mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Chad Talbott <ctalbott@google.com>
To: jaxboe@fusionio.com, vgoyal@redhat.com
Cc: linux-kernel@vger.kernel.org, mrubin@google.com,
	teravest@google.com, Chad Talbott <ctalbott@google.com>
Subject: [PATCH 2/3] cfq-iosched: Fair cross-group preemption (implementation)
Date: Mon, 21 Mar 2011 18:10:44 -0700	[thread overview]
Message-ID: <1300756245-12380-3-git-send-email-ctalbott@google.com> (raw)
In-Reply-To: <1300756245-12380-1-git-send-email-ctalbott@google.com>

Add a new function cfq_group_should_preempt() which uses the new
service classes to decide if a task in one cgroup should preempt a
task in another.  We only allow it if the preempting task's vdisktime
is "behind" the preempted task.

Signed-off-by: Chad Talbott <ctalbott@google.com>
---
 block/cfq-iosched.c |   41 ++++++++++++++++++++++++++++++++++++++---
 1 files changed, 38 insertions(+), 3 deletions(-)

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 55e78b7..dfcce80 100644
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -1157,6 +1157,32 @@ void cfq_unlink_blkio_group(void *key, struct blkio_group *blkg)
 	spin_unlock_irqrestore(cfqd->queue->queue_lock, flags);
 }
 
+/*
+ * Return true if new_cfqg should preempt cfqg.  A return value of
+ * false means "don't care."  In that case, cfq has other heuristics
+ * to decide
+ */
+static bool
+cfq_group_should_preempt(struct cfq_queue *new_cfqq, struct cfq_queue *cfqq,
+			 struct request *rq)
+{
+	struct cfq_group *cfqg = cfqq->cfqg;
+	struct cfq_group *new_cfqg = new_cfqq->cfqg;
+	u64 grace_period;
+
+	/* in-group preemption is handled elsewhere */
+	if (new_cfqg == cfqg)
+		return false;
+
+	if (!(new_cfqg->class == BLKIO_RT_CLASS &&
+	      cfqg->class == BLKIO_BE_CLASS))
+		return false;
+
+	grace_period = cfq_scale_slice(cfqq->allocated_slice, cfqg);
+	return time_before64(new_cfqg->vdisktime,
+			     cfqg->vdisktime + grace_period);
+}
+
 #else /* GROUP_IOSCHED */
 static struct cfq_group *cfq_get_cfqg(struct cfq_data *cfqd, int create)
 {
@@ -1176,6 +1202,12 @@ cfq_link_cfqq_cfqg(struct cfq_queue *cfqq, struct cfq_group *cfqg) {
 static void cfq_release_cfq_groups(struct cfq_data *cfqd) {}
 static inline void cfq_put_cfqg(struct cfq_group *cfqg) {}
 
+static bool
+cfq_group_should_preempt(struct cfq_queue *new_cfqq, struct cfq_queue *cfqq,
+			 struct request *rq)
+{
+	return false;
+}
 #endif /* GROUP_IOSCHED */
 
 /*
@@ -3234,6 +3266,12 @@ cfq_should_preempt(struct cfq_data *cfqd, struct cfq_queue *new_cfqq,
 	if (!cfqq)
 		return false;
 
+	if (cfq_group_should_preempt(new_cfqq, cfqq, rq))
+		return true;
+
+	if (new_cfqq->cfqg != cfqq->cfqg)
+		return false;
+
 	if (cfq_class_idle(new_cfqq))
 		return false;
 
@@ -3253,9 +3291,6 @@ cfq_should_preempt(struct cfq_data *cfqd, struct cfq_queue *new_cfqq,
 	if (rq_is_sync(rq) && !cfq_cfqq_sync(cfqq))
 		return true;
 
-	if (new_cfqq->cfqg != cfqq->cfqg)
-		return false;
-
 	if (cfq_slice_used(cfqq))
 		return true;
 
-- 
1.7.3.1


  parent reply	other threads:[~2011-03-22  1:11 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-03-22  1:10 [PATCH 0/3] cfq-iosched: Fair cross-group preemption Chad Talbott
2011-03-22  1:10 ` [PATCH 1/3] cfq-iosched: Fair cross-group preemption (interface and documentation) Chad Talbott
2011-03-22 10:03   ` Gui Jianfeng
2011-03-22 18:07     ` Chad Talbott
2011-03-22  1:10 ` Chad Talbott [this message]
2011-03-22  1:10 ` [PATCH 3/3] cfq-iosched: Fair cross-group preemption (stats) Chad Talbott
2011-03-22 15:09 ` [PATCH 0/3] cfq-iosched: Fair cross-group preemption Vivek Goyal
2011-03-22 17:39   ` Chad Talbott
2011-03-22 18:12     ` Vivek Goyal
2011-03-22 23:46       ` Chad Talbott
2011-03-23  1:43         ` Vivek Goyal
2011-03-23 20:10       ` Chad Talbott
2011-03-23 20:41         ` Vivek Goyal
2011-03-24 21:47           ` Chad Talbott
2011-03-25  5:43             ` Gui Jianfeng
2011-03-25 21:32             ` Vivek Goyal
2011-03-25 23:53               ` Chad Talbott
2011-03-28 13:15                 ` Vivek Goyal
2011-03-28 16:59                   ` Chad Talbott
2011-03-28 17:24                     ` Vivek Goyal
2011-03-28 13:17                 ` Vivek Goyal
2011-03-28 17:02                   ` Chad Talbott

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=1300756245-12380-3-git-send-email-ctalbott@google.com \
    --to=ctalbott@google.com \
    --cc=jaxboe@fusionio.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mrubin@google.com \
    --cc=teravest@google.com \
    --cc=vgoyal@redhat.com \
    /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

Powered by JetHome