[Cbe-oss-dev] [PATCH] Fix sparsemem on Cell

[Christoph Hellwig]

On Mon, Dec 18, 2006 at 01:13:57PM -0800, Dave Hansen wrote:
> On Sat, 2006-12-16 at 17:03 +0900, KAMEZAWA Hiroyuki wrote:
> >  /* add this memory to iomem resource */
> >  static struct resource *register_memory_resource(u64 start, u64 size)
> >  {
> > @@ -273,10 +284,13 @@
> >  		if (ret)
> >  			goto error;
> >  	}
> > +	atomic_inc(&memory_hotadd_count);
> >  
> >  	/* call arch's memory hotadd */
> >  	ret = arch_add_memory(nid, start, size);
> >  
> > +	atomic_dec(&memory_hotadd_count);
> 
> I'd be willing to be that this will work just fine.  But, I think we can
> do it without any static state at all, if we just pass a runtime-or-not
> flag down into the arch_add_memory() call chain.
> 
> I'll code that up so we can compare to yours.  

Yes, I stronly concur that passing an explicit flag is much much better
than any hack involving global state.