public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: linux-kernel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	stable@vger.kernel.org, Alan Stern <stern@rowland.harvard.edu>,
	syzbot+76629376e06e2c2ad626@syzkaller.appspotmail.com
Subject: [PATCH 4.9 13/48] USB: core: Fix hang in usb_kill_urb by adding memory barriers
Date: Mon,  7 Feb 2022 12:05:46 +0100	[thread overview]
Message-ID: <20220207103752.778460332@linuxfoundation.org> (raw)
In-Reply-To: <20220207103752.341184175@linuxfoundation.org>

From: Alan Stern <stern@rowland.harvard.edu>

commit 26fbe9772b8c459687930511444ce443011f86bf upstream.

The syzbot fuzzer has identified a bug in which processes hang waiting
for usb_kill_urb() to return.  It turns out the issue is not unlinking
the URB; that works just fine.  Rather, the problem arises when the
wakeup notification that the URB has completed is not received.

The reason is memory-access ordering on SMP systems.  In outline form,
usb_kill_urb() and __usb_hcd_giveback_urb() operating concurrently on
different CPUs perform the following actions:

CPU 0					CPU 1
----------------------------		---------------------------------
usb_kill_urb():				__usb_hcd_giveback_urb():
  ...					  ...
  atomic_inc(&urb->reject);		  atomic_dec(&urb->use_count);
  ...					  ...
  wait_event(usb_kill_urb_queue,
	atomic_read(&urb->use_count) == 0);
					  if (atomic_read(&urb->reject))
						wake_up(&usb_kill_urb_queue);

Confining your attention to urb->reject and urb->use_count, you can
see that the overall pattern of accesses on CPU 0 is:

	write urb->reject, then read urb->use_count;

whereas the overall pattern of accesses on CPU 1 is:

	write urb->use_count, then read urb->reject.

This pattern is referred to in memory-model circles as SB (for "Store
Buffering"), and it is well known that without suitable enforcement of
the desired order of accesses -- in the form of memory barriers -- it
is entirely possible for one or both CPUs to execute their reads ahead
of their writes.  The end result will be that sometimes CPU 0 sees the
old un-decremented value of urb->use_count while CPU 1 sees the old
un-incremented value of urb->reject.  Consequently CPU 0 ends up on
the wait queue and never gets woken up, leading to the observed hang
in usb_kill_urb().

The same pattern of accesses occurs in usb_poison_urb() and the
failure pathway of usb_hcd_submit_urb().

The problem is fixed by adding suitable memory barriers.  To provide
proper memory-access ordering in the SB pattern, a full barrier is
required on both CPUs.  The atomic_inc() and atomic_dec() accesses
themselves don't provide any memory ordering, but since they are
present, we can use the optimized smp_mb__after_atomic() memory
barrier in the various routines to obtain the desired effect.

This patch adds the necessary memory barriers.

CC: <stable@vger.kernel.org>
Reported-and-tested-by: syzbot+76629376e06e2c2ad626@syzkaller.appspotmail.com
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Link: https://lore.kernel.org/r/Ye8K0QYee0Q0Nna2@rowland.harvard.edu
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 drivers/usb/core/hcd.c |   14 ++++++++++++++
 drivers/usb/core/urb.c |   12 ++++++++++++
 2 files changed, 26 insertions(+)

--- a/drivers/usb/core/hcd.c
+++ b/drivers/usb/core/hcd.c
@@ -1668,6 +1668,13 @@ int usb_hcd_submit_urb (struct urb *urb,
 		urb->hcpriv = NULL;
 		INIT_LIST_HEAD(&urb->urb_list);
 		atomic_dec(&urb->use_count);
+		/*
+		 * Order the write of urb->use_count above before the read
+		 * of urb->reject below.  Pairs with the memory barriers in
+		 * usb_kill_urb() and usb_poison_urb().
+		 */
+		smp_mb__after_atomic();
+
 		atomic_dec(&urb->dev->urbnum);
 		if (atomic_read(&urb->reject))
 			wake_up(&usb_kill_urb_queue);
@@ -1777,6 +1784,13 @@ static void __usb_hcd_giveback_urb(struc
 
 	usb_anchor_resume_wakeups(anchor);
 	atomic_dec(&urb->use_count);
+	/*
+	 * Order the write of urb->use_count above before the read
+	 * of urb->reject below.  Pairs with the memory barriers in
+	 * usb_kill_urb() and usb_poison_urb().
+	 */
+	smp_mb__after_atomic();
+
 	if (unlikely(atomic_read(&urb->reject)))
 		wake_up(&usb_kill_urb_queue);
 	usb_put_urb(urb);
--- a/drivers/usb/core/urb.c
+++ b/drivers/usb/core/urb.c
@@ -684,6 +684,12 @@ void usb_kill_urb(struct urb *urb)
 	if (!(urb && urb->dev && urb->ep))
 		return;
 	atomic_inc(&urb->reject);
+	/*
+	 * Order the write of urb->reject above before the read
+	 * of urb->use_count below.  Pairs with the barriers in
+	 * __usb_hcd_giveback_urb() and usb_hcd_submit_urb().
+	 */
+	smp_mb__after_atomic();
 
 	usb_hcd_unlink_urb(urb, -ENOENT);
 	wait_event(usb_kill_urb_queue, atomic_read(&urb->use_count) == 0);
@@ -725,6 +731,12 @@ void usb_poison_urb(struct urb *urb)
 	if (!urb)
 		return;
 	atomic_inc(&urb->reject);
+	/*
+	 * Order the write of urb->reject above before the read
+	 * of urb->use_count below.  Pairs with the barriers in
+	 * __usb_hcd_giveback_urb() and usb_hcd_submit_urb().
+	 */
+	smp_mb__after_atomic();
 
 	if (!urb->dev || !urb->ep)
 		return;



  parent reply	other threads:[~2022-02-07 11:13 UTC|newest]

Thread overview: 57+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-02-07 11:05 [PATCH 4.9 00/48] 4.9.300-rc1 review Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 01/48] can: bcm: fix UAF of bcm op Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 02/48] Bluetooth: refactor malicious adv data check Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 03/48] s390/hypfs: include z/VM guests with access control group set Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 04/48] scsi: zfcp: Fix failed recovery on gone remote port with non-NPIV FCP devices Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 05/48] udf: Restore i_lenAlloc when inode expansion fails Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 06/48] udf: Fix NULL ptr deref when converting from inline format Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 07/48] PM: wakeup: simplify the output logic of pm_show_wakelocks() Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 08/48] serial: stm32: fix software flow control transfer Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 09/48] tty: n_gsm: fix SW flow control encoding/handling Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 10/48] tty: Add support for Brainboxes UC cards Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 11/48] usb-storage: Add unusual-devs entry for VL817 USB-SATA bridge Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 12/48] usb: gadget: f_sourcesink: Fix isoc transfer for USB_SPEED_SUPER_PLUS Greg Kroah-Hartman
2022-02-07 11:05 ` Greg Kroah-Hartman [this message]
2022-02-07 11:05 ` [PATCH 4.9 14/48] powerpc/32: Fix boot failure with GCC latent entropy plugin Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 15/48] scsi: bnx2fc: Flush destroy_work queue before calling bnx2fc_interface_put() Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 16/48] ipv6_tunnel: Rate limit warning messages Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 17/48] net: fix information leakage in /proc/net/ptype Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 18/48] ipv4: avoid using shared IP generator for connected sockets Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 19/48] NFSv4: Handle case where the lookup of a directory fails Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 20/48] NFSv4: nfs_atomic_open() can race when looking up a non-regular file Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 21/48] net-procfs: show net devices bound packet types Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 22/48] drm/msm: Fix wrong size calculation Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 23/48] hwmon: (lm90) Reduce maximum conversion rate for G781 Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 24/48] ipv4: raw: lock the socket in raw_bind() Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 25/48] ipv4: tcp: send zero IPID in SYNACK messages Greg Kroah-Hartman
2022-02-07 11:05 ` [PATCH 4.9 26/48] netfilter: nat: remove l4 protocol port rovers Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 27/48] netfilter: nat: limit port clash resolution attempts Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 28/48] ipheth: fix EOVERFLOW in ipheth_rcvbulk_callback Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 29/48] net: amd-xgbe: ensure to reset the tx_timer_active flag Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 30/48] net: amd-xgbe: Fix skb data length underflow Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 31/48] rtnetlink: make sure to refresh master_dev/m_ops in __rtnl_newlink() Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 32/48] af_packet: fix data-race in packet_setsockopt / packet_setsockopt Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 33/48] ASoC: ops: Reject out of bounds values in snd_soc_put_volsw() Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 34/48] ASoC: ops: Reject out of bounds values in snd_soc_put_volsw_sx() Greg Kroah-Hartman
2022-06-03 10:06   ` Pavel Machek
2022-06-03 10:28     ` Mark Brown
2022-06-04 10:54       ` Mark Brown
2022-02-07 11:06 ` [PATCH 4.9 35/48] ASoC: ops: Reject out of bounds values in snd_soc_put_xr_sx() Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 36/48] drm/nouveau: fix off by one in BIOS boundary checking Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 37/48] iommu/amd: Fix loop timeout issue in iommu_ga_log_enable() Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 38/48] spi: bcm-qspi: check for valid cs before applying chip select Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 39/48] spi: mediatek: Avoid NULL pointer crash in interrupt Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 40/48] net: ieee802154: Return meaningful error codes from the netlink helpers Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 41/48] net: macsec: Verify that send_sci is on when setting Tx sci explicitly Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 42/48] ASoC: fsl: Add missing error handling in pcm030_fabric_probe Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 43/48] scsi: bnx2fc: Make bnx2fc_recv_frame() mp safe Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 44/48] nfsd: nfsd4_setclientid_confirm mistakenly expires confirmed client Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 45/48] rtc: cmos: Evaluate century appropriate Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 46/48] EDAC/altera: Fix deferred probing Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 47/48] EDAC/xgene: " Greg Kroah-Hartman
2022-02-07 11:06 ` [PATCH 4.9 48/48] ext4: fix error handling in ext4_restore_inline_data() Greg Kroah-Hartman
2022-02-07 21:28 ` [PATCH 4.9 00/48] 4.9.300-rc1 review Shuah Khan
2022-02-07 22:20 ` Guenter Roeck
2022-02-07 23:07 ` Florian Fainelli
2022-02-08  1:45 ` Slade Watkins
2022-02-08 10:27 ` Naresh Kamboju

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=20220207103752.778460332@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=stable@vger.kernel.org \
    --cc=stern@rowland.harvard.edu \
    --cc=syzbot+76629376e06e2c2ad626@syzkaller.appspotmail.com \
    /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