lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20161118205412.GA1935@tuco>
Date:   Fri, 18 Nov 2016 21:54:12 +0100
From:   Ivan Grimaldi <grimaldi.ivan@...il.com>
To:     Richard Weinberger <richard.weinberger@...il.com>,
        grimaldi.ivan@...il.com
Cc:     "linux-mtd@...ts.infradead.org" <linux-mtd@...ts.infradead.org>,
        LKML <linux-kernel@...r.kernel.org>, grimaldi.ivan@...il.com
Subject: Re: [BUG] mtd: blkdevs: fix deadlock

On 11/17, Richard Weinberger wrote:
> On Thu, Nov 17, 2016 at 10:15 PM, Ivan Grimaldi <grimaldi.ivan@...il.com> wrote:
> > I found this deadlock with kernel version 4.4.30, when insmod the ftl
> > module, which register_mtd_blktrans.
> >
> >   -> register_mtd_blktrans()
> >   -> mutex_lock(&mtd_table_mutex)
> >   -> ftl_add_mtd()
> >   -> add_mtd_blktrans_dev()
> >   -> add_disk()
> >   -> register_disk()
> >   -> blkdev_get()
> >   -> __blkdev_get()
> >   -> blktrans_open()
> >   -> mutex_lock(&mtd_table_mutex); !!!!!!!
> >
> > I think the lock/unlock of the mtd_table_mutex is not necessary in
> > blktrans_open and blktrans_release, the log message of lockdep
> > output follow:
> >
> > =============================================
> > [ INFO: possible recursive locking detected ]
> > 4.4.30+siae-g8b14c5b-dirty #23 Not tainted
> > ---------------------------------------------
> > insmod/440 is trying to acquire lock:
> >  (mtd_table_mutex){+.+.+.}, at: [<802fe6b0>] blktrans_open+0x30/0x1ac
> >
> > but task is already holding lock:
> >  (mtd_table_mutex){+.+.+.}, at: [<802fe93c>] register_mtd_blktrans+0x2c/0x120
> >
> > other info that might help us debug this:
> >  Possible unsafe locking scenario:
> >
> >        CPU0
> >        ----
> >   lock(mtd_table_mutex);
> >   lock(mtd_table_mutex);
> >
> >  *** DEADLOCK ***
> >
> >  May be due to missing lock nesting notation
> >
> > 2 locks held by insmod/440:
> >  #0:  (mtd_table_mutex){+.+.+.}, at: [<802fe93c>] register_mtd_blktrans+0x2c/0x120
> >  #1:  (&bdev->bd_mutex){+.+.+.}, at: [<8013c404>] __blkdev_get+0x40/0x398
> >
> > stack backtrace:
> > CPU: 0 PID: 440 Comm: insmod Not tainted 4.4.30+siae-g8b14c5b-dirty #23
> > Hardware name: SIAE Microelettronica AGS20 (Device Tree)
> > Backtrace:
> > [<800140e8>] (dump_backtrace) from [<800142dc>] (show_stack+0x18/0x1c)
> >  r6:60000093 r5:00000000 r4:8067cb48 r3:00000000
> > [<800142c4>] (show_stack) from [<8024fe1c>] (dump_stack+0xb4/0xe8)
> > [<8024fd68>] (dump_stack) from [<8006dc48>] (__lock_acquire+0x1a94/0x1f14)
> >  r10:807d6db8 r9:8067cc28 r8:bdacd4c0 r7:80e933dc r6:bdacd080 r5:80e84ab4
> >  r4:807d6db8 r3:00000002
> > [<8006c1b4>] (__lock_acquire) from [<8006e8ec>] (lock_acquire+0x78/0x98)
> >  r10:80686c38 r9:00000000 r8:00000001 r7:00000001 r6:802fe6b0 r5:60000013
> >  r4:00000000
> > [<8006e874>] (lock_acquire) from [<804bc708>] (mutex_lock_nested+0x54/0x3ec)
> >  r7:bdacd080 r6:80e84ab4 r5:802fe6b0 r4:00000000
> > [<804bc6b4>] (mutex_lock_nested) from [<802fe6b0>] (blktrans_open+0x30/0x1ac)
> >  r10:00000001 r9:00000000 r8:00000001 r7:be61be10 r6:00000000 r5:bd98a400
> >  r4:be61be00
> > [<802fe680>] (blktrans_open) from [<8013c464>] (__blkdev_get+0xa0/0x398)
> >  r8:00000000 r7:bec24f10 r6:00000000 r5:bd98a400 r4:bec24f00 r3:802fe680
> > [<8013c3c4>] (__blkdev_get) from [<8013c86c>] (blkdev_get+0x110/0x348)
> >  r10:bec24f00 r9:00000001 r8:be831f90 r7:bd98a40c r6:00000000 r5:00000000
> >  r4:bec24f00
> > [<8013c75c>] (blkdev_get) from [<8024204c>] (add_disk+0x318/0x46c)
> >  r10:bec24f00 r9:00000001 r8:be831f90 r7:bd98a40c r6:bd98a480 r5:00000000
> >  r4:bd98a400
> > [<80241d34>] (add_disk) from [<802fde80>] (add_mtd_blktrans_dev+0x314/0x440)
> >  r10:be61be9c r9:00000000 r8:00f68e00 r7:bd98a400 r6:7f00242c r5:00000000
> >  r4:be61be00
> > [<802fdb6c>] (add_mtd_blktrans_dev) from [<7f000c74>] (ftl_add_mtd+0x490/0x7d8 [ftl])
> >  r10:bd843d28 r9:be61bf00 r8:00000000 r7:000000bf r6:00007b47 r5:00000be0
> >  r4:be61be00
> > [<7f0007e4>] (ftl_add_mtd [ftl]) from [<802fe9c4>] (register_mtd_blktrans+0xb4/0x120)
> >  r10:00000000 r9:be618ec0 r8:7f0024c8 r7:7f004000 r6:00000000 r5:7f00242c
> >  r4:be806000
> > [<802fe910>] (register_mtd_blktrans) from [<7f004014>] (init_ftl+0x14/0x1c [ftl])
> >  r6:be618f80 r5:8066ad20 r4:8066ad20 r3:00000000
> > [<7f004000>] (init_ftl [ftl]) from [<800097bc>] (do_one_initcall+0x88/0x1e8)
> > [<80009734>] (do_one_initcall) from [<800bf9f4>] (do_init_module+0x60/0x394)
> >  r10:00000001 r9:be618ec0 r8:7f0024c8 r7:7f002480 r6:be618700 r5:be618ec8
> >  r4:7f002480
> > [<800bf994>] (do_init_module) from [<800a1b7c>] (load_module+0x1814/0x1f34)
> >  r6:00000001 r5:be618ec8 r4:bd843f34
> > [<800a0368>] (load_module) from [<800a2370>] (SyS_init_module+0xd4/0x148)
> >  r10:bd842000 r9:00000051 r8:c6b16780 r7:76fd8788 r6:00000000 r5:00004780
> >  r4:00000000
> > [<800a229c>] (SyS_init_module) from [<8000fb60>] (ret_fast_syscall+0x0/0x1c)
> >  r10:00000000 r9:bd842000 r8:8000fd04 r7:00000080 r6:00000191 r5:7e8cbf31
> >  r4:00024780
> 
> Forgot to include the diff?
> 
> -- 
> Thanks,
> //richard

Hi,

	For me the solution is remove the mutex lock for mtd_table_mutex in
	blktrans_open and blktrans_release, the patch follow.

IVan

Signed-off-by: Ivan Grimaldi <grimaldi.ivan@...il.com>
---
 drivers/mtd/mtd_blkdevs.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/drivers/mtd/mtd_blkdevs.c b/drivers/mtd/mtd_blkdevs.c
index f470118..472d48f 100644
--- a/drivers/mtd/mtd_blkdevs.c
+++ b/drivers/mtd/mtd_blkdevs.c
@@ -192,7 +192,6 @@ static int blktrans_open(struct block_device *bdev, fmode_t mode)
 	if (!dev)
 		return -ERESTARTSYS; /* FIXME: busy loop! -arnd*/
 
-	mutex_lock(&mtd_table_mutex);
 	mutex_lock(&dev->lock);
 
 	if (dev->open)
@@ -218,7 +217,6 @@ static int blktrans_open(struct block_device *bdev, fmode_t mode)
 unlock:
 	dev->open++;
 	mutex_unlock(&dev->lock);
-	mutex_unlock(&mtd_table_mutex);
 	blktrans_dev_put(dev);
 	return ret;
 
@@ -229,7 +227,6 @@ error_put:
 	module_put(dev->tr->owner);
 	kref_put(&dev->ref, blktrans_dev_release);
 	mutex_unlock(&dev->lock);
-	mutex_unlock(&mtd_table_mutex);
 	blktrans_dev_put(dev);
 	return ret;
 }
@@ -241,7 +238,6 @@ static void blktrans_release(struct gendisk *disk, fmode_t mode)
 	if (!dev)
 		return;
 
-	mutex_lock(&mtd_table_mutex);
 	mutex_lock(&dev->lock);
 
 	if (--dev->open)
@@ -257,7 +253,6 @@ static void blktrans_release(struct gendisk *disk, fmode_t mode)
 	}
 unlock:
 	mutex_unlock(&dev->lock);
-	mutex_unlock(&mtd_table_mutex);
 	blktrans_dev_put(dev);
 }
 
-- 
2.10.2

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ