On Sat, 2008-03-22 at 00:25 +0900, Yasunori Goto wrote:
quoted text
> Badari-san.
> 
> > Index: linux-2.6.25-rc2/mm/memory_hotplug.c
> > ===================================================================
> > --- linux-2.6.25-rc2.orig/mm/memory_hotplug.c	2008-02-27 12:58:17.000000000 -0800
> > +++ linux-2.6.25-rc2/mm/memory_hotplug.c	2008-02-27 16:06:50.000000000 -0800
> > @@ -102,6 +102,21 @@ static int __add_section(struct zone *zo
> >  	return register_new_memory(__pfn_to_section(phys_start_pfn));
> >  }
> >  
> > +static int __remove_section(struct zone *zone, struct mem_section *ms)
> > +{
> > +	int ret = -EINVAL;
> > +
> > +	if (!valid_section(ms))
> > +		return ret;
> > +
> > +	ret = unregister_memory_section(ms);
> > +	if (ret)
> > +		return ret;
> > +
> > +	sparse_remove_one_section(zone, ms);
> > +	return 0;
> > +}
> > +
> >  /*
> >   * Reasonably generic function for adding memory.  It is
> >   * expected that archs that support memory hotplug will
> > @@ -135,6 +150,35 @@ int __add_pages(struct zone *zone, unsig
> >  }
> >  EXPORT_SYMBOL_GPL(__add_pages);
> >  
> > +int __remove_pages(struct zone *zone, unsigned long phys_start_pfn,
> > +		 unsigned long nr_pages)
> > +{
> > +	unsigned long i, ret = 0;
> > +	int sections_to_remove;
> > +	unsigned long flags;
> > +	struct pglist_data *pgdat = zone->zone_pgdat;
> > +
> > +	/*
> > +	 * We can only remove entire sections
> > +	 */
> > +	BUG_ON(phys_start_pfn & ~PAGE_SECTION_MASK);
> > +	BUG_ON(nr_pages % PAGES_PER_SECTION);
> > +
> > +	release_mem_region(phys_start_pfn << PAGE_SHIFT, nr_pages * PAGE_SIZE);
> > +
> > +	sections_to_remove = nr_pages / PAGES_PER_SECTION;
> > +	for (i = 0; i < sections_to_remove; i++) {
> > +		unsigned long pfn = phys_start_pfn + i*PAGES_PER_SECTION;
> > +		pgdat_resize_lock(pgdat, &flags);
> > +		ret = __remove_section(zone, __pfn_to_section(pfn));
> > +		pgdat_resize_unlock(pgdat, &flags);
> > +		if (ret)
> > +			break;
> > +	}
> > +	return ret;
> > +}
> > +EXPORT_SYMBOL_GPL(__remove_pages);
> 
> Here may be a bug.
> __remove_section() is called with pgdat_resize_lock() which is
> spin_lock_irqsave().
> 
>   __remove_section()
>       |
>       +--> unregister_memory_section()
>               |
>               +--> remove_memory_block()
>                       |
>                       +--> unregister_memory()
>                              |
>                              +--> sysdev_unregister()
> 
> sysdev_unregister() calls mutex_lock().
> It might sleep with irq disable, right?
> I found BUG()'s messages by this.

Hmm. I think you are right. Let me take a look
and see if we can move sysdev_unregister() outside
the lock.

Thanks for finding it.

Thanks,
Badari

--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/