All of lore.kernel.org
 help / color / mirror / Atom feed
From: Greg Banks <gnb@sgi.com>
To: "J. Bruce Fields" <bfields@fieldses.org>
Cc: Linux NFS ML <linux-nfs@vger.kernel.org>
Subject: [patch 09/14] sunrpc: Remove the global lock queue_io_mutex.
Date: Thu, 08 Jan 2009 19:25:19 +1100	[thread overview]
Message-ID: <20090108082604.299840000@sgi.com> (raw)
In-Reply-To: 20090108082510.050854000@sgi.com

This lock was used for two different purposes in two places.  The use
in cache_read() is redundant because cd->queue_lock is also taken
and can be removed without harm.  The use in cache_write() protects a
global buffer which can be eliminated and replaced with a temporarily
allocated page; this also removes an unnecessary global bottleneck
when multiple rpc.mountd threads are writing down into the kernel.

Signed-off-by: Greg Banks <gnb@sgi.com>
---

 net/sunrpc/cache.c |   34 +++++++++++++++-------------------
 1 file changed, 15 insertions(+), 19 deletions(-)

Index: bfields/net/sunrpc/cache.c
===================================================================
--- bfields.orig/net/sunrpc/cache.c
+++ bfields/net/sunrpc/cache.c
@@ -674,8 +674,6 @@ void cache_clean_deferred(void *owner)
  *
  */
 
-static DEFINE_MUTEX(queue_io_mutex);
-
 struct cache_queue {
 	struct list_head	list;
 	int			reader;	/* if 0, then request */
@@ -703,8 +701,6 @@ cache_read(struct file *filp, char __use
 	if (count == 0)
 		return 0;
 
-	mutex_lock(&queue_io_mutex); /* protect against multiple concurrent
-			      * readers on this file */
  again:
 	spin_lock(&cd->queue_lock);
 	/* need to find next request */
@@ -716,7 +712,6 @@ cache_read(struct file *filp, char __use
 	}
 	if (rp->q.list.next == &cd->queue) {
 		spin_unlock(&cd->queue_lock);
-		mutex_unlock(&queue_io_mutex);
 		BUG_ON(rp->offset);
 		return 0;
 	}
@@ -763,37 +758,38 @@ cache_read(struct file *filp, char __use
 	}
 	if (err == -EAGAIN)
 		goto again;
-	mutex_unlock(&queue_io_mutex);
 	return err ? err :  count;
 }
 
-static char write_buf[8192]; /* protected by queue_io_mutex */
-
 static ssize_t
 cache_write(struct file *filp, const char __user *buf, size_t count,
 	    loff_t *ppos)
 {
 	int err;
 	struct cache_detail *cd = PDE(filp->f_path.dentry->d_inode)->data;
+	char *tmp;
+	int tmp_size = PAGE_SIZE;
 
 	if (count == 0)
 		return 0;
-	if (count >= sizeof(write_buf))
+	if (count >= tmp_size)
 		return -EINVAL;
 
-	mutex_lock(&queue_io_mutex);
+	tmp = kmalloc(tmp_size, GFP_KERNEL);
+	if (!tmp)
+		return -ENOMEM;
 
-	if (copy_from_user(write_buf, buf, count)) {
-		mutex_unlock(&queue_io_mutex);
-		return -EFAULT;
-	}
-	write_buf[count] = '\0';
+	err = -EFAULT;
+	if (copy_from_user(tmp, buf, count))
+		goto out;
+
+	tmp[count] = '\0';
+	err = -EINVAL;
 	if (cd->cache_parse)
-		err = cd->cache_parse(cd, write_buf, count);
-	else
-		err = -EINVAL;
+		err = cd->cache_parse(cd, tmp, count);
 
-	mutex_unlock(&queue_io_mutex);
+out:
+	kfree(tmp);
 	return err ? err : count;
 }
 

--
-- 
Greg Banks, P.Engineer, SGI Australian Software Group.
the brightly coloured sporks of revolution.
I don't speak for SGI.

  parent reply	other threads:[~2009-01-08  8:26 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-01-08  8:25 [patch 00/14] sunrpc: Sunrpc cache cleanups and upcall rework Greg Banks
2009-01-08  8:25 ` [patch 01/14] sunrpc: Use consistent naming for variables of type struct cache_detail* Greg Banks
2009-01-08  8:25 ` [patch 02/14] sunrpc: Use consistent naming for variables of type struct cache_head* Greg Banks
2009-01-08  8:25 ` [patch 03/14] sunrpc: Use consistent naming for variables of type struct cache_request* Greg Banks
2009-01-08  8:25 ` [patch 04/14] sunrpc: Minor indentation cleanup in cache.c Greg Banks
2009-01-08  8:25 ` [patch 05/14] sunrpc: Rename queue_loose() to cache_remove_queued() Greg Banks
2009-01-08  8:25 ` [patch 06/14] sunrpc: Gather forward declarations of static functions in cache.c Greg Banks
2009-01-08  8:25 ` [patch 07/14] sunrpc: Make the global queue_lock per-cache-detail Greg Banks
2009-01-08  8:25 ` [patch 08/14] sunrpc: Make the global queue_wait per-cache-detail Greg Banks
2009-01-08  8:25 ` Greg Banks [this message]
2009-01-08  8:25 ` [patch 10/14] sunrpc: Reorganise the queuing of cache upcalls Greg Banks
2009-01-08 19:57   ` J. Bruce Fields
2009-01-09  2:40     ` Greg Banks
     [not found]       ` <4966B92F.8060008-cP1dWloDopni96+mSzHFpQC/G2K4zDHf@public.gmane.org>
2009-01-09  2:57         ` J. Bruce Fields
2009-01-09  3:12           ` Greg Banks
     [not found]             ` <4966C0AB.7000604-cP1dWloDopni96+mSzHFpQC/G2K4zDHf@public.gmane.org>
2009-01-09 16:53               ` Chuck Lever
2009-01-10  1:28                 ` Greg Banks
2009-01-09 21:29         ` J. Bruce Fields
2009-01-09 21:41           ` J. Bruce Fields
2009-01-09 23:40             ` Greg Banks
2009-01-09 23:29           ` Greg Banks
2009-01-08  8:25 ` [patch 11/14] sunrpc: Allocate cache_requests in a single allocation Greg Banks
2009-01-08  8:25 ` [patch 12/14] sunrpc: Centralise memory management of cache_requests Greg Banks
2009-01-08  8:25 ` [patch 13/14] sunrpc: Move struct cache_request to linux/sunrpc/cache.h Greg Banks
2009-01-08  8:25 ` [patch 14/14] sunrpc: Improve the usefulness of debug printks in the sunrpc cache code Greg Banks
2009-01-08 19:52 ` [patch 00/14] sunrpc: Sunrpc cache cleanups and upcall rework J. Bruce Fields
2009-01-09  1:42   ` Greg Banks

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=20090108082604.299840000@sgi.com \
    --to=gnb@sgi.com \
    --cc=bfields@fieldses.org \
    --cc=linux-nfs@vger.kernel.org \
    /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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.