Re: [PATCH] Register sysfs file for hotpluged new node

[Yasunori Goto <y-goto@jp.fujitsu.com>]

> On Tue, 2006-05-16 at 21:23 +0900, Yasunori Goto wrote:
> > +       /*
> > +        * register this node to sysfs.
> > +        * this is depends on topology. So each arch has its own.
> > +        */
> > +       if (new_pgdat){
> > +               ret = arch_register_node(nid);
> > +               BUG_ON(ret);
> > +       } 
> 
> Please don't do BUG_ON()s for things like this.  Memory hotplug _should_
> handle failures from top to bottom and not screw you over.  It isn't a
> crime or a bug to be out of memory.  

Basically, I would like to agree. 
But, there is no way to roll back from here now.
If online_node_map is set once, then new pgdat might be touched.
There is no way to disable them.

And I suppose it is not good thing that creating sysfs file of new node
before setting online_node_map. It means user interface is shown
before system initialization completion.

(In addition, remove_memory() is not yet....)

If return code of arch_register_node is ignored, 
cpu hotplug will work without new node's file.
When we tried cpu hotplug on it, it was cause of stack dump at last.

> Have you run this past the ppc maintainers?

Nope. I just tried cross compile.
I want powerpc box for test....


Thanks.

-- 
Yasunori Goto 


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>