mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@suse.de>
To: linux-kernel@vger.kernel.org
Cc: Alex Chiang <achiang@hp.com>, Greg Kroah-Hartman <gregkh@suse.de>
Subject: [PATCH 55/61] sysfs: only allow one scheduled removal callback per kobj
Date: Tue, 24 Mar 2009 17:26:59 -0700	[thread overview]
Message-ID: <1237940825-22904-55-git-send-email-gregkh@suse.de> (raw)
In-Reply-To: <20090325001659.GA22461@kroah.com>

From: Alex Chiang <achiang@hp.com>

The only way for a sysfs attribute to remove itself (without
deadlock) is to use the sysfs_schedule_callback() interface.

Vegard Nossum discovered that a poorly written sysfs ->store
callback can repeatedly schedule remove callbacks on the same
device over and over, e.g.

	$ while true ; do echo 1 > /sys/devices/.../remove ; done

If the 'remove' attribute uses the sysfs_schedule_callback API
and also does not protect itself from concurrent accesses, its
callback handler will be called multiple times, and will
eventually attempt to perform operations on a freed kobject,
leading to many problems.

Instead of requiring all callers of sysfs_schedule_callback to
implement their own synchronization, provide the protection in
the infrastructure.

Now, sysfs_schedule_callback will only allow one scheduled
callback per kobject. On subsequent calls with the same kobject,
return -EAGAIN.

This is a short term fix. The long term fix is to allow sysfs
attributes to remove themselves directly, without any of this
callback hokey pokey.

[cornelia.huck@de.ibm.com: s390 ccwgroup bits]

Reported-by: vegard.nossum@gmail.com
Signed-off-by: Alex Chiang <achiang@hp.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
---
 drivers/s390/cio/ccwgroup.c |    5 +++--
 fs/sysfs/file.c             |   26 +++++++++++++++++++++++---
 2 files changed, 26 insertions(+), 5 deletions(-)

diff --git a/drivers/s390/cio/ccwgroup.c b/drivers/s390/cio/ccwgroup.c
index 918e6fc..b91c171 100644
--- a/drivers/s390/cio/ccwgroup.c
+++ b/drivers/s390/cio/ccwgroup.c
@@ -104,8 +104,9 @@ ccwgroup_ungroup_store(struct device *dev, struct device_attribute *attr, const
 	rc = device_schedule_callback(dev, ccwgroup_ungroup_callback);
 out:
 	if (rc) {
-		/* Release onoff "lock" when ungrouping failed. */
-		atomic_set(&gdev->onoff, 0);
+		if (rc != -EAGAIN)
+			/* Release onoff "lock" when ungrouping failed. */
+			atomic_set(&gdev->onoff, 0);
 		return rc;
 	}
 	return count;
diff --git a/fs/sysfs/file.c b/fs/sysfs/file.c
index 1f4a3f8..289c43a 100644
--- a/fs/sysfs/file.c
+++ b/fs/sysfs/file.c
@@ -659,13 +659,16 @@ void sysfs_remove_file_from_group(struct kobject *kobj,
 EXPORT_SYMBOL_GPL(sysfs_remove_file_from_group);
 
 struct sysfs_schedule_callback_struct {
-	struct kobject 		*kobj;
+	struct list_head	workq_list;
+	struct kobject		*kobj;
 	void			(*func)(void *);
 	void			*data;
 	struct module		*owner;
 	struct work_struct	work;
 };
 
+static DEFINE_MUTEX(sysfs_workq_mutex);
+static LIST_HEAD(sysfs_workq);
 static void sysfs_schedule_callback_work(struct work_struct *work)
 {
 	struct sysfs_schedule_callback_struct *ss = container_of(work,
@@ -674,6 +677,9 @@ static void sysfs_schedule_callback_work(struct work_struct *work)
 	(ss->func)(ss->data);
 	kobject_put(ss->kobj);
 	module_put(ss->owner);
+	mutex_lock(&sysfs_workq_mutex);
+	list_del(&ss->workq_list);
+	mutex_unlock(&sysfs_workq_mutex);
 	kfree(ss);
 }
 
@@ -695,15 +701,25 @@ static void sysfs_schedule_callback_work(struct work_struct *work)
  * until @func returns.
  *
  * Returns 0 if the request was submitted, -ENOMEM if storage could not
- * be allocated, -ENODEV if a reference to @owner isn't available.
+ * be allocated, -ENODEV if a reference to @owner isn't available,
+ * -EAGAIN if a callback has already been scheduled for @kobj.
  */
 int sysfs_schedule_callback(struct kobject *kobj, void (*func)(void *),
 		void *data, struct module *owner)
 {
-	struct sysfs_schedule_callback_struct *ss;
+	struct sysfs_schedule_callback_struct *ss, *tmp;
 
 	if (!try_module_get(owner))
 		return -ENODEV;
+
+	mutex_lock(&sysfs_workq_mutex);
+	list_for_each_entry_safe(ss, tmp, &sysfs_workq, workq_list)
+		if (ss->kobj == kobj) {
+			mutex_unlock(&sysfs_workq_mutex);
+			return -EAGAIN;
+		}
+	mutex_unlock(&sysfs_workq_mutex);
+
 	ss = kmalloc(sizeof(*ss), GFP_KERNEL);
 	if (!ss) {
 		module_put(owner);
@@ -715,6 +731,10 @@ int sysfs_schedule_callback(struct kobject *kobj, void (*func)(void *),
 	ss->data = data;
 	ss->owner = owner;
 	INIT_WORK(&ss->work, sysfs_schedule_callback_work);
+	INIT_LIST_HEAD(&ss->workq_list);
+	mutex_lock(&sysfs_workq_mutex);
+	list_add_tail(&ss->workq_list, &sysfs_workq);
+	mutex_unlock(&sysfs_workq_mutex);
 	schedule_work(&ss->work);
 	return 0;
 }
-- 
1.6.2


  parent reply	other threads:[~2009-03-25  0:50 UTC|newest]

Thread overview: 69+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-03-25  0:16 [GIT PATCH] driver core patches for 2.6.30 merge window Greg KH
2009-03-25  0:26 ` [PATCH 01/61] edac: struct device - replace bus_id with dev_name(), dev_set_name() Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 02/61] spi: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 03/61] video: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 04/61] zorro: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 05/61] mmc: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 06/61] mtd: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 07/61] pci: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 08/61] rapidio: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 09/61] s390: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 10/61] serial: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 11/61] sh: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 12/61] tc: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 13/61] pcmcia: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 14/61] rtc: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 15/61] net: " Greg Kroah-Hartman
2009-03-25  3:31   ` Stephen Hemminger
2009-03-25  4:18     ` Greg KH
2009-03-25  6:29   ` David Miller
2009-03-25  9:34     ` Kay Sievers
2009-03-25  0:26 ` [PATCH 16/61] drm: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 17/61] v4l: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 18/61] amba: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 19/61] dio: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 20/61] dma: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 21/61] eisa: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 22/61] gpio: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 23/61] mca: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 24/61] mfd: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 25/61] wimax: " Greg Kroah-Hartman
2009-03-25  1:22   ` Perez-Gonzalez, Inaky
2009-03-25  0:26 ` [PATCH 26/61] usb: " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 27/61] PS3: replace bus_id usage Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 28/61] USB: FHCI: use dev_name() in place of bus_id Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 29/61] driver core: get rid of struct device's bus_id string array Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 30/61] SYSFS: use standard magic.h for sysfs Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 31/61] sysfs: Take sysfs_mutex when fetching the root inode Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 32/61] driver core: check bus->match without holding device lock Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 33/61] platform: make better use of to_platform_{device,driver}() macros Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 34/61] platform: introduce module id table for platform devices Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 35/61] UIO: Add name attributes for mappings and port regions Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 36/61] uio: add the uio_aec driver Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 37/61] UIO: Take offset into account when determining number of pages that can be mapped Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 38/61] platform driver: fix incorrect use of 'platform_bus_type' with 'struct device_driver' Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 39/61] scsi/m68k: Kill NCR_700_detect() warnings Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 40/61] driver-core: do not register a driver with bus_type not registered Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 41/61] sysfs: sysfs_add_one WARNs with full path to duplicate filename Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 42/61] sysfs: reference sysfs_dirent from sysfs inodes Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 43/61] driver core: remove polling for driver_probe_done(v5) Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 44/61] driver core: create a private portion of struct device Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 45/61] driver core: move klist_children into private structure Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 46/61] driver core: move knode_driver " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 47/61] driver core: move knode_bus " Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 48/61] sysfs: don't block indefinitely for unmapped files Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 49/61] driver core: move platform_data into platform_device Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 50/61] driver core: fix passing platform_data Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 51/61] vcs: hook sysfs devices into object lifetime instead of "binding" Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 52/61] Driver core: implement uevent suppress in kobject Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 53/61] Driver core: some cleanup on drivers/base/sys.c Greg Kroah-Hartman
2009-03-25  0:26 ` [PATCH 54/61] Driver core: Fix device_move() vs. dpm list ordering, v2 Greg Kroah-Hartman
2009-03-25  0:26 ` Greg Kroah-Hartman [this message]
2009-03-25  0:27 ` [PATCH 56/61] kobject: don't block for each kobject_uevent Greg Kroah-Hartman
2009-03-25  0:27 ` [PATCH 57/61] sysfs: fix some bin_vm_ops errors Greg Kroah-Hartman
2009-03-25  0:27 ` [PATCH 58/61] dynamic debug: combine dprintk and dynamic printk Greg Kroah-Hartman
2009-03-25 10:50   ` Ingo Molnar
2009-03-25 13:20     ` Jason Baron
2009-03-25  0:27 ` [PATCH 59/61] dynamic debug: update docs Greg Kroah-Hartman
2009-03-25  0:27 ` [PATCH 60/61] Dynamic debug: allow simple quoting of words Greg Kroah-Hartman
2009-03-25  0:27 ` [PATCH 61/61] Dynamic debug: fix pr_fmt() build error Greg Kroah-Hartman

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=1237940825-22904-55-git-send-email-gregkh@suse.de \
    --to=gregkh@suse.de \
    --cc=achiang@hp.com \
    --cc=linux-kernel@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

all inboxes | Powered by JetHome®