public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Tejun Heo <tj@kernel.org>
To: axboe@kernel.dk
Cc: shli@kernel.org, linux-kernel@vger.kernel.org,
	kernel-team@fb.com, lizefan@huawei.com, hannes@cmpxchg.org,
	cgroups@vger.kernel.org, guro@fb.com, Tejun Heo <tj@kernel.org>
Subject: [PATCH 2/7] blkcg: use percpu_ref for blkcg_gq->refcnt
Date: Sun, 12 Nov 2017 14:26:08 -0800	[thread overview]
Message-ID: <20171112222613.3613362-3-tj@kernel.org> (raw)
In-Reply-To: <20171112222613.3613362-1-tj@kernel.org>

blkcg_gq->refcnt is an atomic_t.  This was due to the following two
reasons.

* When blkcg_gq was added, the percpu memory allocator didn't support
  allocations from !GFP_KERNEL contexts.  Because blkcg_gq's are
  created from IO issue paths, it couldn't use GFP_KERNEL allocations.

* A blkcg_gq represents the connection between a cgroup and a
  request_queue.  Because a in-flight bio already pins both, blkcg_gq
  didn't usually need explicit pinning, making the use of atomic_t
  acceptable albeit restrictive and fragile.

Now that the percpu allocator supports !GFP_KERNEL allocations,
there's no reason to keep using atomic_t refcnt.  This will allow
clean separation between bio and request layers helping blkcg support
in blk-mq.

Signed-off-by: Tejun Heo <tj@kernel.org>
---
 block/blk-cgroup.c         | 21 ++++++++++++++++-----
 include/linux/blk-cgroup.h | 13 ++++---------
 2 files changed, 20 insertions(+), 14 deletions(-)

diff --git a/block/blk-cgroup.c b/block/blk-cgroup.c
index 6482be5..60a4486 100644
--- a/block/blk-cgroup.c
+++ b/block/blk-cgroup.c
@@ -78,6 +78,7 @@ static void blkg_free(struct blkcg_gq *blkg)
 
 	blkg_rwstat_exit(&blkg->stat_ios);
 	blkg_rwstat_exit(&blkg->stat_bytes);
+	percpu_ref_exit(&blkg->refcnt);
 	kfree(blkg);
 }
 
@@ -89,7 +90,7 @@ static void blkg_free(struct blkcg_gq *blkg)
  * Having a reference to blkg under an rcu allows accesses to only values
  * local to groups like group stats and group rate limits.
  */
-void __blkg_release_rcu(struct rcu_head *rcu_head)
+static void blkg_release_rcu(struct rcu_head *rcu_head)
 {
 	struct blkcg_gq *blkg = container_of(rcu_head, struct blkcg_gq, rcu_head);
 
@@ -102,7 +103,13 @@ void __blkg_release_rcu(struct rcu_head *rcu_head)
 
 	blkg_free(blkg);
 }
-EXPORT_SYMBOL_GPL(__blkg_release_rcu);
+
+static void blkg_release(struct percpu_ref *refcnt)
+{
+	struct blkcg_gq *blkg = container_of(refcnt, struct blkcg_gq, refcnt);
+
+	call_rcu(&blkg->rcu_head, blkg_release_rcu);
+}
 
 /**
  * blkg_alloc - allocate a blkg
@@ -123,6 +130,11 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct request_queue *q,
 	if (!blkg)
 		return NULL;
 
+	if (percpu_ref_init(&blkg->refcnt, blkg_release, 0, gfp_mask)) {
+		kfree(blkg);
+		return NULL;
+	}
+
 	if (blkg_rwstat_init(&blkg->stat_bytes, gfp_mask) ||
 	    blkg_rwstat_init(&blkg->stat_ios, gfp_mask))
 		goto err_free;
@@ -130,7 +142,6 @@ static struct blkcg_gq *blkg_alloc(struct blkcg *blkcg, struct request_queue *q,
 	blkg->q = q;
 	INIT_LIST_HEAD(&blkg->q_node);
 	blkg->blkcg = blkcg;
-	atomic_set(&blkg->refcnt, 1);
 
 	/* root blkg uses @q->root_rl, init rl only for !root blkgs */
 	if (blkcg != &blkcg_root) {
@@ -266,7 +277,7 @@ static struct blkcg_gq *blkg_create(struct blkcg *blkcg,
 		return blkg;
 
 	/* @blkg failed fully initialized, use the usual release path */
-	blkg_put(blkg);
+	percpu_ref_kill(&blkg->refcnt);
 	return ERR_PTR(ret);
 
 err_put_congested:
@@ -373,7 +384,7 @@ static void blkg_destroy(struct blkcg_gq *blkg)
 	 * Put the reference taken at the time of creation so that when all
 	 * queues are gone, group can be destroyed.
 	 */
-	blkg_put(blkg);
+	percpu_ref_kill(&blkg->refcnt);
 }
 
 /**
diff --git a/include/linux/blk-cgroup.h b/include/linux/blk-cgroup.h
index 8bbc371..c0d4736 100644
--- a/include/linux/blk-cgroup.h
+++ b/include/linux/blk-cgroup.h
@@ -19,7 +19,7 @@
 #include <linux/seq_file.h>
 #include <linux/radix-tree.h>
 #include <linux/blkdev.h>
-#include <linux/atomic.h>
+#include <linux/percpu-refcount.h>
 
 /* percpu_counter batch for blkg_[rw]stats, per-cpu drift doesn't matter */
 #define BLKG_STAT_CPU_BATCH	(INT_MAX / 2)
@@ -123,7 +123,7 @@ struct blkcg_gq {
 	struct request_list		rl;
 
 	/* reference count */
-	atomic_t			refcnt;
+	struct percpu_ref		refcnt;
 
 	/* is this blkg online? protected by both blkcg and q locks */
 	bool				online;
@@ -355,21 +355,16 @@ static inline int blkg_path(struct blkcg_gq *blkg, char *buf, int buflen)
  */
 static inline void blkg_get(struct blkcg_gq *blkg)
 {
-	WARN_ON_ONCE(atomic_read(&blkg->refcnt) <= 0);
-	atomic_inc(&blkg->refcnt);
+	percpu_ref_get(&blkg->refcnt);
 }
 
-void __blkg_release_rcu(struct rcu_head *rcu);
-
 /**
  * blkg_put - put a blkg reference
  * @blkg: blkg to put
  */
 static inline void blkg_put(struct blkcg_gq *blkg)
 {
-	WARN_ON_ONCE(atomic_read(&blkg->refcnt) <= 0);
-	if (atomic_dec_and_test(&blkg->refcnt))
-		call_rcu(&blkg->rcu_head, __blkg_release_rcu);
+	percpu_ref_put(&blkg->refcnt);
 }
 
 /**
-- 
2.9.5

  parent reply	other threads:[~2017-11-12 22:26 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-11-12 22:26 [PATCHSET] blkcg: basic accounting and throttling fixes Tejun Heo
2017-11-12 22:26 ` [PATCH 1/7] blkcg: relocate __blkg_release_rcu() Tejun Heo
2017-11-14 23:12   ` Shaohua Li
2017-11-12 22:26 ` Tejun Heo [this message]
2017-11-14 23:12   ` [PATCH 2/7] blkcg: use percpu_ref for blkcg_gq->refcnt Shaohua Li
2017-11-12 22:26 ` [PATCH 3/7] blkcg: associate a request with its blkcg_gq instead of request_list Tejun Heo
2017-11-13 20:15   ` [PATCH v2 " Tejun Heo
2017-11-14 23:17     ` Shaohua Li
2017-11-15 17:11       ` Tejun Heo
2017-11-12 22:26 ` [PATCH 4/7] blkcg: refactor blkcg_gq lookup and creation in blkcg_bio_issue_check() Tejun Heo
2017-11-14 23:18   ` Shaohua Li
2017-11-12 22:26 ` [PATCH 5/7] blkcg: associate blk-mq requests with the matching blkcg_gqs Tejun Heo
2017-11-12 22:26 ` [PATCH 6/7] blkcg: account requests instead of bios for request based request_queues Tejun Heo
2017-11-14 23:23   ` Shaohua Li
2017-11-15 17:18     ` [PATCH v2 " Tejun Heo
2017-11-15 17:19     ` [PATCH " Jens Axboe
2017-11-15 17:22       ` [PATCH v3 " Tejun Heo
2017-11-12 22:26 ` [PATCH 7/7] blk-throtl: don't throttle the same IO multiple times Tejun Heo
2017-11-13  4:07   ` Shaohua Li
2017-11-13 11:13     ` Tejun Heo
2017-11-13 15:57       ` Tejun Heo
2017-11-13 19:54         ` Shaohua Li
2017-11-13 19:58           ` Tejun Heo
2017-11-13 19:58         ` Shaohua Li

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=20171112222613.3613362-3-tj@kernel.org \
    --to=tj@kernel.org \
    --cc=axboe@kernel.dk \
    --cc=cgroups@vger.kernel.org \
    --cc=guro@fb.com \
    --cc=hannes@cmpxchg.org \
    --cc=kernel-team@fb.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=lizefan@huawei.com \
    --cc=shli@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