linux-stable.git/drivers/usb/core/driver.c, branch linux-3.18.y

USB: core: Fix bug caused by duplicate interface PM usage counter

2019-05-16T07:17:18+00:00

commit c2b71462d294cf517a0bc6e4fd6424d7cee5596f upstream.

The syzkaller fuzzer reported a bug in the USB hub driver which turned
out to be caused by a negative runtime-PM usage counter.  This allowed
a hub to be runtime suspended at a time when the driver did not expect
it.  The symptom is a WARNING issued because the hub's status URB is
submitted while it is already active:

	URB 0000000031fb463e submitted while active
	WARNING: CPU: 0 PID: 2917 at drivers/usb/core/urb.c:363

The negative runtime-PM usage count was caused by an unfortunate
design decision made when runtime PM was first implemented for USB.
At that time, USB class drivers were allowed to unbind from their
interfaces without balancing the usage counter (i.e., leaving it with
a positive count).  The core code would take care of setting the
counter back to 0 before allowing another driver to bind to the
interface.

Later on when runtime PM was implemented for the entire kernel, the
opposite decision was made: Drivers were required to balance their
runtime-PM get and put calls.  In order to maintain backward
compatibility, however, the USB subsystem adapted to the new
implementation by keeping an independent usage counter for each
interface and using it to automatically adjust the normal usage
counter back to 0 whenever a driver was unbound.

This approach involves duplicating information, but what is worse, it
doesn't work properly in cases where a USB class driver delays
decrementing the usage counter until after the driver's disconnect()
routine has returned and the counter has been adjusted back to 0.
Doing so would cause the usage counter to become negative.  There's
even a warning about this in the USB power management documentation!

As it happens, this is exactly what the hub driver does.  The
kick_hub_wq() routine increments the runtime-PM usage counter, and the
corresponding decrement is carried out by hub_event() in the context
of the hub_wq work-queue thread.  This work routine may sometimes run
after the driver has been unbound from its interface, and when it does
it causes the usage counter to go negative.

It is not possible for hub_disconnect() to wait for a pending
hub_event() call to finish, because hub_disconnect() is called with
the device lock held and hub_event() acquires that lock.  The only
feasible fix is to reverse the original design decision: remove the
duplicate interface-specific usage counter and require USB drivers to
balance their runtime PM gets and puts.  As far as I know, all
existing drivers currently do this.

Signed-off-by: Alan Stern 
Reported-and-tested-by: syzbot+7634edaea4d0b341c625@syzkaller.appspotmail.com
CC: 
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Greg Kroah-Hartman

USB: remove LPM management from usb_driver_claim_interface()

2018-10-13T07:09:28+00:00

commit c183813fcee44a249339b7c46e1ad271ca1870aa upstream.

usb_driver_claim_interface() disables and re-enables Link Power
Management, but it shouldn't do either one, for the reasons listed
below.  This patch removes the two LPM-related function calls from the
routine.

The reason for disabling LPM in the analogous function
usb_probe_interface() is so that drivers won't have to deal with
unwanted LPM transitions in their probe routine.  But
usb_driver_claim_interface() doesn't call the driver's probe routine
(or any other callbacks), so that reason doesn't apply here.

Furthermore, no driver other than usbfs will ever call
usb_driver_claim_interface() unless it is already bound to another
interface in the same device, which means disabling LPM here would be
redundant.  usbfs doesn't interact with LPM at all.

Lastly, the error return from usb_unlocked_disable_lpm() isn't handled
properly; the code doesn't clean up its earlier actions before
returning.

Signed-off-by: Alan Stern 
Fixes: 8306095fd2c1 ("USB: Disable USB 3.0 LPM in critical sections.")
CC: 
Signed-off-by: Greg Kroah-Hartman

USB: fix error handling in usb_driver_claim_interface()

2018-10-13T07:09:27+00:00

commit bd729f9d67aa9a303d8925bb8c4f06af25f407d1 upstream.

The syzbot fuzzing project found a use-after-free bug in the USB
core.  The bug was caused by usbfs not unbinding from an interface
when the USB device file was closed, which led another process to
attempt the unbind later on, after the private data structure had been
deallocated.

The reason usbfs did not unbind the interface at the appropriate time
was because it thought the interface had never been claimed in the
first place.  This was caused by the fact that
usb_driver_claim_interface() does not clean up properly when
device_bind_driver() returns an error.  Although the error code gets
passed back to the caller, the iface->dev.driver pointer remains set
and iface->condition remains equal to USB_INTERFACE_BOUND.

This patch adds proper error handling to usb_driver_claim_interface().

Signed-off-by: Alan Stern 
Reported-by: syzbot+f84aa7209ccec829536f@syzkaller.appspotmail.com
CC: 
Signed-off-by: Greg Kroah-Hartman

usb: hub: Do not attempt to autosuspend disconnected devices

2017-05-20T12:18:40+00:00

commit f5cccf49428447dfbc9edb7a04bb8fc316269781 upstream.

While running a bind/unbind stress test with the dwc3 usb driver on rk3399,
the following crash was observed.

Unable to handle kernel NULL pointer dereference at virtual address 00000218
pgd = ffffffc00165f000
[00000218] *pgd=000000000174f003, *pud=000000000174f003,
				*pmd=0000000001750003, *pte=00e8000001751713
Internal error: Oops: 96000005 [#1] PREEMPT SMP
Modules linked in: uinput uvcvideo videobuf2_vmalloc cmac
ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_nat_ipv4 nf_nat rfcomm
xt_mark fuse bridge stp llc zram btusb btrtl btbcm btintel bluetooth
ip6table_filter mwifiex_pcie mwifiex cfg80211 cdc_ether usbnet r8152 mii joydev
snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq snd_seq_device ppp_async
ppp_generic slhc tun
CPU: 1 PID: 29814 Comm: kworker/1:1 Not tainted 4.4.52 #507
Hardware name: Google Kevin (DT)
Workqueue: pm pm_runtime_work
task: ffffffc0ac540000 ti: ffffffc0af4d4000 task.ti: ffffffc0af4d4000
PC is at autosuspend_check+0x74/0x174
LR is at autosuspend_check+0x70/0x174
...
Call trace:
[] autosuspend_check+0x74/0x174
[] usb_runtime_idle+0x20/0x40
[] __rpm_callback+0x48/0x7c
[] rpm_idle+0x1e8/0x498
[] pm_runtime_work+0x88/0xcc
[] process_one_work+0x390/0x6b8
[] worker_thread+0x480/0x610
[] kthread+0x164/0x178
[] ret_from_fork+0x10/0x40

Source:

(gdb) l *0xffffffc00080dcc0
0xffffffc00080dcc0 is in autosuspend_check
(drivers/usb/core/driver.c:1778).
1773		/* We don't need to check interfaces that are
1774		 * disabled for runtime PM.  Either they are unbound
1775		 * or else their drivers don't support autosuspend
1776		 * and so they are permanently active.
1777		 */
1778		if (intf->dev.power.disable_depth)
1779			continue;
1780		if (atomic_read(&intf->dev.power.usage_count) > 0)
1781			return -EBUSY;
1782		w |= intf->needs_remote_wakeup;

Code analysis shows that intf is set to NULL in usb_disable_device() prior
to setting actconfig to NULL. At the same time, usb_runtime_idle() does not
lock the usb device, and neither does any of the functions in the
traceback. This means that there is no protection against a race condition
where usb_disable_device() is removing dev->actconfig->interface[] pointers
while those are being accessed from autosuspend_check().

To solve the problem, synchronize and validate device state between
autosuspend_check() and usb_disconnect().

Acked-by: Alan Stern 
Signed-off-by: Guenter Roeck 
Signed-off-by: Greg Kroah-Hartman

USB: leave LPM alone if possible when binding/unbinding interface drivers

2016-06-03T15:30:33+00:00

[ Upstream commit 6fb650d43da3e7054984dc548eaa88765a94d49f ]

When a USB driver is bound to an interface (either through probing or
by claiming it) or is unbound from an interface, the USB core always
disables Link Power Management during the transition and then
re-enables it afterward.  The reason is because the driver might want
to prevent hub-initiated link power transitions, in which case the HCD
would have to recalculate the various LPM parameters.  This
recalculation takes place when LPM is re-enabled and the new
parameters are sent to the device and its parent hub.

However, if the driver does not want to prevent hub-initiated link
power transitions then none of this work is necessary.  The parameters
don't need to be recalculated, and LPM doesn't need to be disabled and
re-enabled.

It turns out that disabling and enabling LPM can be time-consuming,
enough so that it interferes with user programs that want to claim and
release interfaces rapidly via usbfs.  Since the usbfs kernel driver
doesn't set the disable_hub_initiated_lpm flag, we can speed things up
and get the user programs to work by leaving LPM alone whenever the
flag isn't set.

And while we're improving the way disable_hub_initiated_lpm gets used,
let's also fix its kerneldoc.

Signed-off-by: Alan Stern 
Tested-by: Matthew Giassa 
CC: Mathias Nyman 
CC: 
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Sasha Levin

USB: usb_driver_claim_interface: add sanity checking

2016-04-18T12:49:35+00:00

[ Upstream commit 0b818e3956fc1ad976bee791eadcbb3b5fec5bfd ]

Attacks that trick drivers into passing a NULL pointer
to usb_driver_claim_interface() using forged descriptors are
known. This thwarts them by sanity checking.

Signed-off-by: Oliver Neukum 
CC: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Sasha Levin

USB: add flag for HCDs that can't receive wakeup requests (isp1760-hcd)

2015-03-06T22:53:00+00:00

commit 074f9dd55f9cab1b82690ed7e44bcf38b9616ce0 upstream.

Currently the USB stack assumes that all host controller drivers are
capable of receiving wakeup requests from downstream devices.
However, this isn't true for the isp1760-hcd driver, which means that
it isn't safe to do a runtime suspend of any device attached to a
root-hub port if the device requires wakeup.

This patch adds a "cant_recv_wakeups" flag to the usb_hcd structure
and sets the flag in isp1760-hcd.  The core is modified to prevent a
direct child of the root hub from being put into runtime suspend with
wakeup enabled if the flag is set.

Signed-off-by: Alan Stern 
Tested-by: Nicolas Pitre 
Signed-off-by: Greg Kroah-Hartman 
Signed-off-by: Greg Kroah-Hartman

USB: don't cancel queued resets when unbinding drivers

2015-03-06T22:53:00+00:00

commit 524134d422316a59d5464ccbc12036bbe90c5563 upstream.

The USB stack provides a mechanism for drivers to request an
asynchronous device reset (usb_queue_reset_device()).  The mechanism
uses a work item (reset_ws) embedded in the usb_interface structure
used by the driver, and the reset is carried out by a work queue
routine.

The asynchronous reset can race with driver unbinding.  When this
happens, we try to cancel the queued reset before unbinding the
driver, on the theory that the driver won't care about any resets once
it is unbound.

However, thanks to the fact that lockdep now tracks work queue
accesses, this can provoke a lockdep warning in situations where the
device reset causes another interface's driver to be unbound; see

	http://marc.info/?l=linux-usb&m=141893165203776&w=2

for an example.  The reason is that the work routine for reset_ws in
one interface calls cancel_queued_work() for the reset_ws in another
interface.  Lockdep thinks this might lead to a work routine trying to
cancel itself.  The simplest solution is not to cancel queued resets
when unbinding drivers.

This means we now need to acquire a reference to the usb_interface
when queuing a reset_ws work item and to drop the reference when the
work routine finishes.  We also need to make sure that the
usb_interface structure doesn't outlive its parent usb_device; this
means acquiring and dropping a reference when the interface is created
and destroyed.

In addition, cancelling a queued reset can fail (if the device is in
the middle of an earlier reset), and this can cause usb_reset_device()
to try to rebind an interface that has been deallocated (see
http://marc.info/?l=linux-usb&m=142175717016628&w=2 for details).
Acquiring the extra references prevents this failure.

Signed-off-by: Alan Stern 
Reported-by: Russell King - ARM Linux 
Reported-by: Olivier Sobrie 
Tested-by: Olivier Sobrie 
Signed-off-by: Greg Kroah-Hartman

USB: shutdown all URBs after controller death

2014-07-17T23:59:27+00:00

When a host controller dies, we don't need to wait for a driver to
time out.  We can shut down its URBs immediately.  Without this
change, we can end up waiting 30 seconds for a mass-storage transfer
to time out.

Signed-off-by: Alan Stern 
Signed-off-by: Greg Kroah-Hartman

USB: Avoid runtime suspend loops for HCDs that can't handle suspend/resume

2014-05-23T19:06:00+00:00

Not all host controller drivers have bus-suspend and bus-resume
methods.  When one doesn't, it will cause problems if runtime PM is
enabled in the kernel.  The PM core will attempt to suspend the
controller's root hub, the suspend will fail because there is no
bus-suspend routine, and a -EBUSY error code will be returned to the
PM core.  This will cause the suspend attempt to be repeated shortly
thereafter, in a never-ending loop.

Part of the problem is that the original error code -ENOENT gets
changed to -EBUSY in usb_runtime_suspend(), on the grounds that the PM
core will interpret -ENOENT as meaning that the root hub has gotten
into a runtime-PM error state.  While this change is appropriate for
real USB devices, it's not such a good idea for a root hub.  In fact,
considering the root hub to be in a runtime-PM error state would not
be far from the truth.  Therefore this patch updates
usb_runtime_suspend() so that it adjusts error codes only for
non-root-hub devices.

Furthermore, the patch attempts to prevent the problem from occurring
in the first place by not enabling runtime PM by default for root hubs
whose host controller driver doesn't have bus_suspend and bus_resume
methods.

Signed-off-by: Alan Stern 
Reported-by: Will Deacon 
Tested-by: Will Deacon 
CC: 
Signed-off-by: Greg Kroah-Hartman