From:   Imran Khan <imran.f.khan@oracle.com>
To:     gregkh@linuxfoundation.org, tj@kernel.org
Cc:     linux-kernel@vger.kernel.org
Subject: [RFC PATCH] kernfs: use kernfs_node specific mutex and spinlock.
Date:   Thu, 16 Dec 2021 02:06:38 +1100
Message-Id: <20211215150638.390466-1-imran.f.khan@oracle.com>
Content-Transfer-Encoding: 8bit
Content-Type: text/plain
MIME-Version: 1.0
X-MS-Exchange-AntiSpam-MessageData-ChunkCount: 1
X-MS-Exchange-AntiSpam-MessageData-0: =?us-ascii?Q?ewMAUztFOYjp+KlCIutJpKc4DLacw/pzKgsMRuiaZVxQxN7+AlghMbuBmnr2?=
 =?us-ascii?Q?7WBetiOGZrjjtJBAFimQreG70HvTpAh63FMQiZ8Lh65qR3RuxkIKHSxYE0gz?=
 =?us-ascii?Q?Sw8R3G5G3BY1aJO2DPrLJKLOYOt9omaMDc3EKf/5Bk/sAIUj+3KZ8cffwv44?=
 =?us-ascii?Q?AbFkJ9mOVDud+ip7eyC12lCM7CWfZgjXwlaLEoaeM0O8jRp4RKFLTM5LSvEW?=
 =?us-ascii?Q?xGYNrXbWe3u1nMtqR9CjsY0YxJhJ1tqxjfuelcUYyTGiEriaj6f+xwlULgE3?=
 =?us-ascii?Q?ddKRnqC3x88v/pKG49kSJZ5v4HI9WGHngzpqnxgQotS9U45j5ywcZ2QMEtuX?=
 =?us-ascii?Q?oDhL13r+OMEkpt3co349PY9Tc8/u9CPBcrpVhIRgd7bzQ8SkUgDNVgQHnlwl?=
 =?us-ascii?Q?JjqcZEs5u45QLWaELmcaC9L7j4SLrlflLQ82LGvHluONuPPW5wNmlDmoenb7?=
 =?us-ascii?Q?8zgrcz40pLCOcmT1iGKUQohXNHwOsljuc6fThVUY3xAlqVPYmJF5HDwOm8dB?=
 =?us-ascii?Q?Nyz48cpbPyGpfc2RYa1VTrq06YZuGn+uubrM3VAV3kg2UHHVj2zixdCWBGxD?=
 =?us-ascii?Q?iT3rH8u6bWeebm8jfL7Te7m5iDGbz6oF4F1nmqdw9u2q8Yvy1soDyGuSYQZq?=
 =?us-ascii?Q?YhgQc16zL1ENwv3PC5/qG2CdJvz8AtkVvwCPQBXNsYAPUQH7stUnWLf0b70E?=
 =?us-ascii?Q?Dgpq77sq6wCfTRRE5Y7adHoU1WWMlO9HMbDcIhx7dVw41++WT4xkRifv2sbO?=
 =?us-ascii?Q?bHeefq5gC4TYnC/Fw3Roedap1R5zPmu+MnB3JVHDTgixO8luE5u/ubmOOaIw?=
 =?us-ascii?Q?lFYMn3BJ5Q61FYLAX4ZOUHRJPVKhmOw8p29zx+tbYn8GAg5nItC5RtWjXbez?=
 =?us-ascii?Q?OcypTewPuKVK3hG4HN0t0YDmRbA5YWy75HWy6uf7t1gOa+gML4N6i+dktENP?=
 =?us-ascii?Q?VK/OGWCqRBkAvTmY0sNeHUF8D27b2tMaX20/RNvgU8aqF3EE+JSXZ4YvRBBl?=
 =?us-ascii?Q?Sxy6zlvnWGEC6r5uJUdgUlsFphMMmzDYmlcp1VJ2S3pydvaZX4CwoifqFrPO?=
 =?us-ascii?Q?sxQufHvt0esA94WfNIAgMp8IHChRltrcou4qowF2EsI7cm7gpXkwLmOR2cgB?=
 =?us-ascii?Q?Nt56IlCA2gvCMfRr3r1lq/MyU+5DSHH33FgRHJpLaxl97zTzFTWBXUBdocCl?=
 =?us-ascii?Q?u5kNruJk/i5TcM0fhoPJ70ulGlgiJ3Fp0S2kCSpWD8S9NKCIwSSaIsN1XEhA?=
 =?us-ascii?Q?093m5NLAiPhNS0As3MvAGJomKNlXRNg9R39Lhuq7igvJeSSwDUFUFTh7CXhL?=
 =?us-ascii?Q?TWzpdoczocRSIVBiCH9cPgFUz6BfS1a02W0bbEUlxHxJhQezn501nk6NzoSO?=
 =?us-ascii?Q?58MreIHhH1RpamaqFmDT6raNpUZyDcEIP7zN1UdgRR/99ggPa8osWq5HZxtS?=
 =?us-ascii?Q?UMD/oXFfo7YeWXrgEsFGC1k5FuvVz3CaGfFr58IH6dnPXGQmNi7U1oTy5g7j?=
 =?us-ascii?Q?+vXAaoh+3t6DS6rahrzzmUSsdr/Rh30ksrTvIMWb+yzB4Rmw6y8lC0D2mDck?=
 =?us-ascii?Q?g/KNwOehJ+zASGEn4RGN6rufX1mwJWFkAO9weCSydt2zE2uETuMmx02iM3jS?=
 =?us-ascii?Q?EcQsk9X0nnUWzFLfEFrindE=3D?=
X-OriginatorOrg: oracle.com
X-MS-Exchange-CrossTenant-Network-Message-Id: 9e8b85e6-907d-4069-2be5-08d9bfdc8816
X-MS-Exchange-CrossTenant-AuthSource: CO1PR10MB4468.namprd10.prod.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Internal
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 15 Dec 2021 15:06:54.9625
 (UTC)
X-MS-Exchange-CrossTenant-FromEntityHeader: Hosted
X-MS-Exchange-CrossTenant-Id: 4e2c6054-71cb-48f1-bd6c-3a9705aca71b
X-MS-Exchange-CrossTenant-MailboxType: HOSTED
X-MS-Exchange-CrossTenant-UserPrincipalName: nQAQqFcsdSemj8jmQX1995tiPB0awQwdohyPjBaWDZXRMkPLDqxGUye1swGyyh+ebN0PM1OLwlH4qWZYVlEJeA==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: MW4PR10MB5861
X-Proofpoint-Virus-Version: vendor=nai engine=6300 definitions=10198 signatures=668683
X-Proofpoint-Spam-Details: rule=notspam policy=default score=0 phishscore=0 spamscore=0 suspectscore=0
 malwarescore=0 mlxlogscore=999 bulkscore=0 mlxscore=0 adultscore=0
 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2110150000
 definitions=main-2112150086
X-Proofpoint-ORIG-GUID: srMnC8DB3fT1rPpH3BFcrOYkoY12h0D2
X-Proofpoint-GUID: srMnC8DB3fT1rPpH3BFcrOYkoY12h0D2
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

Right now a global mutex (kernfs_open_file_mutex) protects list of
kernfs_open_file instances corresponding to a sysfs attribute, so even
if different tasks are opening or closing different sysfs files they
can contend on osq_lock of this mutex. The contention is more apparent
in large scale systems with few hundred CPUs where most of the CPUs have
running tasks that are opening, accessing or closing sysfs files at any
point of time. Since each list of kernfs_open_file belongs to a
kernfs_open_node instance which in turn corresponds to one kernfs_node,
move global kernfs_open_file_mutex within kernfs_node so that it does
not block access to kernfs_open_file lists corresponding to other
kernfs_node.

Also since kernfs_node->attr.open points to kernfs_open_node instance
corresponding to the kernfs_node, we can use a kernfs_node specific
spinlock in place of current global spinlock i.e kernfs_open_node_lock.
So make this spinlock local to kernfs_node instance as well.

Signed-off-by: Imran Khan <imran.f.khan@oracle.com>
---
I have kept this patch as RFC, as I am not sure if I have overlooked any
scenario(s) where these global locks are needed.

 fs/kernfs/dir.c        |  2 ++
 fs/kernfs/file.c       | 48 +++++++++++++++---------------------------
 include/linux/kernfs.h |  2 ++
 3 files changed, 21 insertions(+), 31 deletions(-)

diff --git a/fs/kernfs/dir.c b/fs/kernfs/dir.c
index e6d9772ddb4c..cd68ac30f71b 100644
--- a/fs/kernfs/dir.c
+++ b/fs/kernfs/dir.c
@@ -603,6 +603,8 @@ static struct kernfs_node *__kernfs_new_node(struct kernfs_root *root,
 	atomic_set(&kn->count, 1);
 	atomic_set(&kn->active, KN_DEACTIVATED_BIAS);
 	RB_CLEAR_NODE(&kn->rb);
+	spin_lock_init(&kn->kernfs_open_node_lock);
+	mutex_init(&kn->kernfs_open_file_mutex);
 
 	kn->name = name;
 	kn->mode = mode;
diff --git a/fs/kernfs/file.c b/fs/kernfs/file.c
index 9414a7a60a9f..4114745d80d5 100644
--- a/fs/kernfs/file.c
+++ b/fs/kernfs/file.c
@@ -18,20 +18,6 @@
 
 #include "kernfs-internal.h"
 
-/*
- * There's one kernfs_open_file for each open file and one kernfs_open_node
- * for each kernfs_node with one or more open files.
- *
- * kernfs_node->attr.open points to kernfs_open_node.  attr.open is
- * protected by kernfs_open_node_lock.
- *
- * filp->private_data points to seq_file whose ->private points to
- * kernfs_open_file.  kernfs_open_files are chained at
- * kernfs_open_node->files, which is protected by kernfs_open_file_mutex.
- */
-static DEFINE_SPINLOCK(kernfs_open_node_lock);
-static DEFINE_MUTEX(kernfs_open_file_mutex);
-
 struct kernfs_open_node {
 	atomic_t		refcnt;
 	atomic_t		event;
@@ -526,8 +512,8 @@ static int kernfs_get_open_node(struct kernfs_node *kn,
 	struct kernfs_open_node *on, *new_on = NULL;
 
  retry:
-	mutex_lock(&kernfs_open_file_mutex);
-	spin_lock_irq(&kernfs_open_node_lock);
+	mutex_lock(&kn->kernfs_open_file_mutex);
+	spin_lock_irq(&kn->kernfs_open_node_lock);
 
 	if (!kn->attr.open && new_on) {
 		kn->attr.open = new_on;
@@ -540,8 +526,8 @@ static int kernfs_get_open_node(struct kernfs_node *kn,
 		list_add_tail(&of->list, &on->files);
 	}
 
-	spin_unlock_irq(&kernfs_open_node_lock);
-	mutex_unlock(&kernfs_open_file_mutex);
+	spin_unlock_irq(&kn->kernfs_open_node_lock);
+	mutex_unlock(&kn->kernfs_open_file_mutex);
 
 	if (on) {
 		kfree(new_on);
@@ -577,8 +563,8 @@ static void kernfs_put_open_node(struct kernfs_node *kn,
 	struct kernfs_open_node *on = kn->attr.open;
 	unsigned long flags;
 
-	mutex_lock(&kernfs_open_file_mutex);
-	spin_lock_irqsave(&kernfs_open_node_lock, flags);
+	mutex_lock(&kn->kernfs_open_file_mutex);
+	spin_lock_irqsave(&kn->kernfs_open_node_lock, flags);
 
 	if (of)
 		list_del(&of->list);
@@ -588,8 +574,8 @@ static void kernfs_put_open_node(struct kernfs_node *kn,
 	else
 		on = NULL;
 
-	spin_unlock_irqrestore(&kernfs_open_node_lock, flags);
-	mutex_unlock(&kernfs_open_file_mutex);
+	spin_unlock_irqrestore(&kn->kernfs_open_node_lock, flags);
+	mutex_unlock(&kn->kernfs_open_file_mutex);
 
 	kfree(on);
 }
@@ -733,7 +719,7 @@ static void kernfs_release_file(struct kernfs_node *kn,
 	 * here because drain path may be called from places which can
 	 * cause circular dependency.
 	 */
-	lockdep_assert_held(&kernfs_open_file_mutex);
+	lockdep_assert_held(&kn->kernfs_open_file_mutex);
 
 	if (!of->released) {
 		/*
@@ -752,9 +738,9 @@ static int kernfs_fop_release(struct inode *inode, struct file *filp)
 	struct kernfs_open_file *of = kernfs_of(filp);
 
 	if (kn->flags & KERNFS_HAS_RELEASE) {
-		mutex_lock(&kernfs_open_file_mutex);
+		mutex_lock(&kn->kernfs_open_file_mutex);
 		kernfs_release_file(kn, of);
-		mutex_unlock(&kernfs_open_file_mutex);
+		mutex_unlock(&kn->kernfs_open_file_mutex);
 	}
 
 	kernfs_put_open_node(kn, of);
@@ -773,15 +759,15 @@ void kernfs_drain_open_files(struct kernfs_node *kn)
 	if (!(kn->flags & (KERNFS_HAS_MMAP | KERNFS_HAS_RELEASE)))
 		return;
 
-	spin_lock_irq(&kernfs_open_node_lock);
+	spin_lock_irq(&kn->kernfs_open_node_lock);
 	on = kn->attr.open;
 	if (on)
 		atomic_inc(&on->refcnt);
-	spin_unlock_irq(&kernfs_open_node_lock);
+	spin_unlock_irq(&kn->kernfs_open_node_lock);
 	if (!on)
 		return;
 
-	mutex_lock(&kernfs_open_file_mutex);
+	mutex_lock(&kn->kernfs_open_file_mutex);
 
 	list_for_each_entry(of, &on->files, list) {
 		struct inode *inode = file_inode(of->file);
@@ -793,7 +779,7 @@ void kernfs_drain_open_files(struct kernfs_node *kn)
 			kernfs_release_file(kn, of);
 	}
 
-	mutex_unlock(&kernfs_open_file_mutex);
+	mutex_unlock(&kn->kernfs_open_file_mutex);
 
 	kernfs_put_open_node(kn, NULL);
 }
@@ -922,13 +908,13 @@ void kernfs_notify(struct kernfs_node *kn)
 		return;
 
 	/* kick poll immediately */
-	spin_lock_irqsave(&kernfs_open_node_lock, flags);
+	spin_lock_irqsave(&kn->kernfs_open_node_lock, flags);
 	on = kn->attr.open;
 	if (on) {
 		atomic_inc(&on->event);
 		wake_up_interruptible(&on->poll);
 	}
-	spin_unlock_irqrestore(&kernfs_open_node_lock, flags);
+	spin_unlock_irqrestore(&kn->kernfs_open_node_lock, flags);
 
 	/* schedule work to kick fsnotify */
 	spin_lock_irqsave(&kernfs_notify_lock, flags);
diff --git a/include/linux/kernfs.h b/include/linux/kernfs.h
index 9f650986a81b..22cd01477129 100644
--- a/include/linux/kernfs.h
+++ b/include/linux/kernfs.h
@@ -162,6 +162,8 @@ struct kernfs_node {
 	unsigned short		flags;
 	umode_t			mode;
 	struct kernfs_iattrs	*iattr;
+	spinlock_t kernfs_open_node_lock;
+	struct mutex kernfs_open_file_mutex;
 };
 
 /*

base-commit: 0bafb8f3ebc84525d0ae0fcea22d12151b99312f
-- 
2.30.2