On Sat, Jun 04, 2011 at 01:25:52AM +0100, Al Viro wrote:
> On Thu, Jun 02, 2011 at 05:01:01PM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@xxxxxxxxxx>
> > The inode unused list is currently a global LRU. This does not match
> > the other global filesystem cache - the dentry cache - which uses
> > per-superblock LRU lists. Hence we have related filesystem object
> > types using different LRU reclaimation schemes.
> > To enable a per-superblock filesystem cache shrinker, both of these
> > caches need to have per-sb unused object LRU lists. Hence this patch
> > converts the global inode LRU to per-sb LRUs.
> > The patch only does rudimentary per-sb propotioning in the shrinker
> > infrastructure, as this gets removed when the per-sb shrinker
> > callouts are introduced later on.
> What protects s_nr_inodes_unused?
For this patch, the modifications are protected by the
inode_lru_lock, but the reads are unprotected. That's the same
protection as the inode_stat.nr_unused field, and the same as the
existing dentry cache per-sb LRU accounting. In the next patch
modifcations are moved under the sb->s_inode_lru_lock, but reads
still remain unprotected.
I can see how the multiple reads in shrink_icache_sb() could each
return a different value during the proportioning, but I don't think
that is a big problem. That proportioning code goes away in the next
patch and is replaced by different code in prune_super(), so if you
want the reads protected by locks or a single snapshot used for the
proportioning calculations I'll do it in the new code in