Re: [ckrm-tech] [PATCH 4/6] CKRM: Add guarantee support for mem controller

[Chandra Seetharaman <sekharan@us.ibm.com>]

On Thu, 2005-06-30 at 11:23 -0700, Dave Hansen wrote:
> On Thu, 2005-06-30 at 11:11 -0700, Chandra Seetharaman wrote:
> > On Thu, 2005-06-30 at 14:52 +0900, YAMAMOTO Takashi wrote:
> > > > +static inline void
> > > > +ckrm_clear_page_class(struct page *page)
> > > > +{
> > > > +	struct ckrm_zone *czone = page_ckrmzone(page);
> > > > +	if (czone == NULL)
> > > > +		return;
> > > > +	sub_use_count(czone->memcls, 0, page_zonenum(page), 1);
> > > > +	kref_put(&czone->memcls->nr_users, memclass_release);
> > > > +	set_page_ckrmzone(page, NULL);
> > > >  }
> > > 
> > > are you sure if it's safe?
> > > this function is called with zone->lock held,
> > > and memclass_release calls kfree.
> > 
> > i don't understand why you think it is a problem to call kfree with lock
> > held( i agree calling kmalloc with wait flag when a lock is being held
> > is not correct).
> 
> kfree->__cache_free
> 	->cache_flusharray
> 	->free_block
> 	->slab_destroy
> 	->kmem_freepages
> 	->free_pages
> 	->__free_pages
> 	->__free_pages_ok
> 	->free_pages_bulk:
> 
> 	spin_lock_irqsave(&zone->lock, flags);
> 
Oh.... I did not realize Takashi was mentioning zone->lock... i assumed
zone->lru_lock...

memory controller does not use zone->lock, it only uses zone->lru_lock..
I 'll look thru to see if that leads to a deadlock...

> Deadlock.
> 
> Whew!!!
> 
> -- Dave
> 
-- 

----------------------------------------------------------------------
    Chandra Seetharaman               | Be careful what you choose....
              - sekharan@us.ibm.com   |      .......you may get it.
----------------------------------------------------------------------


--
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:"aart@kvack.org"> aart@kvack.org </a>