LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: David Rientjes <rientjes@google.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Christoph Lameter <cl@linux-foundation.org>,
	Nick Piggin <npiggin@suse.de>, Paul Menage <menage@google.com>,
	Derek Fults <dfults@sgi.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [patch 1/7] cpusets: add dirty map to struct address_space
Date: Tue, 28 Oct 2008 18:13:38 -0700 (PDT)	[thread overview]
Message-ID: <alpine.DEB.1.10.0810281722290.23743@chino.kir.corp.google.com> (raw)
In-Reply-To: <alpine.DEB.1.10.0810281337530.6470@chino.kir.corp.google.com>

On Tue, 28 Oct 2008, David Rientjes wrote:

> Yeah, if we don't serialize with tree_lock then we'll need to protect the 
> attachment of mapping->dirty_nodes with a new spinlock in struct 
> address_space (and only for configs where MAX_NUMNODES > BITS_PER_LONG). 
> That locking overhead is negligible when mapping->dirty_nodes is non-NULL 
> since there's no requirement to protect the setting of the node in the 
> nodemask.
> 
> Are your concurrent pagecache patches in the latest mmotm?  If so, I can 
> rebase this entire patchset off that.

We're still taking mapping->tree_lock in both __set_page_dirty() and 
__set_page_dirty_nobuffers() in today's mmotm.

When tree_lock is removed with your patchset, we can add a spinlock to 
protect mapping->dirty_nodes when MAX_NUMNODES > BITS_PER_LONG.

Would you like to fold this patch into your series (which assumes we're 
not taking mapping->tree_lock in either of the two callers above)?
---
diff --git a/fs/inode.c b/fs/inode.c
--- a/fs/inode.c
+++ b/fs/inode.c
@@ -223,6 +223,9 @@ void inode_init_once(struct inode *inode)
 	INIT_LIST_HEAD(&inode->inotify_watches);
 	mutex_init(&inode->inotify_mutex);
 #endif
+#if MAX_NUMNODES > BITS_PER_LONG
+	spin_lock_init(&inode->i_data.dirty_nodes_lock);
+#endif
 }
 
 EXPORT_SYMBOL(inode_init_once);
diff --git a/include/linux/fs.h b/include/linux/fs.h
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -554,6 +554,7 @@ struct address_space {
 	nodemask_t		dirty_nodes;	/* nodes with dirty pages */
 #else
 	nodemask_t		*dirty_nodes;	/* pointer to mask, if dirty */
+	spinlock_t		dirty_nodes_lock; /* protects the above */
 #endif
 #endif
 } __attribute__((aligned(sizeof(long))));
diff --git a/kernel/cpuset.c b/kernel/cpuset.c
--- a/kernel/cpuset.c
+++ b/kernel/cpuset.c
@@ -2413,10 +2413,7 @@ EXPORT_SYMBOL_GPL(cpuset_mem_spread_node);
 #if MAX_NUMNODES > BITS_PER_LONG
 /*
  * Special functions for NUMA systems with a large number of nodes.  The
- * nodemask is pointed to from the address_space structure.  The attachment of
- * the dirty_nodes nodemask is protected by the tree_lock.  The nodemask is
- * freed only when the inode is cleared (and therefore unused, thus no locking
- * is necessary).
+ * nodemask is pointed to from the address_space structure.
  */
 void cpuset_update_dirty_nodes(struct address_space *mapping,
 			       struct page *page)
@@ -2424,14 +2421,18 @@ void cpuset_update_dirty_nodes(struct address_space *mapping,
 	nodemask_t *nodes = mapping->dirty_nodes;
 	int node = page_to_nid(page);
 
+	spin_lock_irq(&mapping->dirty_nodes_lock);
 	if (!nodes) {
 		nodes = kmalloc(sizeof(nodemask_t), GFP_ATOMIC);
-		if (!nodes)
+		if (!nodes) {
+			spin_unlock_irq(&mapping->dirty_nodes_lock);
 			return;
+		}
 
 		*nodes = NODE_MASK_NONE;
 		mapping->dirty_nodes = nodes;
 	}
+	spin_unlock_irq(&mapping->dirty_nodes_lock);
 	node_set(node, *nodes);
 }
 
@@ -2446,8 +2447,8 @@ void cpuset_clear_dirty_nodes(struct address_space *mapping)
 }
 
 /*
- * Called without tree_lock.  The nodemask is only freed when the inode is
- * cleared and therefore this is safe.
+ * The nodemask is only freed when the inode is cleared and therefore this
+ * requires no locking.
  */
 int cpuset_intersects_dirty_nodes(struct address_space *mapping,
 				  nodemask_t *mask)

  reply	other threads:[~2008-10-29  1:14 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-10-28 16:08 David Rientjes
2008-10-28 16:08 ` [patch 2/7] pdflush: allow the passing of a nodemask parameter David Rientjes
2008-10-28 16:08 ` [patch 3/7] mm: make page writeback obey cpuset constraints David Rientjes
2008-10-28 17:31   ` Peter Zijlstra
2008-10-28 19:16     ` David Rientjes
2008-10-28 17:32   ` Peter Zijlstra
2008-10-28 19:18     ` David Rientjes
2008-10-30  8:42       ` Peter Zijlstra
2008-10-30  9:10         ` David Rientjes
2008-10-30  9:34           ` Peter Zijlstra
2008-10-28 16:08 ` [patch 4/7] mm: cpuset aware reclaim writeout David Rientjes
2008-10-28 16:08 ` [patch 5/7] mm: throttle writeout with cpuset awareness David Rientjes
2008-10-28 16:08 ` [patch 6/7] cpusets: per cpuset dirty ratios David Rientjes
2008-10-30  6:59   ` Paul Menage
2008-10-30  8:48     ` David Rientjes
2008-10-30 15:28     ` Christoph Lameter
2008-10-30  8:44   ` Peter Zijlstra
2008-10-30  9:03     ` David Rientjes
2008-10-30  9:34       ` Peter Zijlstra
2008-10-30 10:02         ` David Rientjes
2008-10-28 16:08 ` [patch 7/7] cpusets: update documentation for writeback throttling David Rientjes
2008-10-30 16:06   ` Christoph Lameter
2008-10-28 17:37 ` [patch 1/7] cpusets: add dirty map to struct address_space Peter Zijlstra
2008-10-28 20:48   ` David Rientjes
2008-10-29  1:13     ` David Rientjes [this message]
2008-10-29  2:24       ` David Rientjes
2008-10-30  8:38       ` Peter Zijlstra
2008-10-28 17:46 ` Peter Zijlstra
2008-10-28 19:19   ` David Rientjes
2008-10-30 19:23 [patch 0/7] cpuset writeback throttling David Rientjes
2008-10-30 19:23 ` [patch 1/7] cpusets: add dirty map to struct address_space David Rientjes
2008-11-04 21:09   ` Andrew Morton
2008-11-04 21:20     ` Christoph Lameter
2008-11-04 21:42       ` Andrew Morton

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=alpine.DEB.1.10.0810281722290.23743@chino.kir.corp.google.com \
    --to=rientjes@google.com \
    --cc=akpm@linux-foundation.org \
    --cc=cl@linux-foundation.org \
    --cc=dfults@sgi.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=menage@google.com \
    --cc=npiggin@suse.de \
    --cc=peterz@infradead.org \
    --subject='Re: [patch 1/7] cpusets: add dirty map to struct address_space' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).