Re: Regression - locking (all from 2.6.28)

[Catalin Marinas <catalin.marinas@arm.com>]

On Fri, 2009-03-06 at 09:26 -0800, Dave Hansen wrote:
> On Fri, 2009-03-06 at 17:18 +0000, Catalin Marinas wrote:
> > > > Are the pgdat->node_start_pfn and pgdat->node_spanned_pages always
> > > > valid? Thanks.
> > > 
> > > The variables themselves?  I'm sure there's a window in early boot where
> > > they aren't valid, but other than that they should be OK unless you're
> > > int the middle of a hotplug operation.
> > > 
> > > See pgdat_resize_(un)lock() in include/linux/memory_hotplug.h.
> > 
> > I wouldn't hold a lock for that long. It's not really critical to scan
> > all the page structures at a time as there are subsequent scans as well,
> > so some can be missed.
> 
> I think you should be more worried about consistency rather than missing
> entries.  Take these two lines of code:
> 
> 	start_pfn = node->node_start_pfn;
> 	/* hotplug occurs here */
> 	end_pfn = start_pfn + node->node_spanned_pages;
> 
> What if someone comes in and adds memory to the node, at the beginning
> of the node, after you have calculated start_pfn?  Try to think of what
> value you'll get for end_pfn and whether it is consistent and was *ever*
> valid at all.  Would that oops the kernel?

I assume pfn_valid() should handle this and kmemleak wouldn't scan the
page, unless we need locks around pfn_valid as well but I haven't seen
any used in the kernel.

-- 
Catalin