From 61c4caf1e9a90a5ac023cb0d9443fdda8c88e4e5 Mon Sep 17 00:00:00 2001
From: Jens Axboe <jens.axboe@oracle.com>
Date: Tue, 6 Oct 2009 20:53:44 +0200
Subject: [PATCH] --- yaml --- r: 167433 b: refs/heads/master c:
 b9c8946b192397394a0ccd4fcecb31bc060f79f8 h: refs/heads/master i:   167431:
 58feae97dd1a120bbd8bfa6de44aeaa27e3bd71b v: v3

---
 [refs]                    | 2 +-
 trunk/block/cfq-iosched.c | 8 +++++++-
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/[refs] b/[refs]
index 0698afb6bdd9..d94e0bc5bb88 100644
--- a/[refs]
+++ b/[refs]
@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 0b182d617eb50762b483658dd6dd9a9fbcb25758
+refs/heads/master: b9c8946b192397394a0ccd4fcecb31bc060f79f8
diff --git a/trunk/block/cfq-iosched.c b/trunk/block/cfq-iosched.c
index 5c3cee93329a..4ab33d8a20b2 100644
--- a/trunk/block/cfq-iosched.c
+++ b/trunk/block/cfq-iosched.c
@@ -507,8 +507,14 @@ static void cfq_service_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq,
 		} else
 			rb_key += jiffies;
 	} else if (!add_front) {
+		/*
+		 * Get our rb key offset. Subtract any residual slice
+		 * value carried from last service. A negative resid
+		 * count indicates slice overrun, and this should position
+		 * the next service time further away in the tree.
+		 */
 		rb_key = cfq_slice_offset(cfqd, cfqq) + jiffies;
-		rb_key += cfqq->slice_resid;
+		rb_key -= cfqq->slice_resid;
 		cfqq->slice_resid = 0;
 	} else {
 		rb_key = -HZ;