On Tue 02-01-18 10:21:03, Mel Gorman wrote:
> On Sat, Dec 23, 2017 at 10:36:53AM +0900, Minchan Kim wrote:
> > > code path. It appears that similar situation is possible for them too.
> > >
> > > The file cache pages will be delete from file cache address_space before
> > > address_space (embedded in inode) is freed. But they will be deleted
> > > from LRU list only when its refcount dropped to zero, please take a look
> > > at put_page() and release_pages(). While address_space will be freed
> > > after putting reference to all file cache pages. If someone holds a
> > > reference to a file cache page for quite long time, it is possible for a
> > > file cache page to be in LRU list after the inode/address_space is
> > > freed.
> > >
> > > And I found inode/address_space is freed witch call_rcu(). I don't know
> > > whether this is related to page_mapping().
> > >
> > > This is just my understanding.
> >
> > Hmm, it smells like a bug of __isolate_lru_page.
> >
> > Ccing Mel:
> >
> > What locks protects address_space destroying when race happens between
> > inode trauncation and __isolate_lru_page?
> >
>
> I'm just back online and have a lot of catching up to do so this is a rushed
> answer and I didn't read the background of this. However the question is
> somewhat ambiguous and the scope is broad as I'm not sure which race you
> refer to. For file cache pages, I wouldnt' expect the address_space to be
> destroyed specifically as long as the inode exists which is the structure
> containing the address_space in this case. A page on the LRU being isolated
> in __isolate_lru_page will have an elevated reference count which will
> pin the inode until remove_mapping is called which holds the page lock
> while inode truncation looking at a page for truncation also only checks
> page_mapping under the page lock. Very broadly speaking, pages avoid being
> added back to an inode being freed by checking the I_FREEING state.
So I'm wondering what prevents the following:
CPU1 CPU2
truncate(inode) __isolate_lru_page()
...
truncate_inode_page(mapping, page);
delete_from_page_cache(page)
spin_lock_irqsave(&mapping->tree_lock, flags);
__delete_from_page_cache(page, NULL)
page_cache_tree_delete(..)
... mapping = page_mapping(page);
page->mapping = NULL;
...
spin_unlock_irqrestore(&mapping->tree_lock, flags);
page_cache_free_page(mapping, page)
put_page(page)
if (put_page_testzero(page)) -> false
- inode now has no pages and can be freed including embedded address_space
if (mapping && !mapping->a_ops->migratepage)
- we've dereferenced mapping which is potentially already free.
This all seems very theoretical but in principle possible...
Honza
--
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
