Message-Id: <20110311102012.0901e551.kamezawa.hiroyu@jp.fujitsu.com>
Date:	Fri, 11 Mar 2011 10:20:12 +0900
From:	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
To:	Chris Mason <chris.mason@...cle.com>
Cc:	Vivek Goyal <vgoyal@...hat.com>,
	Andreas Dilger <adilger@...ger.ca>,
	Justin TerAvest <teravest@...gle.com>,
	m-ikeda <m-ikeda@...jp.nec.com>, jaxboe <jaxboe@...ionio.com>,
	linux-kernel <linux-kernel@...r.kernel.org>,
	ryov <ryov@...inux.co.jp>, taka <taka@...inux.co.jp>,
	"righi.andrea" <righi.andrea@...il.com>,
	guijianfeng <guijianfeng@...fujitsu.com>,
	balbir <balbir@...ux.vnet.ibm.com>,
	ctalbott <ctalbott@...gle.com>, nauman <nauman@...gle.com>,
	mrubin <mrubin@...gle.com>,
	linux-fsdevel <linux-fsdevel@...r.kernel.org>
Subject: Re: [RFC] Storing cgroup id in page->private (Was: Re: [RFC] [PATCH
 0/6] Provide cgroup isolation for buffered writes.)

On Thu, 10 Mar 2011 16:43:31 -0500
Chris Mason <chris.mason@...cle.com> wrote:

> Excerpts from Vivek Goyal's message of 2011-03-10 16:38:32 -0500:
> > On Thu, Mar 10, 2011 at 02:24:07PM -0700, Andreas Dilger wrote:
> > > On 2011-03-10, at 2:15 PM, Chris Mason wrote:
> > > > Excerpts from Vivek Goyal's message of 2011-03-10 14:41:06 -0500:
> > > >> On Thu, Mar 10, 2011 at 02:11:15PM -0500, Vivek Goyal wrote:
> > > >>>>> I think the person who dirtied the page can store the information in
> > > >>>>> page->private (assuming buffer heads were not generated) and if flusher
> > > >>>>> thread later ends up generating buffer heads and ends up modifying
> > > >>>>> page->private, this can be copied in buffer heads?
> > > >>>> 
> > > >>>> This scares me a bit.
> > > >>>> 
> > > >>>> As I understand it, fs/ code expects total ownership of page->private.
> > > >>>> This adds a responsibility for every user to copy the data through and
> > > >>>> store it in the buffer head (or anything else). btrfs seems to do
> > > >>>> something entirely different in some cases and store a different kind
> > > >>>> of value.
> > > >>> 
> > > >>> If filesystems are using page->private for some other purpose also, then
> > > >>> I guess we have issues. 
> > > >>> 
> > > >>> I am ccing linux-fsdevel to have some feedback on the idea of trying
> > > >>> to store cgroup id of page dirtying thread in page->private and/or buffer
> > > >>> head for tracking which group originally dirtied the page in IO controller
> > > >>> during writeback.
> > > >> 
> > > >> A quick "grep" showed that btrfs, ceph and logfs are using page->private
> > > >> for other purposes also.
> > > >> 
> > > >> I was under the impression that either page->private is null or it 
> > > >> points to buffer heads for the writeback case. So storing the info
> > > >> directly in either buffer head directly or first in page->private and
> > > >> then transferring it to buffer heads would have helped. 
> > > > 
> > > > Right, btrfs has its own uses for page->private, and we expect to own
> > > > it.  With a proper callback, the FS could store the extra information you
> > > > need in out own structs.
> > > 
> > > There is no requirement that page->private ever points to a buffer_head, and Lustre clients use it for its own tracking structure (never touching buffer_heads at all).  Any assumption about what a filesystem is storing in page->private in other parts of the code is just broken.
> > 
> > Andreas,
> > 
> > As Chris mentioned, will providing callbacks so that filesystem can
> > save/restore page->private be reasonable?
> 
> Just to clarify, I think saving/restoring page->private is going to be
> hard.  I'd rather just have a call back that says here's a page, storage
> this for the block io controller please, and another one that returns
> any previously stored info.
> 

Hmm, Vivek,
for dynamic allocation of io-record, how about this kind of tagging ?
(just an idea. not compiled at all.)

Pros.
- much better than consuming 2bytes for all pages including pages other 
  than file caches. 
- this will allow lockless lookup of iotag.
- setting iotag can be done at the same time PAGECACHE_TAG_DIRTY...
  no extra lock will be required.
- At clearing, we can expect lock for radix-tree is already held.
Cons.
- makes radix-tree struct larger and not good for cacheline.
- some special care? will be required at page-migration.


==
@@ -51,6 +51,9 @@ struct radix_tree_node {
        struct rcu_head rcu_head;
        void __rcu      *slots[RADIX_TREE_MAP_SIZE];
        unsigned long   tags[RADIX_TREE_MAX_TAGS][RADIX_TREE_TAG_LONGS];
+#ifdef CONFIG_BLK_CGROUP
+       unsigned short  iotag[RADIX_TREE_MAP_SIZE];
+#endif
 };

 struct radix_tree_path {
@@ -487,6 +490,36 @@ void *radix_tree_tag_set(struct radix_tr
 }
 EXPORT_SYMBOL(radix_tree_tag_set);

+#ifdef CONFIG_BLK_CGROUP
+void *radix_tree_iotag_set(struct radix_tree_root *root,
+                       unsigned long index, unsigned short tag)
+{
+       unsigned int height, shift;
+       struct radix_tree_node *node;
+
+       height = root->height;
+       BUG_ON(index > radix_tree_maxindex(height));
+
+       node = indirect_to_ptr(root->rnode);
+       shift = (height - 1) * RADIX_TREE_MAP_SHIFT;
+
+       while (height > 0) {
+               int offset;
+
+               offset = (index >> shift) & RADIX_TREE_MAP_MASK;
+               node = node->slots[offset];
+               BUG(!node);
+               shift -= RADIX_TREE_MAP_SHIFT;
+               height--;
+       }
+       node->iotag[offset] = tag;
+
+       return;
+}
+EXPORT_SYMBOL(radix_tree_iotag_set);
+

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

Powered by blists - more mailing lists