The command cancellation code doesn't check whether find_trb_seg()
couldn't find the segment that contains the TRB to be canceled. This
could cause a NULL pointer deference later in the function when next_trb
is called. It's unlikely to happen unless something is wrong with the
command ring pointers, so add some debugging in case it happens.
This patch should be backported to stable kernels as old as 3.0, that
contain the commit b63f4053cc "xHCI:
handle command after aborting the command ring".
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org
When a device with an isochronous endpoint is behind a hub plugged into
the Intel Panther Point xHCI host controller, and the driver submits
multiple frames per URB, the xHCI driver will set the Block Event
Interrupt (BEI) flag on all but the last TD for the URB. This causes
the host controller to place an event on the event ring, but not send an
interrupt. When the last TD for the URB completes, BEI is cleared, and
we get an interrupt for the whole URB.
However, under a Panther Point xHCI host controller, if the parent hub
is unplugged when one or more events from transfers with BEI set are on
the event ring, a port status change event is placed on the event ring,
but no interrupt is generated. This means URBs stop completing, and the
USB device disconnect is not noticed. Something like a USB headset will
cause mplayer to hang when the device is disconnected.
If another transfer is sent (such as running `sudo lsusb -v`), the next
transfer event seems to "unstick" the event ring, the xHCI driver gets
an interrupt, and the disconnect is reported to the USB core.
The fix is not to use the BEI flag under the Panther Point xHCI host.
This will impact power consumption and system responsiveness, because
the xHCI driver will receive an interrupt for every frame in all
isochronous URBs instead of once per URB.
Intel chipset developers confirm that this bug will be hit if the BEI
flag is used on any endpoint, not just ones that are behind a hub.
This patch should be backported to kernels as old as 3.0, that contain
the commit 69e848c209 "Intel xhci: Support
EHCI/xHCI port switching."
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org
removes unnecessary semicolon
Found by Coccinelle: http://coccinelle.lip6.fr/
Signed-off-by: Peter Senna Tschudin <peter.senna@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
drivers/usb/host/xhci.c:1826:14: warning: symbol 'xhci_get_block_size' was not declared. Should it be static?
drivers/usb/host/xhci.c:1844:14: warning: symbol 'xhci_get_largest_overhead' was not declared. Should it be static?
drivers/usb/host/xhci-ring.c:2304:36: warning: context imbalance in 'handle_tx_event' - unexpected unlock
drivers/usb/host/xhci-hub.c:425:6: warning: symbol 'xhci_set_remote_wake_mask' was not declared. Should it be static?
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
According to xHCI spec section 4.6.1.1 and section 4.6.1.2,
after aborting a command on the command ring, xHC will
generate a command completion event with its completion
code set to Command Ring Stopped at least. If a command is
currently executing at the time of aborting a command, xHC
also generate a command completion event with its completion
code set to Command Abort. When the command ring is stopped,
software may remove, add, or rearrage Command Descriptors.
To cancel a command, software will initialize a command
descriptor for the cancel command, and add it into a
cancel_cmd_list of xhci. When the command ring is stopped,
software will find the command trbs described by command
descriptors in cancel_cmd_list and modify it to No Op
command. If software can't find the matched trbs, we can
think it had been finished.
This patch should be backported to kernels as old as 3.0, that contain
the commit 7ed603ecf8 "xhci: Add an
assertion to check for virt_dev=0 bug." That commit papers over a NULL
pointer dereference, and this patch fixes the underlying issue that
caused the NULL pointer dereference.
Signed-off-by: Elric Fu <elricfu1@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Miroslav Sabljic <miroslav.sabljic@avl.com>
Cc: stable@vger.kernel.org
Software have to abort command ring and cancel command
when a command is failed or hang. Otherwise, the command
ring will hang up and can't handle the others. An example
of a command that may hang is the Address Device Command,
because waiting for a SET_ADDRESS request to be acknowledged
by a USB device is outside of the xHC's ability to control.
To cancel a command, software will initialize a command
descriptor for the cancel command, and add it into a
cancel_cmd_list of xhci.
Sarah: Fixed missing newline on "Have the command ring been stopped?"
debugging statement.
This patch should be backported to kernels as old as 3.0, that contain
the commit 7ed603ecf8 "xhci: Add an
assertion to check for virt_dev=0 bug." That commit papers over a NULL
pointer dereference, and this patch fixes the underlying issue that
caused the NULL pointer dereference.
Signed-off-by: Elric Fu <elricfu1@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Miroslav Sabljic <miroslav.sabljic@avl.com>
Cc: stable@vger.kernel.org
Adding cmd_ring_state for command ring. It helps to verify
the current command ring state for controlling the command
ring operations.
This patch should be backported to kernels as old as 3.0. The commit
7ed603ecf8 "xhci: Add an assertion to
check for virt_dev=0 bug." papers over the NULL pointer dereference that
I now believe is related to a timed out Set Address command. This (and
the four patches that follow it) contain the real fix that also allows
VIA USB 3.0 hubs to consistently re-enumerate during the plug/unplug
stress tests.
Signed-off-by: Elric Fu <elricfu1@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Miroslav Sabljic <miroslav.sabljic@avl.com>
Cc: stable@vger.kernel.org
This patch fixes a particularly nasty bug that was revealed by the ring
expansion patches. The bug has been present since the very beginning of
the xHCI driver history, and could have caused general protection faults
from bad memory accesses.
The first thing to note is that a Set TR Dequeue Pointer command can
move the dequeue pointer to a link TRB, if the canceled or stalled
transfer TD ended just before a link TRB. The function to increment the
dequeue pointer, inc_deq, was written before cancellation and stall
support was added. It assumed that the dequeue pointer could never
point to a link TRB. It would unconditionally increment the dequeue
pointer at the start of the function, check if the pointer was now on a
link TRB, and move it to the top of the next segment if so.
This means that if a Set TR Dequeue Point command moved the dequeue
pointer to a link TRB, a subsequent call to inc_deq() would move the
pointer off the segment and into la-la-land. It would then read from
that memory to determine if it was a link TRB. Other functions would
often call inc_deq() until the dequeue pointer matched some other
pointer, which means this function would quite happily read all of
system memory before wrapping around to the right pointer value.
Often, there would be another endpoint segment from a different ring
allocated from the same DMA pool, which would be contiguous to the
segment inc_deq just stepped off of. inc_deq would eventually find the
link TRB in that segment, and blindly move the dequeue pointer back to
the top of the correct ring segment.
The only reason the original code worked at all is because there was
only one ring segment. With the ring expansion patches, the dequeue
pointer would eventually wrap into place, but the dequeue segment would
be out-of-sync. On the second TD after the dequeue pointer was moved to
a link TRB, trb_in_td() would fail (because the dequeue pointer and
dequeue segment were out-of-sync), and this message would appear:
ERROR Transfer event TRB DMA ptr not part of current TD
This fixes bugzilla entry 4333 (option-based modem unhappy on USB 3.0
port: "Transfer event TRB DMA ptr not part of current TD", "rejecting
I/O to offline device"),
https://bugzilla.kernel.org/show_bug.cgi?id=43333
and possibly other general protection fault bugs as well.
This patch should be backported to kernels as old as 2.6.31. A separate
patch will be created for kernels older than 3.4, since inc_deq was
modified in 3.4 and this patch will not apply.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: James Ettle <theholyettlz@googlemail.com>
Tested-by: Matthew Hall <mhall@mhcomputing.net>
Cc: stable@vger.kernel.org
When we encounter an xHCI host that needs the XHCI_TRUST_TX_LENGTH
quirk, the xHCI driver ends up spewing messages about the quirk into
dmesg every time a short packet occurs. Change the xHCI driver to
rate-limit such warnings.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reported-by: Matthew Hall <mhall@mhcomputing.net>
Reported-by: Gary E. Miller <gem@rellim.com>
The Microsoft LifeChat 3000 USB headset was causing a very reproducible
hang whenever it was plugged in. At first, I thought the host
controller was producing bad transfer events, because the log was filled
with errors like:
xhci_hcd 0000:00:14.0: ERROR Transfer event TRB DMA ptr not part of current TD
However, it turned out to be an xHCI driver bug in the ring expansion
patches. The bug is triggered When there are two ring segments, and a
TD that ends just before a link TRB, like so:
______________ _____________
| | ---> | setup TRB B |
______________ | _____________
| | | | data TRB B |
______________ | _____________
| setup TRB A | <-- deq | | data TRB B |
______________ | _____________
| data TRB A | | | | <-- enq, deq''
______________ | _____________
| status TRB A | | | |
______________ | _____________
| link TRB |--------------- | link TRB |
_____________ <--- deq' _____________
TD A (the first control transfer) stalls on the data phase. That halts
the ring. The xHCI driver moves the hardware dequeue pointer to the
first TRB after the stalled transfer, which happens to be the link TRB.
Once the Set TR dequeue pointer command completes, the function
update_ring_for_set_deq_completion runs. That function is supposed to
update the xHCI driver's dequeue pointer to match the internal hardware
dequeue pointer. On the first call this would work fine, and the
software dequeue pointer would move to deq'.
However, if the transfer immediately after that stalled (TD B in this
case), another Set TR Dequeue command would be issued. That would move
the hardware dequeue pointer to deq''. Once that command completed,
update_ring_for_set_deq_completion would run again.
The original code would unconditionally increment the software dequeue
pointer, which moved the pointer off the ring segment into la-la-land.
The while loop would happy increment the dequeue pointer (possibly
wrapping it) until it matched the hardware pointer value.
The while loop would also access all the memory in between the first
ring segment and the second ring segment to determine if it was a link
TRB. This could cause general protection faults, although it was
unlikely because the ring segments came from a DMA pool, and would often
have consecutive memory addresses.
If nothing in that space looked like a link TRB, the deq_seg pointer for
the ring would remain on the first segment. Thus, the deq_seg and the
software dequeue pointer would get out of sync.
When the next transfer event came in after the stalled transfer, the
xHCI driver code would attempt to convert the software dequeue pointer
into a DMA address in order to compare the DMA address for the completed
transfer. Since the deq_seg and the dequeue pointer were out of sync,
xhci_trb_virt_to_dma would return NULL.
The transfer event would get ignored, the transfer would eventually
timeout, and we would mistakenly convert the finished transfer to no-op
TRBs. Some kernel driver (maybe xHCI?) would then get stuck in an
infinite loop in interrupt context, and the whole machine would hang.
This patch should be backported to kernels as old as 3.4, that contain
the commit b008df60c6 "xHCI: count free
TRBs on transfer ring"
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Andiry Xu <andiry.xu@amd.com>
Cc: stable@vger.kernel.org
The upcoming USB 3.0 Link PM patches will introduce new API to enable
and disable low-power link states. We must be able to disable LPM in
order to reset a device, or place the device into U3 (device suspend).
Therefore, we need to make sure the Evaluate Context command to disable
the LPM timeouts can't fail due to there being no room on the command
ring.
Introduce a new flag to the function that queues the Evaluate Context
command, command_must_succeed. This tells the ring handler that a TRB
has already been reserved for the command (by incrementing
xhci->cmd_ring_reserved_trbs), and basically ensures that prepare_ring()
won't fail. A similar flag was already implemented for the Configure
Endpoint command queuing function.
All functions that currently call xhci_configure_endpoint() to issue an
Evaluate Context command pass "false" for the "must_succeed" parameter,
so this patch should have no effect on current xHCI driver behavior.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Sergio reported that when he recorded audio from a USB headset mic
plugged into the USB 3.0 port on his ASUS N53SV-DH72, the audio sounded
"robotic". When plugged into the USB 2.0 port under EHCI on the same
laptop, the audio sounded fine. The device is:
Bus 002 Device 004: ID 046d:0a0c Logitech, Inc. Clear Chat Comfort USB Headset
The problem was tracked down to the Fresco Logic xHCI host controller
not correctly reporting short transfers on isochronous IN endpoints.
The driver would submit a 96 byte transfer, the device would only send
88 or 90 bytes, and the xHCI host would report the transfer had a
"successful" completion code, with an untransferred buffer length of 8
or 6 bytes.
The successful completion code and non-zero untransferred length is a
contradiction. The xHCI host is supposed to only mark a transfer as
successful if all the bytes are transferred. Otherwise, the transfer
should be marked with a short packet completion code. Without the EHCI
bus trace, we wouldn't know whether the xHCI driver should trust the
completion code or the untransferred length. With it, we know to trust
the untransferred length.
Add a new xHCI quirk for the Fresco Logic host controller. If a
transfer is reported as successful, but the untransferred length is
non-zero, print a warning. For the Fresco Logic host, change the
completion code to COMP_SHORT_TX and process the transfer like a short
transfer.
This should be backported to stable kernels that contain the commit
f5182b4155 "xhci: Disable MSI for some
Fresco Logic hosts." That commit was marked for stable kernels as old
as 2.6.36.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reported-by: Sergio Correia <lists@uece.net>
Tested-by: Sergio Correia <lists@uece.net>
Cc: stable@vger.kernel.org
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
This patch changes the output format specifier of a debugging line in
the xhci-hcd driver. An URB's transfer_buffer_length should be
printed in decimal; there's no reason to print it in hex. Especially
since the actual_length value, printed earlier on the same line, is
already in decimal.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Cc: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
While testing unplugging an UVC HD webcam with usb-redirection (so through
usbdevfs), my userspace usb-redir code was getting a value of -1 in
iso_frame_desc[n].status, which according to Documentation/usb/error-codes.txt
is not a valid value.
The source of this -1 is the default case in xhci-ring.c:process_isoc_td()
adding a kprintf there showed the value of trb_comp_code to be COMP_TX_ERR
in this case, so this patch adds handling for that completion code to
process_isoc_td().
This was observed and tested with the following xhci controller:
1033:0194 NEC Corporation uPD720200 USB 3.0 Host Controller (rev 04)
Note: I also wonder if setting frame->status to -1 (-EPERM) is the best we can
do, but since I cannot come up with anything better I've left that as is.
This patch should be backported to kernels as old as 2.6.36, which contain the
commit 04e51901dd "USB: xHCI: Isochronous
transfer implementation".
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org
This commit adds a bit-array to xhci bus_state for keeping track of
which ports are undergoing a resume transition. If any of the bits
are set when xhci_hub_status_data() is called, the routine will return
a non-zero value even if no ports have any status changes pending.
This will allow usbcore to handle races between root-hub suspend and
port wakeup.
This patch should be backported to kernels as old as 3.4, that contain
the commit 879d38e6bc "USB: fix race
between root-hub suspend and remote wakeup".
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@vger.kernel.org
The caller is allowed to specify the GFP flags for these functions.
We should prefer their flags unless we have good reason. For
example, if we take a spin_lock ourselves we'd need to use
GFP_ATOMIC. But in this case it's safe to use the callers GFP
flags.
The callers all pass GFP_ATOMIC here, so this change doesn't affect
how the kernel behaves but we may add other callers later and this
is a cleanup.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
While we're at that, define IMAN bitfield to aid readability.
The interrupt enable bit should be set once on driver init, and we
shouldn't need to continually re-enable it. Commit c21599a3 introduced
a read of the irq_pending register, and that allows us to preserve the
state of the IE bit. Before that commit, we were blindly writing 0x3 to
the register.
This patch should be backported to kernels as old as 2.6.36, or ones
that contain the commit c21599a361 "USB:
xhci: Reduce reads and writes of interrupter registers".
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org
When a urb is submitted to xHCI driver, check if queueing the urb will make
the enqueue pointer advance into dequeue seg and expand the ring if it
occurs. This is to guarantee the safety of ring expansion.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Paul Zimmerman <Paul.Zimmerman@synopsys.com>
If room_on_ring() check fails, try to expand the ring and check again.
When expand a ring, use a cached ring or allocate new segments, link
the original ring and the new ring or segments, update the original ring's
segment numbers and the last segment pointer.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Paul Zimmerman <Paul.Zimmerman@synopsys.com>
In the past, the room_on_ring() check was implemented by walking all over
the ring, which is wasteful and complicated.
Count the number of free TRBs instead. The free TRBs number should be
updated when enqueue/dequeue pointer is updated, or upon the completion
of a set dequeue pointer command.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Paul Zimmerman <Paul.Zimmerman@synopsys.com>
When allocate a ring, store its type - four transfer types for endpoint,
TYPE_STREAM for stream transfer, and TYPE_COMMAND/TYPE_EVENT for xHCI host.
This helps to get rid of three bool function parameters: link_trbs, isoc
and consumer.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Paul Zimmerman <Paul.Zimmerman@synopsys.com>
There's really no point in having hcd->irq as a
signed integer when we consider the fact that
IRQ 0 means NO_IRQ. In order to avoid confusion,
make hcd->irq unsigned and fix users who were
passing -1 as the IRQ number to usb_add_hcd.
Tested-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
USB 3.0 hubs don't have a port suspend change bit (that bit is now
reserved). Instead, when a host-initiated resume finishes, the hub sets
the port link state change bit.
When a USB 3.0 device initiates remote wakeup, the parent hubs with
their upstream links in U3 will pass the LFPS up the chain. The first
hub that has an upstream link in U0 (which may be the roothub) will
reflect that LFPS back down the path to the device.
However, the parent hubs in the resumed path will not set their link
state change bit. Instead, the device that initiated the resume has to
send an asynchronous "Function Wake" Device Notification up to the host
controller. Therefore, we need a way to notify the USB core of a device
resume without going through the normal hub URB completion method.
First, make the xHCI roothub act like an external USB 3.0 hub and not
pass up the port link state change bit when a device-initiated resume
finishes. Introduce a new xHCI bit field, port_remote_wakeup, so that
we can tell the difference between a port coming out of the U3Exit state
(host-initiated resume) and the RExit state (ending state of
device-initiated resume).
Since the USB core can't tell whether a port on a hub has resumed by
looking at the Hub Status buffer, we need to introduce a bitfield,
wakeup_bits, that indicates which ports have resumed. When the xHCI
driver notices a port finishing a device-initiated resume, we call into
a new USB core function, usb_wakeup_notification(), that will set
the right bit in wakeup_bits, and kick khubd for that hub.
We also call usb_wakeup_notification() when the Function Wake Device
Notification is received by the xHCI driver. This covers the case where
the link between the roothub and the first-tier hub is in U0, and the
hub reflects the resume signaling back to the device without giving any
indication it has done so until the device sends the Function Wake
notification.
Change the code in khubd that handles the remote wakeup to look at the
state the USB core thinks the device is in, and handle the remote wakeup
if the port's wakeup bit is set.
This patch only takes care of the case where the device is attached
directly to the roothub, or the USB 3.0 hub that is attached to the root
hub is the device sending the Function Wake Device Notification (e.g.
because a new USB device was attached). The other cases will be covered
in a second patch.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
When the USB 3.0 hub support went in, I disabled selective suspend for
all external USB 3.0 hubs because they used a different mechanism to
enable remote wakeup. In fact, other USB 3.0 devices that could signal
remote wakeup would have been prevented from going into suspend because
they would have stalled the SetFeature Device Remote Wakeup request.
This patch adds support for the USB 3.0 way of enabling remote wake up
(with a SetFeature Function Suspend request), and enables selective
suspend for all hubs during hub_probe. It assumes that all USB 3.0 have
only one "function" as defined by the interface association descriptor,
which is true of all the USB 3.0 devices I've seen so far. FIXME if
that turns out to change later.
After a device signals a remote wakeup, it is supposed to send a Device
Notification packet to the host controller, signaling which function
sent the remote wakeup. The host can then put any other functions back
into function suspend. Since we don't have support for function suspend
(and no devices currently support it), we'll just assume the hub
function will resume the device properly when it received the port
status change notification, and simply ignore any device notification
events from the xHCI host controller.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
xHCI roothubs go through slightly different port state machines when
either a device initiates a remote wakeup and signals resume, or when
the host initiates a resume.
According to section 4.19.1.2.13 of the xHCI 1.0 spec, on host-initiated
resume, the xHC port state machine automatically goes through the U3Exit
state into the U0 state, setting the port link state change (PLC) bit in
the process.
When a device initiates resume, the xHCI port state machine goes into
the "Resume" state and sets the PLC bit. Then the xHCI driver writes U0
into the port link state register to transition the port to U0 from the
Resume state.
We can't be sure the device is actually in the U0 state until we receive
the next port status change event with the PLC bit set. We really don't
want khubd to be polling the roothub port status bits until the device
is really in U0.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Andiry Xu <andiry.xu@amd.com>
When a TD length mismatch is found during isoc TRB enqueue, it directly
returns -EINVAL. However, isoc transfer is partially enqueued at this time,
and the ring should be cleared.
This should be backported to kernels as old as 2.6.36, which contain the
commit 522989a27c "xhci: Fix failed
enqueue in the middle of isoch TD."
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org
The xHCI hub port code gets passed a zero-based port number by the USB
core. It then adds one to in order to find a device slot by port number
and device speed by calling xhci_find_slot_id_by_port. That function
clearly states it requires a one-based port number. The xHCI port
status change event handler was using a zero-based port number that it
got from find_faked_portnum_from_hw_portnum, not a one-based port
number. This lead to the doorbells never being rung for a device after
a resume, or worse, a different device with the same speed having its
doorbell rung (which could lead to bad power management in the xHCI host
controller).
This patch should be backported to kernels as old as 2.6.39.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Andiry Xu <andiry.xu@amd.com>
Cc: stable@vger.kernel.org
Randy Dunlap points out that commit 9258c0b2 "xhci: Better debugging for
critical host errors." introduces some new build warnings on 32-bit
builds:
drivers/usb/host/xhci-ring.c:1936:3: warning: format '%016llx' expects type 'long long unsigned int', but argument 3 has type 'dma_addr_t'
drivers/usb/host/xhci-ring.c:1958:3: warning: format '%016llx' expects type 'long long unsigned int', but argument 3 has type 'dma_addr_t'
Cast the results of xhci_trb_virt_to_dma() from a dma_addr_t to an
unsigned long long.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reported-by: Randy Dunlap <rdunlap@xenotime.net>
When a host controller gives a bad event TRB, we should print out the
contents of the TRB as a warning so that users don't have to recompile
their kernel to get information about what went wrong. Also, print out
the event ring if they have xHCI debugging turned on, since previous
events can often explain what happened before the bad TRB occurred.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
With devices that can need up to 128 segments (with 64 TRBs per
segment), we can't afford to print out the entire endpoint ring every
time an URB is canceled. Instead, print the offset of the TRB, along
with device pathname and endpoint number.
Only print DMA addresses, since virtual addresses of internal structures
are not useful. Change the cancellation code to be more clear about
what steps of the cancellation it is in the process of doing (queueing
the request, handling the stop endpoint command, turning the TDs into
no-ops, or moving the dequeue pointers).
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The code for toggling the cycle bits when the ring wraps around has
worked for years. The print statement alone is not enough to indicate
there's something wrong with that code. Now that full transfer tracing
has been ripped out, the print statement or lack thereof won't help
without context of where the enqueue pointer is.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Users can trace the submission of URBs through USBmon, so it makes no
sense to have duplicate debugging in the xHCI driver.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Remove verbose debugging about scatter-gather lists, as we haven't had
an issue with scatter gather list math for about a year now. The
debugging didn't help before, and just clutters up the log file when
trying to debug other issues.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Getting a short packet or a babble error is usually a recoverable error,
so stop scaring users with warnings in dmesg when xHCI debugging is turned
off.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Add a new field num_mapped_sgs to struct urb so that we have a place to
store the number of mapped entries and can also retain the original
value of entries in num_sgs. Previously, usb_hcd_map_urb_for_dma()
would overwrite this with the number of mapped entries, which would
break dma_unmap_sg() because it requires the original number of entries.
This fixes warnings like the following when using USB storage devices:
------------[ cut here ]------------
WARNING: at lib/dma-debug.c:902 check_unmap+0x4e4/0x695()
ehci_hcd 0000:00:12.2: DMA-API: device driver frees DMA sg list with different entry count [map count=4] [unmap count=1]
Modules linked in: ohci_hcd ehci_hcd
Pid: 0, comm: kworker/0:1 Not tainted 3.2.0-rc2+ #319
Call Trace:
<IRQ> [<ffffffff81036d3b>] warn_slowpath_common+0x80/0x98
[<ffffffff81036de7>] warn_slowpath_fmt+0x41/0x43
[<ffffffff811fa5ae>] check_unmap+0x4e4/0x695
[<ffffffff8105e92c>] ? trace_hardirqs_off+0xd/0xf
[<ffffffff8147208b>] ? _raw_spin_unlock_irqrestore+0x33/0x50
[<ffffffff811fa84a>] debug_dma_unmap_sg+0xeb/0x117
[<ffffffff8137b02f>] usb_hcd_unmap_urb_for_dma+0x71/0x188
[<ffffffff8137b166>] unmap_urb_for_dma+0x20/0x22
[<ffffffff8137b1c5>] usb_hcd_giveback_urb+0x5d/0xc0
[<ffffffffa0000d02>] ehci_urb_done+0xf7/0x10c [ehci_hcd]
[<ffffffffa0001140>] qh_completions+0x429/0x4bd [ehci_hcd]
[<ffffffffa000340a>] ehci_work+0x95/0x9c0 [ehci_hcd]
...
---[ end trace f29ac88a5a48c580 ]---
Mapped at:
[<ffffffff811faac4>] debug_dma_map_sg+0x45/0x139
[<ffffffff8137bc0b>] usb_hcd_map_urb_for_dma+0x22e/0x478
[<ffffffff8137c494>] usb_hcd_submit_urb+0x63f/0x6fa
[<ffffffff8137d01c>] usb_submit_urb+0x2c7/0x2de
[<ffffffff8137dcd4>] usb_sg_wait+0x55/0x161
Signed-off-by: Clemens Ladisch <clemens@ladisch.de>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This pulls in the latest USB bugfixes and helps a few of the drivers
merge nicer in the future due to changes in both branches.
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The HCD_FLAG_SAW_IRQ flag was introduced in order to catch IRQ routing
errors: If an URB was unlinked and the host controller hadn't gotten
any IRQs, it seemed likely that the IRQs were directed to the wrong
vector.
This warning hasn't come up in many years, as far as I know; interrupt
routing now seems to be well under control. Therefore there's no
reason to keep the flag around any more. This patch (as1495) finally
removes it.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb: (260 commits)
usb: renesas_usbhs: fixup inconsistent return from usbhs_pkt_push()
usb/isp1760: Allow to optionally trigger low-level chip reset via GPIOLIB.
USB: gadget: midi: memory leak in f_midi_bind_config()
USB: gadget: midi: fix range check in f_midi_out_open()
QE/FHCI: fixed the CONTROL bug
usb: renesas_usbhs: tidyup for smatch warnings
USB: Fix USB Kconfig dependency problem on 85xx/QoirQ platforms
EHCI: workaround for MosChip controller bug
usb: gadget: file_storage: fix race on unloading
USB: ftdi_sio.c: Use ftdi async_icount structure for TIOCMIWAIT, as in other drivers
USB: ftdi_sio.c:Fill MSR fields of the ftdi async_icount structure
USB: ftdi_sio.c: Fill LSR fields of the ftdi async_icount structure
USB: ftdi_sio.c:Fill TX field of the ftdi async_icount structure
USB: ftdi_sio.c: Fill the RX field of the ftdi async_icount structure
USB: ftdi_sio.c: Basic icount infrastructure for ftdi_sio
usb/isp1760: Let OF bindings depend on general CONFIG_OF instead of PPC_OF .
USB: ftdi_sio: Support TI/Luminary Micro Stellaris BD-ICDI Board
USB: Fix runtime wakeup on OHCI
xHCI/USB: Make xHCI driver have a BOS descriptor.
usb: gadget: add new usb gadget for ACM and mass storage
...
Setting the chain (CH) bit in the link TRB of isochronous transfer rings
is required by AMD 0.96 xHCI host controller to successfully transverse
multi-TRB TD that span through different memory segments.
When a Missed Service Error event occurs, if the chain bit is not set in
the link TRB and the host skips TDs which just across a link TRB, the
host may falsely recognize the link TRB as a normal TRB. You can see
this may cause big trouble - the host does not jump to the right address
which is pointed by the link TRB, but continue fetching the memory which
is after the link TRB address, which may not even belong to the host,
and the result cannot be predicted.
This causes some big problems. Without the former patch I sent: "xHCI:
prevent infinite loop when processing MSE event", the system may hang.
With that patch applied, system does not hang, but the host still access
wrong memory address and isoc transfer will fail. With this patch,
isochronous transfer works as expected.
This patch should be applied to kernels as old as 2.6.36, which was when
the first isochronous support was added for the xHCI host controller.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the link state changes, xHC will report a port status change event
and set the PORT_PLC bit, for both USB3 and USB2 root hub ports.
The PLC will be cleared by usbcore for USB3 root hub ports, but not for
USB2 ports, because they do not report USB_PORT_STAT_C_LINK_STATE in
wPortChange.
Clear it for USB2 root hub ports in handle_port_status().
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When a xHC host is unable to handle isochronous transfer in the
interval, it reports a Missed Service Error event and skips some tds.
Currently xhci driver handles MSE event in the following ways:
1. When encounter a MSE event, set ep->skip flag, update event ring
dequeue pointer and return.
2. When encounter the next event on this ep, the driver will run the
do-while loop, fetch td from ep's td_list to find the td
corresponding to this event. All tds missed are marked as short
transfer(-EXDEV).
The do-while loop will end in two ways:
1. If the td pointed by the event trb is found;
2. If the ep ring's td_list is empty.
However, if a buggy HW reports some unpredicted event (for example, an
overrun event following a MSE event while the ep ring is actually not
empty), the driver will never find the td, and it will loop until the
td_list is empty.
Unfortunately, the spinlock is dropped when give back a urb in the
do-while loop. During the spinlock released period, the class driver
may still submit urbs and add tds to the td_list. This may cause
disaster, since the td_list will never be empty and the loop never ends,
and the system hangs.
To fix this, count the number of TDs on the ep ring before skipping TDs,
and quit the loop when skipped that number of tds. This guarantees the
do-while loop will end after certain number of cycles, and driver will
not be trapped in an infinite loop.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Now that the xHCI driver always return a status value of zero for isochronous
URBs, when the last TD of an isochronous URB is short, the local variable
"status" stays set to -EINPROGRESS. When xHCI driver debugging is turned on,
this causes the log file to fill with messages like this:
[ 38.859282] xhci_hcd 0000:00:14.0: Giveback URB ffff88013ad47800, len = 1408, expected = 580, status = -115
Don't print out the status of an URB for isochronous URBs.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Now ${LINUX}/drivers/usb/* can use usb_endpoint_maxp(desc) to get maximum packet size
instead of le16_to_cpu(desc->wMaxPacketSize).
This patch fix it up
Cc: Armin Fuerst <fuerst@in.tum.de>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Johannes Erdfelt <johannes@erdfelt.com>
Cc: Vojtech Pavlik <vojtech@suse.cz>
Cc: Oliver Neukum <oliver@neukum.name>
Cc: David Kubicek <dave@awk.cz>
Cc: Johan Hovold <jhovold@gmail.com>
Cc: Brad Hards <bhards@bigpond.net.au>
Acked-by: Felipe Balbi <balbi@ti.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Dahlmann <dahlmann.thomas@arcor.de>
Cc: David Brownell <david-b@pacbell.net>
Cc: David Lopo <dlopo@chipidea.mips.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Michal Nazarewicz <m.nazarewicz@samsung.com>
Cc: Xie Xiaobo <X.Xie@freescale.com>
Cc: Li Yang <leoli@freescale.com>
Cc: Jiang Bo <tanya.jiang@freescale.com>
Cc: Yuan-hsin Chen <yhchen@faraday-tech.com>
Cc: Darius Augulis <augulis.darius@gmail.com>
Cc: Xiaochen Shen <xiaochen.shen@intel.com>
Cc: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Cc: OKI SEMICONDUCTOR, <toshiharu-linux@dsn.okisemi.com>
Cc: Robert Jarzmik <robert.jarzmik@free.fr>
Cc: Ben Dooks <ben@simtec.co.uk>
Cc: Thomas Abraham <thomas.ab@samsung.com>
Cc: Herbert Pötzl <herbert@13thfloor.at>
Cc: Arnaud Patard <arnaud.patard@rtp-net.org>
Cc: Roman Weissgaerber <weissg@vienna.at>
Acked-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Tony Olech <tony.olech@elandigitalsystems.com>
Cc: Florian Floe Echtler <echtler@fs.tum.de>
Cc: Christian Lucht <lucht@codemercs.com>
Cc: Juergen Stuber <starblue@sourceforge.net>
Cc: Georges Toth <g.toth@e-biz.lu>
Cc: Bill Ryder <bryder@sgi.com>
Cc: Kuba Ober <kuba@mareimbrium.org>
Cc: Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com>
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
For a long time, the xHCI driver has had this note:
/* FIXME: Ignoring zero-length packets, can those happen? */
It turns out that, yes, there are drivers that need to queue zero-length
transfers for isochronous OUT transfers. Without this patch, users will
see kernel hang messages when a driver attempts to enqueue an isochronous
URB with a zero length transfer (because count_isoc_trbs_needed will return
zero for that TD, xhci_td->last_trb will never be set, and updating the
dequeue pointer will cause an infinite loop).
Matěj ran into this issue when using an NI Audio4DJ USB soundcard
with the snd-usb-caiaq driver. See
https://bugzilla.kernel.org/show_bug.cgi?id=40702
Fix count_isoc_trbs_needed() to return 1 for zero-length transfers (thanks
Alan on the math help). Update the various TRB field calculations to deal
with zero-length transfers. We're still transferring one packet with a
zero-length data payload, so the total_packet_count should be 1. The
Transfer Burst Count (TBC) and Transfer Last Burst Packet Count (TLBPC)
fields should be set to zero.
This patch should be backported to kernels as old as 2.6.36.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Matěj Laitl <matej@laitl.cz>
Cc: Daniel Mack <zonque@gmail.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: stable@kernel.org
When a driver tries to cancel an URB, and the host controller is dying,
xhci_urb_dequeue will giveback the URB without removing the xhci_tds
that comprise that URB from the td_list or the cancelled_td_list. This
can cause a race condition between the driver calling URB dequeue and
the stop endpoint command watchdog timer.
If the timer fires on a dying host, and a driver attempts to resubmit
while the watchdog timer has dropped the xhci->lock to giveback a
cancelled URB, URBs may be given back by the xhci_urb_dequeue() function.
At that point, the URB's priv pointer will be freed and set to NULL, but
the TDs will remain on the td_list. This will cause an oops in
xhci_giveback_urb_in_irq() when the watchdog timer attempts to loop
through the endpoints' td_lists, giving back killed URBs.
Make sure that xhci_urb_dequeue() removes TDs from the TD lists and
canceled TD lists before it gives back the URB.
This patch should be backported to kernels as old as 2.6.36.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Andiry Xu <andiry.xu@amd.com>
Cc: stable@kernel.org
When an isochronous transfer is enqueued, xhci_queue_isoc_tx_prepare()
will ensure that there is enough room on the transfer rings for all of the
isochronous TDs for that URB. However, when xhci_queue_isoc_tx() is
enqueueing individual isoc TDs, the prepare_transfer() function can fail
if the endpoint state has changed to disabled, error, or some other
unknown state.
With the current code, if Nth TD (not the first TD) fails, the ring is
left in a sorry state. The partially enqueued TDs are left on the ring,
and the first TRB of the TD is not given back to the hardware. The
enqueue pointer is left on the TRB after the last successfully enqueued
TD. This means the ring is basically useless. Any new transfers will be
enqueued after the failed TDs, which the hardware will never read because
the cycle bit indicates it does not own them. The ring will fill up with
untransferred TDs, and the endpoint will be basically unusable.
The untransferred TDs will also remain on the TD list. Since the td_list
is a FIFO, this basically means the ring handler will be waiting on TDs
that will never be completed (or worse, dereference memory that doesn't
exist any more).
Change the code to clean up the isochronous ring after a failed transfer.
If the first TD failed, simply return and allow the xhci_urb_enqueue
function to free the urb_priv. If the Nth TD failed, first remove the TDs
from the td_list. Then convert the TRBs that were enqueued into No-op
TRBs. Make sure to flip the cycle bit on all enqueued TRBs (including any
link TRBs in the middle or between TDs), but leave the cycle bit of the
first TRB (which will show software-owned) intact. Then move the ring
enqueue pointer back to the first TRB and make sure to change the
xhci_ring's cycle state to what is appropriate for that ring segment.
This ensures that the No-op TRBs will be overwritten by subsequent TDs,
and the hardware will not start executing random TRBs because the cycle
bit was left as hardware-owned.
This bug is unlikely to be hit, but it was something I noticed while
tracking down the watchdog timer issue. I verified that the fix works by
injecting some errors on the 250th isochronous URB queued, although I
could not verify that the ring is in the correct state because uvcvideo
refused to talk to the device after the first usb_submit_urb() failed.
Ring debugging shows that the ring looks correct, however.
This patch should be backported to kernels as old as 2.6.36.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Andiry Xu <andiry.xu@amd.com>
Cc: stable@kernel.org
