From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
stable@vger.kernel.org, Anatol Pomozov <anatol.pomozov@gmail.com>,
Al Viro <viro@zeniv.linux.org.uk>,
Linus Torvalds <torvalds@linux-foundation.org>
Subject: [ 33/68] loop: prevent bdev freeing while device in use
Date: Tue, 2 Apr 2013 15:13:22 -0700 [thread overview]
Message-ID: <20130402221333.776278124@linuxfoundation.org> (raw)
In-Reply-To: <20130402221329.915209206@linuxfoundation.org>
3.4-stable review patch. If anyone has any objections, please let me know.
------------------
From: Anatol Pomozov <anatol.pomozov@gmail.com>
commit c1681bf8a7b1b98edee8b862a42c19c4e53205fd upstream.
struct block_device lifecycle is defined by its inode (see fs/block_dev.c) -
block_device allocated first time we access /dev/loopXX and deallocated on
bdev_destroy_inode. When we create the device "losetup /dev/loopXX afile"
we want that block_device stay alive until we destroy the loop device
with "losetup -d".
But because we do not hold /dev/loopXX inode its counter goes 0, and
inode/bdev can be destroyed at any moment. Usually it happens at memory
pressure or when user drops inode cache (like in the test below). When later in
loop_clr_fd() we want to use bdev we have use-after-free error with following
stack:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000280
bd_set_size+0x10/0xa0
loop_clr_fd+0x1f8/0x420 [loop]
lo_ioctl+0x200/0x7e0 [loop]
lo_compat_ioctl+0x47/0xe0 [loop]
compat_blkdev_ioctl+0x341/0x1290
do_filp_open+0x42/0xa0
compat_sys_ioctl+0xc1/0xf20
do_sys_open+0x16e/0x1d0
sysenter_dispatch+0x7/0x1a
To prevent use-after-free we need to grab the device in loop_set_fd()
and put it later in loop_clr_fd().
The issue is reprodusible on current Linus head and v3.3. Here is the test:
dd if=/dev/zero of=loop.file bs=1M count=1
while [ true ]; do
losetup /dev/loop0 loop.file
echo 2 > /proc/sys/vm/drop_caches
losetup -d /dev/loop0
done
[ Doing bdgrab/bput in loop_set_fd/loop_clr_fd is safe, because every
time we call loop_set_fd() we check that loop_device->lo_state is
Lo_unbound and set it to Lo_bound If somebody will try to set_fd again
it will get EBUSY. And if we try to loop_clr_fd() on unbound loop
device we'll get ENXIO.
loop_set_fd/loop_clr_fd (and any other loop ioctl) is called under
loop_device->lo_ctl_mutex. ]
Signed-off-by: Anatol Pomozov <anatol.pomozov@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
drivers/block/loop.c | 9 ++++++++-
fs/block_dev.c | 1 +
2 files changed, 9 insertions(+), 1 deletion(-)
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -908,6 +908,11 @@ static int loop_set_fd(struct loop_devic
lo->lo_flags |= LO_FLAGS_PARTSCAN;
if (lo->lo_flags & LO_FLAGS_PARTSCAN)
ioctl_by_bdev(bdev, BLKRRPART, 0);
+
+ /* Grab the block_device to prevent its destruction after we
+ * put /dev/loopXX inode. Later in loop_clr_fd() we bdput(bdev).
+ */
+ bdgrab(bdev);
return 0;
out_clr:
@@ -1004,8 +1009,10 @@ static int loop_clr_fd(struct loop_devic
memset(lo->lo_encrypt_key, 0, LO_KEY_SIZE);
memset(lo->lo_crypt_name, 0, LO_NAME_SIZE);
memset(lo->lo_file_name, 0, LO_NAME_SIZE);
- if (bdev)
+ if (bdev) {
+ bdput(bdev);
invalidate_bdev(bdev);
+ }
set_capacity(lo->lo_disk, 0);
loop_sysfs_exit(lo);
if (bdev) {
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -604,6 +604,7 @@ struct block_device *bdgrab(struct block
ihold(bdev->bd_inode);
return bdev;
}
+EXPORT_SYMBOL(bdgrab);
long nr_blockdev_pages(void)
{
next prev parent reply other threads:[~2013-04-02 22:14 UTC|newest]
Thread overview: 71+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-04-02 22:12 [ 00/68] 3.4.39-stable review Greg Kroah-Hartman
2013-04-02 22:12 ` [ 01/68] signal: Define __ARCH_HAS_SA_RESTORER so we know whether to clear sa_restorer Greg Kroah-Hartman
2013-04-02 22:12 ` [ 02/68] kernel/signal.c: use __ARCH_HAS_SA_RESTORER instead of SA_RESTORER Greg Kroah-Hartman
2013-04-02 22:12 ` [ 03/68] SUNRPC: Add barriers to ensure read ordering in rpc_wake_up_task_queue_locked Greg Kroah-Hartman
2013-04-02 22:12 ` [ 04/68] tile: expect new initramfs name from hypervisor file system Greg Kroah-Hartman
2013-04-02 22:12 ` [ 05/68] Bluetooth: Fix not closing SCO sockets in the BT_CONNECT2 state Greg Kroah-Hartman
2013-04-02 22:12 ` [ 06/68] Bluetooth: Add support for Dell[QCA 0cf3:0036] Greg Kroah-Hartman
2013-04-02 22:12 ` [ 07/68] Bluetooth: Add support for Dell[QCA 0cf3:817a] Greg Kroah-Hartman
2013-04-02 22:12 ` [ 08/68] staging: comedi: s626: fix continuous acquisition Greg Kroah-Hartman
2013-04-02 22:12 ` [ 09/68] sysfs: fix race between readdir and lseek Greg Kroah-Hartman
2013-04-02 22:12 ` [ 10/68] sysfs: handle failure path correctly for readdir() Greg Kroah-Hartman
2013-04-02 22:13 ` [ 11/68] can: sja1000: fix define conflict on SH Greg Kroah-Hartman
2013-04-02 22:13 ` [ 12/68] ath9k_hw: revert chainmask to user configuration after calibration Greg Kroah-Hartman
2013-04-02 22:13 ` [ 13/68] HID: usbhid: quirk for Realtek Multi-card reader Greg Kroah-Hartman
2013-04-02 22:13 ` [ 14/68] rtlwifi: usb: add missing freeing of skbuff Greg Kroah-Hartman
2013-04-02 22:13 ` [ 15/68] b43: N-PHY: increase initial value of "mind" in RSSI calibration Greg Kroah-Hartman
2013-04-02 22:13 ` [ 16/68] b43: A fix for DMA transmission sequence errors Greg Kroah-Hartman
2013-04-02 22:13 ` [ 17/68] b43: N-PHY: use more bits for offset in RSSI calibration Greg Kroah-Hartman
2013-04-02 22:13 ` [ 18/68] tg3: fix length overflow in VPD firmware parsing Greg Kroah-Hartman
2013-04-02 22:13 ` [ 19/68] iommu/amd: Make sure dma_ops are set for hotplug devices Greg Kroah-Hartman
2013-04-02 22:13 ` [ 20/68] xen/blkback: correctly respond to unknown, non-native requests Greg Kroah-Hartman
2013-04-02 22:13 ` [ 21/68] xen-blkback: fix dispatch_rw_block_io() error path Greg Kroah-Hartman
2013-04-02 22:13 ` [ 22/68] tty: atmel_serial_probe(): index of atmel_ports[] fix Greg Kroah-Hartman
2013-04-02 22:13 ` [ 23/68] usb: ftdi_sio: Add support for Mitsubishi FX-USB-AW/-BD Greg Kroah-Hartman
2013-04-02 22:13 ` [ 24/68] vt: synchronize_rcu() under spinlock is not nice Greg Kroah-Hartman
2013-04-02 22:13 ` [ 25/68] mwifiex: cancel cmd timer and free curr_cmd in shutdown process Greg Kroah-Hartman
2013-04-08 18:01 ` Bing Zhao
2013-04-02 22:13 ` [ 26/68] pnfs-block: removing DM device maybe cause oops when call dev_remove Greg Kroah-Hartman
2013-04-02 22:13 ` [ 27/68] net/irda: add missing error path release_sock call Greg Kroah-Hartman
2013-04-02 22:13 ` [ 28/68] usb: xhci: Fix TRB transfer length macro used for Event TRB Greg Kroah-Hartman
2013-04-02 22:13 ` [ 29/68] Btrfs: fix race between mmap writes and compression Greg Kroah-Hartman
2013-04-02 22:13 ` [ 30/68] Btrfs: limit the global reserve to 512mb Greg Kroah-Hartman
2013-04-02 22:13 ` [ 31/68] Btrfs: dont drop path when printing out tree errors in scrub Greg Kroah-Hartman
2013-04-02 22:13 ` [ 32/68] usb: gadget: udc-core: fix a regression during gadget driver unbinding Greg Kroah-Hartman
2013-04-02 22:13 ` Greg Kroah-Hartman [this message]
2013-04-02 22:13 ` [ 34/68] ARM: cns3xxx: fix mapping of private memory region Greg Kroah-Hartman
2013-04-02 22:13 ` [ 35/68] nfsd4: reject "negative" acl lengths Greg Kroah-Hartman
2013-04-02 22:13 ` [ 36/68] drm/i915: Dont clobber crtc->fb when queue_flip fails Greg Kroah-Hartman
2013-04-02 22:13 ` [ 37/68] Btrfs: fix space leak when we fail to reserve metadata space Greg Kroah-Hartman
2013-04-02 22:13 ` [ 38/68] efivars: explicitly calculate length of VariableName Greg Kroah-Hartman
2013-04-02 22:13 ` [ 39/68] efivars: Handle duplicate names from get_next_variable() Greg Kroah-Hartman
2013-04-02 22:13 ` [ 40/68] ext4: convert number of blocks to clusters properly Greg Kroah-Hartman
2013-04-02 22:13 ` [ 41/68] ext4: use atomic64_t for the per-flexbg free_clusters count Greg Kroah-Hartman
2013-04-02 22:13 ` [ 42/68] tracing: Protect tracer flags with trace_types_lock Greg Kroah-Hartman
2013-04-02 22:13 ` [ 43/68] tracing: Prevent buffer overwrite disabled for latency tracers Greg Kroah-Hartman
2013-04-02 22:13 ` [ 44/68] net: remove a WARN_ON() in net_enable_timestamp() Greg Kroah-Hartman
2013-04-02 22:13 ` [ 45/68] sky2: Receive Overflows not counted Greg Kroah-Hartman
2013-04-02 22:13 ` [ 46/68] sky2: Threshold for Pause Packet is set wrong Greg Kroah-Hartman
2013-04-02 22:13 ` [ 47/68] tcp: preserve ACK clocking in TSO Greg Kroah-Hartman
2013-04-02 22:13 ` [ 48/68] tcp: undo spurious timeout after SACK reneging Greg Kroah-Hartman
2013-04-02 22:13 ` [ 49/68] 8021q: fix a potential use-after-free Greg Kroah-Hartman
2013-04-02 22:13 ` [ 50/68] thermal: shorten too long mcast group name Greg Kroah-Hartman
2013-04-02 22:13 ` [ 51/68] unix: fix a race condition in unix_release() Greg Kroah-Hartman
2013-04-02 22:13 ` [ 52/68] af_unix: dont send SCM_CREDENTIAL when dest socket is NULL Greg Kroah-Hartman
2013-04-02 22:13 ` [ 53/68] bonding: remove already created master sysfs link on failure Greg Kroah-Hartman
2013-04-02 22:13 ` [ 54/68] bonding: fix miimon and arp_interval delayed work race conditions Greg Kroah-Hartman
2013-04-02 22:13 ` [ 55/68] bonding: fix disabling of arp_interval and miimon Greg Kroah-Hartman
2013-04-02 22:13 ` [ 56/68] drivers: net: ethernet: davinci_emac: use netif_wake_queue() while restarting tx queue Greg Kroah-Hartman
2013-04-02 22:13 ` [ 57/68] drivers: net: ethernet: cpsw: " Greg Kroah-Hartman
2013-04-02 22:13 ` [ 58/68] net: fix *_DIAG_MAX constants Greg Kroah-Hartman
2013-04-02 22:13 ` [ 59/68] aoe: reserve enough headroom on skbs Greg Kroah-Hartman
2013-04-02 22:13 ` [ 60/68] atl1e: drop pci-msi support because of packet corruption Greg Kroah-Hartman
2013-04-02 22:13 ` [ 61/68] DM9000B: driver initialization upgrade Greg Kroah-Hartman
2013-04-02 22:13 ` [ 62/68] ipv6: dont accept multicast traffic with scope 0 Greg Kroah-Hartman
2013-04-02 22:13 ` [ 63/68] ipv6: fix bad free of addrconf_init_net Greg Kroah-Hartman
2013-04-02 22:13 ` [ 64/68] ipv6: dont accept node local multicast traffic from the wire Greg Kroah-Hartman
2013-04-02 22:13 ` [ 65/68] ks8851: Fix interpretation of rxlen field Greg Kroah-Hartman
2013-04-02 22:13 ` [ 66/68] net: add a synchronize_net() in netdev_rx_handler_unregister() Greg Kroah-Hartman
2013-04-02 22:13 ` [ 67/68] pch_gbe: fix ip_summed checksum reporting on rx Greg Kroah-Hartman
2013-04-02 22:13 ` [ 68/68] smsc75xx: fix jumbo frame support Greg Kroah-Hartman
2013-04-03 15:19 ` [ 00/68] 3.4.39-stable review Shuah Khan
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=20130402221333.776278124@linuxfoundation.org \
--to=gregkh@linuxfoundation.org \
--cc=anatol.pomozov@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=stable@vger.kernel.org \
--cc=torvalds@linux-foundation.org \
--cc=viro@zeniv.linux.org.uk \
/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