000 +0530
> > @@ -32,6 +32,12 @@ struct kobject {
> >  	struct kset		* kset;
> >  	struct kobj_type	* ktype;
> >  	struct dentry		* dentry;
> > + 	struct list_head	k_sibling;
> > + 	struct list_head	k_children;
> > +	struct list_head	attr;
> > +	struct list_head	attr_group;
> > +	struct rw_semaphore	k_rwsem;
> > +	char 			*k_symlink;
> >  };
> 
> Ouch.  Like Al said, this is too bloated.  Remember, not all kobjects

That is not what LowFree numbers after mounting sysfs says. Sure
you add some 48 bytes to kobject, but you are no longer pinning
256-byte(??) dentries and possibly bigger inodes for kobjects.
Doesn't that count ?

> are registered for use in sysfs.  This makes the overhead for such
> usages pretty high :(

The only way to confirm this is with numbers. Maneesh posted some
numbers that clearly show that memory usage is lower even with
the added fields in kobjects. Now, the question is are there
scenerios where kobject size increase in overhead. It will be nice
to have some numbers to demonstrate that.

Thanks
Dipankar