KASAN reported use-after-free bug when xhci host controller died:
[ 176.952537] BUG: KASAN: use-after-free in xhci_handle_command_timeout+0x68/0x224
[ 176.960846] Write of size 4 at addr ffffffc0cbb01608 by task kworker/3:3/1680
...
[ 177.180644] Freed by task 0:
[ 177.183882] kasan_slab_free+0x90/0x15c
[ 177.188194] kfree+0x114/0x28c
[ 177.191630] xhci_cleanup_command_queue+0xc8/0xf8
[ 177.196916] xhci_hc_died+0x84/0x358
Problem here is that when the cmd_timer fired, it would try to access
current_cmd while the command queue is already freed by xhci_hc_died().
Cleanup current_cmd in xhci_cleanup_command_queue() to avoid that.
Fixes: d9f11ba9f1 ("xhci: Rework how we handle unresponsive or hoptlug removed hosts")
Cc: <stable@vger.kernel.org> # v4.12+
Signed-off-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Track the port status in a human readble way each time we get a
port status change event
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
temp and temp1 variables are used for port status (portsc) and
command register. Give them more descriptive names
No functional changes
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This off by one in stream_id indexing caused NULL pointer dereference and
soft lockup on machines with USB attached SCSI devices connected to a
hotpluggable xhci controller.
The code that cleans up pending URBs for dead hosts tried to dereference
a stream ring at the invalid stream_id 0.
ep->stream_info->stream_rings[0] doesn't point to a ring.
Start looping stream_id from 1 like in all the other places in the driver,
and check that the ring exists before trying to kill URBs on it.
Reported-by: rocko r <rockorequin@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
finish_td() could be called with a skip option to bypass most of the
function and only call xhci_td_cleanup() at the end.
Remove this skip option and call xhci_td_cleanup() directly instead
when needed
No functional changes
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Get rid of stopped_stream member in virtual endpoint structure as
it is only used in one case when cleaning a halted endpoint.
Pass it as function parameter instead.
No functional changes
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Most transfer events have a TRB pointer indicating which TRB caused
the event.
In the case of streams, transfer events such as
USB Transaction error may have its TRB pointer set to zero.
driver won't know which stream or what TRB on that stream caused
the error, but it can issue a soft reset to recover the transfer.
A soft reset will clear the host side halt of the endpoint without
clearing Data toggle or sequence number, and let the transfer
continue from where it halted.
see xhci section 4.12 streams and 4.6.8.2 soft retry.
USB Transaction errors with a zero TRB pointer are seen with
UAS usb devices.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add soft reset support to cleanup_halted_endpoint().
using soft reset will prevent it from setting a new dequeue pointer to
start the transfer from. Let it continue where it halted.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
xhci supports soft retry recovery when the host halted the host side of an
endopint but the connected USB device is not aware of the halt.
In this case xhci needs to issue a reset endopint command with a TSP
(Transfer State Preserve) flag set which preserves the Data toggle
and Sequence number of the endpoint.
This feature is needed to handle a few special transfer event types
such as USB Transaction error that don't always point to a causing TRB.
see xhci 4.6.8.1 for more details
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Parse the transfer event first, and remove duplicate debugging
code.
Reorder completion codes according to endpoint state.
No functional changes
We are not handling some transfer events correcly and need to
clean up this before fixing it
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add xhci_get_hw_deq() helper to retrieve the hardware dequeue pointer an
endpoint or stream stopped on.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The values for the new dequeue segment, new dequeue pointer and new cycle
state are needed for manually moving the xHC ring dequeue pointer.
These are conveniently stored in a xhci_dequeue_state structure.
stream support was added later and stream_id was carried
as a function parameter.
Move the stream_id to the xhci_dequeue_state structure instead.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When xHC is asked to stop an endpoint it will save the position it
stopped on in the endpoint or stream context.
xhci driver needs to know if the controller stopped on the exact same
TRB that the driver was asked to cancel as it then needs to move past
the TD instead of turning the TD to no-op TRBs.
xhci driver used to get the stopped position from a "stopped" transfer
event before the stop endpoint command completed, but if the ring
is already stopped, or in a halted or error state this event is missing.
Get the stopped position from the endpoint or stream context instead
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We no longer keep track of where we stopped in a stopped_td pointer.
We get the ring dequeue pointer from the endpoint or stream context
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In 4.11 TRB completion codes were renamed to match spec.
Completion codes for command ring stopped and endpoint stopped
were mixed, leading to failures while handling a stopped command ring.
Use the correct completion code for command ring stopped events.
Fixes: 0b7c105a04 ("usb: host: xhci: rename completion codes to match spec")
Cc: <stable@vger.kernel.org> # 4.11
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With threaded interrupts, bottom-half handlers are called with
interrupts enabled. Therefore they can't safely use spin_lock(); they
have to use spin_lock_irqsave(). Lockdep warns about a violation
occurring in xhci_irq():
=========================================================
[ INFO: possible irq lock inversion dependency detected ]
4.11.0-rc8-dbg+ #1 Not tainted
---------------------------------------------------------
swapper/7/0 just changed the state of lock:
(&(&ehci->lock)->rlock){-.-...}, at: [<ffffffffa0130a69>]
ehci_hrtimer_func+0x29/0xc0 [ehci_hcd]
but this lock took another, HARDIRQ-unsafe lock in the past:
(hcd_urb_list_lock){+.....}
and interrupts could create inverse lock ordering between them.
other info that might help us debug this:
Possible interrupt unsafe locking scenario:
CPU0 CPU1
---- ----
lock(hcd_urb_list_lock);
local_irq_disable();
lock(&(&ehci->lock)->rlock);
lock(hcd_urb_list_lock);
<Interrupt>
lock(&(&ehci->lock)->rlock);
*** DEADLOCK ***
no locks held by swapper/7/0.
the shortest dependencies between 2nd lock and 1st lock:
-> (hcd_urb_list_lock){+.....} ops: 252 {
HARDIRQ-ON-W at:
__lock_acquire+0x602/0x1280
lock_acquire+0xd5/0x1c0
_raw_spin_lock+0x2f/0x40
usb_hcd_unlink_urb_from_ep+0x1b/0x60 [usbcore]
xhci_giveback_urb_in_irq.isra.45+0x70/0x1b0 [xhci_hcd]
finish_td.constprop.60+0x1d8/0x2e0 [xhci_hcd]
xhci_irq+0xdd6/0x1fa0 [xhci_hcd]
usb_hcd_irq+0x26/0x40 [usbcore]
irq_forced_thread_fn+0x2f/0x70
irq_thread+0x149/0x1d0
kthread+0x113/0x150
ret_from_fork+0x2e/0x40
This patch fixes the problem.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Reported-and-tested-by: Bart Van Assche <bart.vanassche@sandisk.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
According to xHCI spec Figure 30: Interrupt Throttle Flow Diagram
If PCI Message Signaled Interrupts (MSI or MSI-X) are enabled,
then the assertion of the Interrupt Pending (IP) flag in Figure 30
generates a PCI Dword write. The IP flag is automatically cleared
by the completion of the PCI write.
the MSI enabled HCs don't need to clear interrupt pending bit, but
hcd->irq = 0 doesn't equal to MSI enabled HCD. At some Dual-role
controller software designs, it sets hcd->irq as 0 to avoid HCD
requesting interrupt, and they want to decide when to call usb_hcd_irq
by software.
Signed-off-by: Peter Chen <peter.chen@nxp.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Don't access any members of a URB after giving it back.
URB might be freed by then already.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Every XHCI TRB has already been traced by the trb trace events.
It is unnecessary to put the same message in kernel log. This
patch removes xhci_debug_trb().
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
XHCI ring changes have already been traced by the ring trace
events. It's unnecessary to put the same messages in kernel
log. This patch removes the debugging code for a ring.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
enq_updates and deq_updates were introduced in the first place
to check whether an xhci hardware is able to respond to trbs
enqueued in the ring. We now have trb tracers to trace every
single enqueue/dequeue trb. It's time to remove them and the
associated debugging code.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch creates a new event class called xhci_log_ring, and
defines the events used for tracing the change of all kinds of
rings used by an xhci host. An xHCI ring is basically a memory
block shared between software and hardware. By tracing changes
of rings, it makes the life easier for debugging hardware or
software problems.
This info can be used, later, to print, in a human readable way,
the life cycle of an xHCI ring using the trace-cmd tool and the
appropriate plugin.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Reviewed-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Introduce a new xhci_hc_died() function that takes care of handling
pending commands and URBs if a host controller becomes unresponsive.
This addresses issues on hotpluggable xhci controllers that disappear
from the bus suddenly, often while the bus (PCI) remove function is
still being processed.
xhci_hc_died() sets a XHCI_STATUS_DYING flag to prevent new URBs and
commands or to be queued. The flag also ensures xhci_hc_died() will
give back pending commands and URBs once.
Host is considered dead if register read returns 0xffffffff, or host
fails to abort the command ring, or fails stopping an endpoint after
trying for 5 seconds.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There's one annoyance in how xhci prints debug messages, we often
get logs with messages but it's hard to say from which device and
endpoint the message originates. Add slot_id, ep_index messages
in handle_tx_event.
Signed-off-by: Zhengjun Xing <zhengjun.xing@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
With these, we can track what's happening with the HW while executing
each and every command. It will give us visibility into how the
different contexts are being modified by xHC which can bring insight
into problems while debugging.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A control transfer that stopped at the status stage incorrectly
warned about a "unexpected TRB Type 4", and did not set the
transferred actual_length for the URB.
The URB actual_length for control transfers should contain the
bytes transferred in the data stage.
Bytes of a partially sent setup stage and missing bytes from
status stage should be left out.
Cc: <stable@vger.kernel.org>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of storing a zero length array of td pointers, and then
allocate memory both for the td pointer array and the td's, just
use a zero length array of actual td's in urb private data.
old:
struct urb_priv {
struct xhci_td *td[0]
}
new:
struct urb_priv {
struct xhci_td td[0]
}
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
urb_priv structure has a count on how many TDs the
URB contains, and how many of those TD's we have handled.
rename:
length -> num_tds
td_cnt -> num_tds_done
No functional changes
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
These three new tracers will help us tie TRBs into URBs by *also*
looking into URB lifetime.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
instead of having a tracer that can only trace command completions,
let's promote this tracer so it can trace and decode any TRB.
With that, it will be easier to extrapolate the lifetime of any TRB
which might help debugging certain issues.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
instead of using while(!list_empty()) followed by list_first_entry(), we
can actually use list_for_each_entry_safe().
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove duplicate code by using trb_to_noop() when
handling Aborted commads
Based on earlier code by Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Useful for turning both transfer and command trbs
into no-ops.
Based on earlier code by Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
xhci_unmap_td_bounce_buffer() already checks for a valid td->bounce_seg
and bails out early if that's invalid. There's no need to check for this
twice.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This way we can remove checks for valid ring from call sites of
xhci_unmap_td_bounce_buffer()
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
handle_tx_event() is not releasing xhci->lock nor reacquiring it, remove
the bogus annotation.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
By extracting xhci_td_cleanup() from finish_td(), code before clearer
and easier to follow.
There are no functional changes with this patch. It's merely a cleanup.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
no functional changes. Simple cleanup to make sure variables are ordered
in a 'reverse christmas tree' fashion. While at that, also remove an
obsolete comment which doesn't apply anymore.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Replace list_entry() with list_first_entry() and list_for_each() with
list_for_each_entry(). This makes the code slightly more readable.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of having several return points, let's use a local variable and
a single place to return. This makes the code slightly easier to read.
[set ret = IRQ_HANDLED in default working case -Mathias]
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cleanup only. This patch is a mechaninal rename to make sure our macros
for TRB completion codes match what the specification uses to refer to
such errors. The idea behind this is that it makes it far easier to grep
the specification and match it with implementation.
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use list_is_singular() to check if cmd_list has only one entry.
[use list_empty() in queue command instead -Mathias]
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
No need to calculate remainder and length_field, if there is
no data phase of a control transfer.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Drop an unnecessary assignment in prepare_transfer().
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The second try was a workaround for (what we thought was) command
ring failing to stop in the first place. But this turns out to be
due to the race that we have fixed(see "xhci: Fix race related to
abort operation"). With that fix, it is time to remove the second
try.
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Checking if the command timeout timer is pending when queueing the
first command to the command ring is not really useful, remove it.
Suggested-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
A counter was used to find out if the stop endpoint completion raced with
the stop endpoint timeout timer. This was needed in case the stop ep
completion failed to delete the timer as it was running on anoter cpu.
The EP_STOP_CMD_PENDING flag was not enough as a new stop endpoint command
may be queued between the command completion and timeout function, which
would set the flag back.
Instead of the separate counter that was used we can detect the race by
checking both the STOP_EP_PENDING flag and timer_pending in the timeout
function.
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We don't want to confuse halted and stalled endpoint states with
a flag indicating we are waiting for a stop endpoint command to
finish or timeout
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
No functional change, De Morgan !(A && B) = (!A || !B)
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
