linux-nfs.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: NeilBrown <neilb@suse.de>
To: "J. Bruce Fields" <bfields@fieldses.org>
Cc: linux-nfs@vger.kernel.org
Subject: [PATCH 1/6] sunrpc/cache: allow threads to block while waiting for cache update.
Date: Thu, 12 Aug 2010 17:04:06 +1000	[thread overview]
Message-ID: <20100812070405.11459.91402.stgit@localhost.localdomain> (raw)
In-Reply-To: <20100812065722.11459.18978.stgit@localhost.localdomain>

The current practice of waiting for cache updates by queueing the
whole request to be retried has (at least) two problems.

1/ With NFSv4, requests can be quite complex and re-trying a whole
  request when a later part fails should only be a last-resort, not a
  normal practice.

2/ Large requests, and in particular any 'write' request, will not be
  queued by the current code and doing so would be undesirable.

In many cases only a very sort wait is needed before the cache gets
valid data.

So, providing the underlying transport permits it by setting
 ->thread_wait,
arrange to wait briefly for an upcall to be completed (as reflected in
the clearing of CACHE_PENDING).
If the short wait was not long enough and CACHE_PENDING is still set,
fall back on the old approach.

The 'thread_wait' value is set to 5 seconds when there are spare
threads, and 1 second when there are no spare threads.

These values are probably much higher than needed, but will ensure
some forward progress.

Note that as we only request an update for a non-valid item, and as
non-valid items are updated in place it is extremely unlikely that
cache_check will return -ETIMEDOUT.  Normally cache_defer_req will
sleep for a short while and then find that the item is_valid.


Signed-off-by: NeilBrown <neilb@suse.de>
---
 include/linux/sunrpc/cache.h |    3 +++
 net/sunrpc/cache.c           |   47 +++++++++++++++++++++++++++++++++++++++++-
 net/sunrpc/svc_xprt.c        |   11 ++++++++++
 3 files changed, 60 insertions(+), 1 deletions(-)

diff --git a/include/linux/sunrpc/cache.h b/include/linux/sunrpc/cache.h
index 7bf3e84..61f521f 100644
--- a/include/linux/sunrpc/cache.h
+++ b/include/linux/sunrpc/cache.h
@@ -125,6 +125,9 @@ struct cache_detail {
  */
 struct cache_req {
 	struct cache_deferred_req *(*defer)(struct cache_req *req);
+	int thread_wait;  /* How long (jiffies) we can block the
+			   * current thread to wait for updates.
+			   */
 };
 /* this must be embedded in a deferred_request that is being
  * delayed awaiting cache-fill
diff --git a/net/sunrpc/cache.c b/net/sunrpc/cache.c
index 2b06410..2fdd66b 100644
--- a/net/sunrpc/cache.c
+++ b/net/sunrpc/cache.c
@@ -509,10 +509,22 @@ static LIST_HEAD(cache_defer_list);
 static struct list_head cache_defer_hash[DFR_HASHSIZE];
 static int cache_defer_cnt;
 
+struct thread_deferred_req {
+	struct cache_deferred_req handle;
+	struct completion completion;
+};
+static void cache_restart_thread(struct cache_deferred_req *dreq, int too_many)
+{
+	struct thread_deferred_req *dr =
+		container_of(dreq, struct thread_deferred_req, handle);
+	complete(&dr->completion);
+}
+
 static int cache_defer_req(struct cache_req *req, struct cache_head *item)
 {
 	struct cache_deferred_req *dreq, *discard;
 	int hash = DFR_HASH(item);
+	struct thread_deferred_req sleeper;
 
 	if (cache_defer_cnt >= DFR_MAX) {
 		/* too much in the cache, randomly drop this one,
@@ -521,7 +533,15 @@ static int cache_defer_req(struct cache_req *req, struct cache_head *item)
 		if (net_random()&1)
 			return -ENOMEM;
 	}
-	dreq = req->defer(req);
+	if (req->thread_wait) {
+		dreq = &sleeper.handle;
+		sleeper.completion =
+			COMPLETION_INITIALIZER_ONSTACK(sleeper.completion);
+		dreq->revisit = cache_restart_thread;
+	} else
+		dreq = req->defer(req);
+
+ retry:
 	if (dreq == NULL)
 		return -ENOMEM;
 
@@ -555,6 +575,31 @@ static int cache_defer_req(struct cache_req *req, struct cache_head *item)
 		cache_revisit_request(item);
 		return -EAGAIN;
 	}
+
+	if (dreq == &sleeper.handle) {
+		wait_for_completion_interruptible_timeout(
+			&sleeper.completion, req->thread_wait);
+		spin_lock(&cache_defer_lock);
+		if (!list_empty(&sleeper.handle.hash)) {
+			list_del_init(&sleeper.handle.recent);
+			list_del_init(&sleeper.handle.hash);
+			cache_defer_cnt--;
+		}
+		spin_unlock(&cache_defer_lock);
+		if (test_bit(CACHE_PENDING, &item->flags)) {
+			/* item is still pending, try request
+			 * deferral
+			 */
+			dreq = req->defer(req);
+			goto retry;
+		}
+		/* only return success if we actually deferred the
+		 * request.  In this case we waited until it was
+		 * answered so no deferral has happened - rather
+		 * an answer already exists.
+		 */
+		return -EEXIST;
+	}
 	return 0;
 }
 
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index cbc0849..8ff6840 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -651,6 +651,11 @@ int svc_recv(struct svc_rqst *rqstp, long timeout)
 	if (signalled() || kthread_should_stop())
 		return -EINTR;
 
+	/* Normally we will wait up to 5 seconds for any required
+	 * cache information to be provided.
+	 */
+	rqstp->rq_chandle.thread_wait = 5*HZ;
+
 	spin_lock_bh(&pool->sp_lock);
 	xprt = svc_xprt_dequeue(pool);
 	if (xprt) {
@@ -658,6 +663,12 @@ int svc_recv(struct svc_rqst *rqstp, long timeout)
 		svc_xprt_get(xprt);
 		rqstp->rq_reserved = serv->sv_max_mesg;
 		atomic_add(rqstp->rq_reserved, &xprt->xpt_reserved);
+
+		/* As there is a shortage of threads and this request
+		 * had to be queue, don't allow the thread to wait so
+		 * long for cache updates.
+		 */
+		rqstp->rq_chandle.thread_wait = 1*HZ;
 	} else {
 		/* No data pending. Go to sleep */
 		svc_thread_enqueue(pool, rqstp);



  reply	other threads:[~2010-08-12  7:05 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-08-12  7:04 [PATCH 0/6] Cache deferral improvements - try N+2 NeilBrown
2010-08-12  7:04 ` NeilBrown [this message]
2010-08-12  7:04 ` [PATCH 2/6] nfsd/idmap: drop special request deferal in favour of improved default NeilBrown
     [not found]   ` <20100812070406.11459.89468.stgit-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob@public.gmane.org>
2010-09-21 21:08     ` J. Bruce Fields
2010-08-12  7:04 ` [PATCH 5/6] svcauth_gss: replace a trivial 'switch' with an 'if' NeilBrown
2010-08-12  7:04 ` [PATCH 4/6] nfsd: disable deferral for NFSv4 NeilBrown
     [not found]   ` <20100812070407.11459.2929.stgit-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob@public.gmane.org>
2010-09-21 21:01     ` J. Bruce Fields
2010-08-12  7:04 ` [PATCH 3/6] sunrpc: close connection when a request is irretrievably lost NeilBrown
2010-09-21 20:53   ` J. Bruce Fields
2010-09-21 23:37     ` Neil Brown
2010-09-22  2:13       ` J. Bruce Fields
2010-08-12  7:04 ` [PATCH 6/6] sunrpc/cache: change deferred-request hash table to use hlist NeilBrown
     [not found] ` <20100812065722.11459.18978.stgit-bi+AKbBUZKY6gyzm1THtWbp2dZbC/Bob@public.gmane.org>
2010-08-12 12:03   ` [PATCH 0/6] Cache deferral improvements - try N+2 J. Bruce Fields
2010-08-17  5:15 ` [PATCH 1.5/6] Fix race in new request delay code NeilBrown
2010-08-26 21:08   ` J. Bruce Fields
2010-08-29 23:36     ` Neil Brown
2010-09-01 11:31       ` J. Bruce Fields
2010-09-21  8:35       ` Neil Brown
2010-09-22  2:15         ` J. Bruce Fields

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=20100812070405.11459.91402.stgit@localhost.localdomain \
    --to=neilb@suse.de \
    --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 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).