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
When the isochronous transfer support was introduced, and the xHCI driver
switched to using urb->hcpriv to store an "urb_priv" pointer, a couple of
memory leaks were introduced into the URB enqueue function in its error
handling paths.
xhci_urb_enqueue allocates urb_priv, but it doesn't free it if changing
the control endpoint's max packet size fails or the bulk endpoint is in
the middle of allocating or deallocating streams.
xhci_urb_enqueue also doesn't free urb_priv if any of the four endpoint
types' enqueue functions fail. Instead, it expects those functions to
free urb_priv if an error occurs. However, the bulk, control, and
interrupt enqueue functions do not free urb_priv if the endpoint ring is
NULL. It will, however, get freed if prepare_transfer() fails in those
enqueue functions.
Several of the error paths in the isochronous endpoint enqueue function
also fail to free it. xhci_queue_isoc_tx_prepare() doesn't free urb_priv
if prepare_ring() indicates there is not enough room for all the
isochronous TDs in this URB. If individual isochronous TDs fail to be
queued (perhaps due to an endpoint state change), urb_priv is also leaked.
This argues that the freeing of urb_priv should be done in the function
that allocated it, xhci_urb_enqueue.
This patch looks rather ugly, but refactoring the code will have to wait
because this patch needs to be backported to stable kernels.
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
* 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (115 commits)
EHCI: fix direction handling for interrupt data toggles
USB: serial: add IDs for WinChipHead USB->RS232 adapter
USB: OHCI: fix another regression for NVIDIA controllers
usb: gadget: m66592-udc: add pullup function
usb: gadget: m66592-udc: add function for external controller
usb: gadget: r8a66597-udc: add pullup function
usb: renesas_usbhs: support multi driver
usb: renesas_usbhs: inaccessible pipe is not an error
usb: renesas_usbhs: care buff alignment when dma handler
USB: PL2303: correctly handle baudrates above 115200
usb: r8a66597-hcd: fixup USB_PORT_STAT_C_SUSPEND shift
usb: renesas_usbhs: compile/config are rescued
usb: renesas_usbhs: fixup comment-out
usb: update email address in ohci-sh and r8a66597-hcd
usb: r8a66597-hcd: add function for external controller
EHCI: only power off port if over-current is active
USB: mon: Allow to use usbmon without debugfs
USB: EHCI: go back to using the system clock for QH unlinks
ehci: add pci quirk for Ordissimo and RM Slate 100 too
ehci: refactor pci quirk to use standard dmi_check_system method
...
Fix up trivial conflicts in Documentation/feature-removal-schedule.txt
When the xHCI driver encounters a Missed Service Interval event for an
isochronous endpoint ring, it means the host controller skipped over
one or more isochronous TDs. For TD that is skipped, skip_isoc_td() is
called. This sets the frame descriptor status to -EXDEV, and also sets
the value stored in the int pointed to by status to -EXDEV.
If the isochronous TD happens to be the last TD in an URB,
handle_tx_event() will use the status variable to give back the URB to
the USB core. That means drivers will see urb->status as -EXDEV.
It turns out that EHCI, UHCI, and OHCI always set urb->status to zero for
an isochronous urb, regardless of what the frame status is. See
itd_complete() in ehci-sched.c:
} else {
/* URB was too late */
desc->status = -EXDEV;
}
}
/* handle completion now? */
if (likely ((urb_index + 1) != urb->number_of_packets))
goto done;
/* ASSERT: it's really the last itd for this urb
list_for_each_entry (itd, &stream->td_list, itd_list)
BUG_ON (itd->urb == urb);
*/
/* give urb back to the driver; completion often (re)submits */
dev = urb->dev;
ehci_urb_done(ehci, urb, 0);
ehci_urb_done() completes the URB with the status of the third argument, which
is always zero in this case.
It turns out that many USB webcam drivers, such as uvcvideo, cannot
handle urb->status set to a non-zero value. They will not resubmit
their isochronous URBs in that case, and userspace will see a frozen
video.
Change the xHCI driver to be consistent with the EHCI and UHCI driver,
and always set urb->status to 0 for isochronous URBs.
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: "Xu, Andiry" <Andiry.Xu@amd.com>
Cc: stable@kernel.org
It is one new TRB Completion Code for the xHCI spec v1.0.
Asserted if the xHC detects a problem with a device that does not allow it to
be successfully accessed, e.g. due to a device compliance or compatibility
problem. This error may be returned by any command or transfer, and is fatal
as far as the Slot is concerned. Return -EPROTO by urb->status or frame->status
of ISOC for transfer case. And return -ENODEV for configure endpoint command,
evaluate context command and address device command if there is an incompatible
Device Error. The error codes will be sent back to the USB core to decide how
to do. It's unnecessary for other commands because after the three commands run
successfully means that the device has been accepted.
Signed-off-by: Alex He <alex.he@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
FSE shall occur on the TD natural boundary. The software ep_ring dequeue pointer
exceed the hardware ep_ring dequeue pointer in these cases of Table-3. As a
result, the event_trb(pointed by hardware dequeue pointer) of the FSE can't be
found in the current TD(pointed by software dequeue pointer). What should we do
is to figured out the FSE case and skip over it.
Signed-off-by: Alex He <alex.he@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Some of the recently-added cpu_to_leXX and leXX_to_cpu made things somewhat
messy; this patch neatens some of these areas, removing unnecessary casts
in those parts also. In some places (where Y & Z are constants) a
comparison of (leXX_to_cpu(X) & Y) == Z has been replaced with
(X & cpu_to_leXX(Y)) == cpu_to_leXX(Z). The endian reversal of the
constants should wash out at compile time.
Signed-off-by: Matt Evans <matt@ozlabs.org>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Commit 926008c938 "USB: xhci: simplify logic
of skipping missed isoc TDs" added a small endian bug. This patch
fixes skip_isoc_td() to read the DMA pointer correctly.
Signed-off-by: Matt Evans <matt@ozlabs.org>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The Panther Point chipset has an xHCI host controller that has a limit to
the number of active endpoints it can handle. Ideally, it would signal
that it can't handle anymore endpoints by returning a Resource Error for
the Configure Endpoint command, but they don't. Instead it needs software
to keep track of the number of active endpoints, across configure endpoint
commands, reset device commands, disable slot commands, and address device
commands.
Add a new endpoint context counter, xhci_hcd->num_active_eps, and use it
to track the number of endpoints the xHC has active. This gets a little
tricky, because commands to change the number of active endpoints can
fail. This patch adds a new xHCI quirk for these Intel hosts, and the new
code should not have any effect on other xHCI host controllers.
Fail a new device allocation if we don't have room for the new default
control endpoint. Use the endpoint ring pointers to determine what
endpoints were active before a Reset Device command or a Disable Slot
command, and drop those once the command completes.
Fail a configure endpoint command if it would add too many new endpoints.
We have to be a bit over zealous here, and only count the number of new
endpoints to be added, without subtracting the number of dropped
endpoints. That's because a second configure endpoint command for a
different device could sneak in before we know if the first command is
completed. If the first command dropped resources, the host controller
fails the command for some reason, and we're nearing the limit of
endpoints, we could end up oversubscribing the host.
To fix this race condition, when evaluating whether a configure endpoint
command will fix in our bandwidth budget, only add the new endpoints to
xhci->num_active_eps, and don't subtract the dropped endpoints. Ignore
changed endpoints (ones that are dropped and then re-added), as that
shouldn't effect the host's endpoint resources. When the configure
endpoint command completes, subtract off the dropped endpoints.
This may mean some configuration changes may temporarily fail, but it's
always better to under-subscribe than over-subscribe resources.
(Originally my plan had been to push the resource allocation down into the
ring allocation functions. However, that would cause us to allocate
unnecessary resources when endpoints were changed, because the xHCI driver
allocates a new ring for the changed endpoint, and only deletes the old
ring once the Configure Endpoint command succeeds. A further complication
would have been dealing with the per-device endpoint ring cache.)
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xHCI host controller in the Panther Point chipset sometimes produces
spurious events on the event ring. If it receives a short packet, it
first puts a Transfer Event with a short transfer completion code on the
event ring. Then it puts a Transfer Event with a successful completion
code on the ring for the same TD. The xHCI driver correctly processes the
short transfer completion code, gives the URB back to the driver, and then
prints a warning in dmesg about the spurious event. These warning
messages really fill up dmesg when an HD webcam is plugged into xHCI.
This spurious successful event behavior isn't technically disallowed by
the xHCI specification, so make the xHCI driver just ignore the spurious
completion event.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Unsurprisingly, URBs get submitted and completed a lot in the xHCI
driver. If we have to print 10 lines of debug for every URB submitted
or completed, then that can cause the whole system to stay in the
interrupt handler too long, and can cause Missed Service completion
codes for isochronous transfers.
Cut down the debugging in the URB submission and completion paths:
- Don't squawk about successful transfers, only unsuccessful ones.
- Only print the number of bytes transferred if this was a short
transfer.
- Don't print the endpoint index for successful transfers (will add
more debug to failed transfers to show endpoint index there later).
- Stop printing MMIO writes. This debugging shows up when the endpoint
doorbell is rung a to start a transfer (basically for every URB).
- Don't print out the ring enqueue and dequeue pointers
- Stop printing when we're pointing to a link TRB.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Stop printing out the event ring dequeue pointer and status register in
the operational register set. The host will report an OK status 99% of
the time the interrupt handler is called, and usually when it's really
hosed, a host controller won't even call the interrupt handler. So the
line is really useless.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Remove unnecessary debugging from the xHCI driver. We don't need to
know what function we're calling or returning from. Now I know how to
use markup-oops.pl to de-mystify stack dumps of crashes.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
When an URB is cancelled, the xHCI driver issues a Stop Endpoint command
so that it can manipulate the ring and remove the transfer. The xHC
hardware then places a transfer event with the completion code "Stopped"
or "Stopped Invalid" to let the driver know what TD it was in the middle
of processing. This TD and TRB is stored in ep->stopped_td and
ep->stopped_trb. These pointers are also used in handling stalled
endpoints.
By design, the Stop Endpoint command can race with URB completion. By
the time the Stop Endpoint command is handled, the URBs to be cancelled
may have been given back to the driver. Unfortunately, the stopped_td
and stopped_trb pointers were not getting cleared in this case.
The USB core unconditionally tries to reset the toggle bits on any
endpoints when a new alternate interface setting is installed. When the
xHCI driver saw that ep->stopped_td was still set from the Stop Endpoint
command, xhci_reset_endpoint assumed the endpoint was actually stalled,
and attempted to clean up the endpoint rings. This would manifest
itself in a failed Reset Endpoint command and failed Set TR dequeue
Pointer command after a successful Configure Endpoint command. It may
have also been causing driver oops when the stopped_td was accessed.
This patch should be backported to stable kernels since 2.6.31. Before
2.6.33, stopped_td was found in the xhci_endpoint_ring, not the
xhci_virt_ep.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
When the xHCI driver attempts to cancel a transfer, it issues a Stop
Endpoint command and waits for the host controller to indicate which TRB
it was in the middle of processing. The host will put an event TRB with
completion code COMP_STOP on the event ring if it stops on a control
transfer TRB (or other types of transfer TRBs). The ring handling code
is supposed to set ep->stopped_trb to the TRB that the host stopped on
when this happens.
Unfortunately, there is a long-standing bug in the control transfer
completion code. It doesn't actually check to see if COMP_STOP is set
before attempting to process the transfer based on which part of the
control TD completed. So when we get an event on the data phase of the
control TRB with COMP_STOP set, it thinks it's a normal completion of
the transfer and doesn't set ep->stopped_td or ep->stopped_trb.
When the ring handling code goes on to process the completion of the Stop
Endpoint command, it sees that ep->stopped_trb is not a part of the TD
it's trying to cancel. It thinks the hardware has its enqueue pointer
somewhere further up in the ring, and thinks it's safe to turn the control
TRBs into no-op TRBs. Since the hardware was in the middle of the control
TRBs to be cancelled, the proper software behavior is to issue a Set TR
dequeue pointer command.
It turns out that the NEC host controllers can handle active TRBs being
set to no-op TRBs after a stop endpoint command, but other host
controllers have issues with this out-of-spec software behavior. Fix this
behavior.
This patch should be backported to kernels as far back as 2.6.31, but it
may be a bit challenging, since process_ctrl_td() was introduced in some
refactoring done in 2.6.36, and some endian-safe patches added in 2.6.40
that touch the same lines.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: stable@kernel.org
Currently an isoc URB is divided into multiple TDs, and every TD will
trigger an interrupt when it's processed. However, software can schedule
multiple TDs at a time, and it only needs an interrupt every URB.
xHCI 1.0 introduces the Block Event Interrupt(BEI) flag which allows Normal
and Isoch Transfer TRBs to place an Event TRB on an Event Ring but not
assert an intrrupt to the host, and the interrupt rate is significantly
reduced and the system performance is improved.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Setup Stage Transfer Type field is added to indicate the presence and the
direction of the Data Stage TD, and determines the direction of the Status
Stage TD so the wLength length field should be ignored by the xHC.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xHCI 1.0 specification defines a new isochronous TRB field, called
transfer burst last packet count (TBLPC). This field defines the number
of packets in the last "burst" of packets in a TD. Only SuperSpeed
endpoints can handle more than one burst, so this is set to the number for
packets in a TD for all non-SuperSpeed devices (minus one, since the field
is zero based).
This patch should have no effect on host controllers that don't advertise
the xHCI 1.0 (0x100) version number in their hci_version field.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xHCI 1.0 specification adds a new field to the fourth dword in an
isochronous TRB: the transfer burst count (TBC). This field is only
non-zero for SuperSpeed devices. Each SS endpoint sets the bMaxBurst
field in the SuperSpeed endpoint companion descriptor, which indicates how
many max-packet-sized "bursts" it can handle in one service interval. The
device driver may choose to burst less max packet sized chunks each
service interval (which is defined by one TD). The xHCI driver indicates
to the host controller how many bursts it needs to schedule through the
transfer burst count field.
This patch will only effect xHCI hosts that advertise 1.0 support (0x100)
in the HCI version field of their capabilities register.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xHCI 1.0 specification changes the format of the TD size field in
Normal and Isochronous TRBs. The field in control TRBs is still set to
reserved zero. Instead of representing the number of bytes left to
transfer in the TD (including the current TRB's buffer), it now represents
the number of packets left to transfer (*not* including this TRB).
See section 4.11.2.4 of the xHCI 1.0 specification for details. The math
is basically copied straight from there.
Create a new function, xhci_v1_0_td_remainder(), that should be called for
all xHCI 1.0 host controllers. The field location and maximum value is
still the same, so reuse the old function, xhci_td_remainder(), to handle
the bit shifting.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
It doesn't make sense to set the interrupt on short packet (TRB_ISP) flag
for TRBs queued to endpoints that only receive packets from the host
controller (i.e. OUT endpoints). Packets can only be short when they are
sent from a USB device. Plus, the xHCI 1.0 specification forbids setting
the flag for anything but IN endpoints.
While we're at it, remove some of my snide remarks about the inefficiency
of event data TRBs.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Make the caller loop while there are events to handle, instead.
Signed-off-by: Matt Evans <matt@ozlabs.org>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
On weakly-ordered systems, the reading of an event's content must occur
after reading the event's validity.
Signed-off-by: Matt Evans <matt@ozlabs.org>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
This patch changes the struct members defining access to xHCI device-visible
memory to use __le32/__le64 where appropriate, and then adds swaps where
required. Checked with sparse that all accesses are correct.
MMIO accesses use readl/writel so already are performed LE, but prototypes
now reflect this with __le*.
There were a couple of (debug) instances of DMA pointers being truncated to
32bits which have been fixed too.
Signed-off-by: Matt Evans <matt@ozlabs.org>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
This patch disable the optional PM feature inside the Hudson3 platform under
the following conditions:
1. If an isochronous device is connected to xHCI port and is active;
2. Optional PM feature that powers down the internal Bus PLL when the link is
in low power state is enabled.
The PM feature needs to be disabled to eliminate PLL startup delays when the
link comes out of low power state. The performance of DMA data transfer could
be impacted if system delay were encountered and in addition to the PLL start
up delays. Disabling the PM would leave room for unpredictable system delays
in order to guarantee uninterrupted data transfer to isochronous audio or
video stream devices that require time sensitive information. If data in an
audio/video stream was interrupted then erratic audio or video performance
may be encountered.
AMD PLL quirk is already implemented in OHCI/EHCI driver. After moving the
quirk code to pci-quirks.c and export them, xHCI driver can call it directly
without having the quirk implementation in itself.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
When we get a port status change event, we need to figure out what type of
port it came from: a USB 3.0 port, or a USB 2.0/1.1 port. We can't know
which usb_hcd to use until that point, so hcd will be NULL for part of the
function. Unfortunately, if any of the sanity checks fail, we'll jump to
the cleanup label before hcd is set to a valid pointer, and then we'll
attempt to tell the USB core to kick the hcd, which is NULL.
Skip kicking the roothub if the sanity checks fail.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The logic of the handling Missed Service Error Events was pretty
confusing as we were checking the same condition several times.
In addition, it caused compiler warning since the compiler could
not figure out that event_trb is actually unused in case we are
skipping current TD.
Fix that by rearranging "skip" condition checks, and factor out
skip_isoc_td() so that it is called explicitly.
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Remove 'inline' markings from file-local functions and let compiler
do its job and inline what makes sense for given architecture.
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
There were some places that compared port_speed == -1 where port_speed
is a u8. This doesn't work unless we cast the -1 to u8. Some places
did it correctly.
Instead of using -1 directly, I've created a DUPLICATE_ENTRY define
which does the cast and is more descriptive as well.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
* 'usb-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (172 commits)
USB: Add support for SuperSpeed isoc endpoints
xhci: Clean up cycle bit math used during stalls.
xhci: Fix cycle bit calculation during stall handling.
xhci: Update internal dequeue pointers after stalls.
USB: Disable auto-suspend for USB 3.0 hubs.
USB: Remove bogus USB_PORT_STAT_SUPER_SPEED symbol.
xhci: Return canceled URBs immediately when host is halted.
xhci: Fixes for suspend/resume of shared HCDs.
xhci: Fix re-init on power loss after resume.
xhci: Make roothub functions deal with device removal.
xhci: Limit roothub ports to 15 USB3 & 31 USB2 ports.
xhci: Return a USB 3.0 hub descriptor for USB3 roothub.
xhci: Register second xHCI roothub.
xhci: Change xhci_find_slot_id_by_port() API.
xhci: Refactor bus suspend state into a struct.
xhci: Index with a port array instead of PORTSC addresses.
USB: Set usb_hcd->state and flags for shared roothubs.
usb: Make core allocate resources per PCI-device.
usb: Store bus type in usb_hcd, not in driver flags.
usb: Change usb_hcd->bandwidth_mutex to a pointer.
...
Use XOR to invert the cycle bit, instead of a more complicated
calculation. Eliminate a check for the link TRB type in find_trb_seg().
We know that there will always be a link TRB at the end of a segment, so
xhci_segment->trbs[TRBS_PER_SEGMENT - 1] will always have a link TRB type.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Takashi Iwai <tiwai@suse.de>
When an endpoint stalls, we need to update the xHCI host's internal
dequeue pointer to move it past the stalled transfer. This includes
updating the cycle bit (TRB ownership bit) if we have moved the dequeue
pointer past a link TRB with the toggle cycle bit set.
When we're trying to find the new dequeue segment, find_trb_seg() is
supposed to keep track of whether we've passed any link TRBs with the
toggle cycle bit set. However, this while loop's body
while (cur_seg->trbs > trb ||
&cur_seg->trbs[TRBS_PER_SEGMENT - 1] < trb) {
Will never get executed if the ring only contains one segment.
find_trb_seg() will return immediately, without updating the new cycle
bit. Since find_trb_seg() has no idea where in the segment the TD that
stalled was, make the caller, xhci_find_new_dequeue_state(), check for
this special case and update the cycle bit accordingly.
This patch should be queued to kernels all the way back to 2.6.31.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Takashi Iwai <tiwai@suse.de>
Cc: stable@kernel.org
When an endpoint stalls, the xHCI driver must move the endpoint ring's
dequeue pointer past the stalled transfer. To do that, the driver issues
a Set TR Dequeue Pointer command, which will complete some time later.
Takashi was having issues with USB 1.1 audio devices that stalled, and his
analysis of the code was that the old code would not update the xHCI
driver's ring dequeue pointer after the command completes. However, the
dequeue pointer is set in xhci_find_new_dequeue_state(), just before the
set command is issued to the hardware.
Setting the dequeue pointer before the Set TR Dequeue Pointer command
completes is a dangerous thing to do, since the xHCI hardware can fail the
command. Instead, store the new dequeue pointer in the xhci_virt_ep
structure, and update the ring's dequeue pointer when the Set TR dequeue
pointer command completes.
While we're at it, make sure we can't queue another Set TR Dequeue Command
while the first one is still being processed. This just won't work with
the internal xHCI state code. I'm still not sure if this is the right
thing to do, since we might have a case where a driver queues multiple
URBs to a control ring, one of the URBs Stalls, and then the driver tries
to cancel the second URB. There may be a race condition there where the
xHCI driver might try to issue multiple Set TR Dequeue Pointer commands,
but I would have to think very hard about how the Stop Endpoint and
cancellation code works. Keep the fix simple until when/if we run into
that case.
This patch should be queued to kernels all the way back to 2.6.31.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Takashi Iwai <tiwai@suse.de>
Cc: stable@kernel.org
Make sure the HCD_FLAG_HW_ACCESSIBLE flag is mirrored by both roothubs,
since it refers to whether the shared hardware is accessible. Make sure
each bus is marked as suspended by setting usb_hcd->state to
HC_STATE_SUSPENDED when the PCI host controller is resumed.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
This patch changes the xHCI driver to allocate two roothubs. This touches
the driver initialization and shutdown paths, roothub emulation code, and
port status change event handlers. This is a rather large patch, but it
can't be broken up, or it would break git-bisect.
Make the xHCI driver register its own PCI probe function. This will call
the USB core to create the USB 2.0 roothub, and then create the USB 3.0
roothub. This gets the code for registering a shared roothub out of the
USB core, and allows other HCDs later to decide if and how many shared
roothubs they want to allocate.
Make sure the xHCI's reset method marks the xHCI host controller's primary
roothub as the USB 2.0 roothub. This ensures that the high speed bus will
be processed first when the PCI device is resumed, and any USB 3.0 devices
that have migrated over to high speed will migrate back after being reset.
This ensures that USB persist works with these odd devices.
The reset method will also mark the xHCI USB2 roothub as having an
integrated TT. Like EHCI host controllers with a "rate matching hub" the
xHCI USB 2.0 roothub doesn't have an OHCI or UHCI companion controller.
It doesn't really have a TT, but we'll lie and say it has an integrated
TT. We need to do this because the USB core will reject LS/FS devices
under a HS hub without a TT.
Other details:
-------------
The roothub emulation code is changed to return the correct number of
ports for the two roothubs. For the USB 3.0 roothub, it only reports the
USB 3.0 ports. For the USB 2.0 roothub, it reports all the LS/FS/HS
ports. The code to disable a port now checks the speed of the roothub,
and refuses to disable SuperSpeed ports under the USB 3.0 roothub.
The code for initializing a new device context must be changed to set the
proper roothub port number. Since we've split the xHCI host into two
roothubs, we can't just use the port number in the ancestor hub. Instead,
we loop through the array of hardware port status register speeds and find
the Nth port with a similar speed.
The port status change event handler is updated to figure out whether the
port that reported the change is a USB 3.0 port, or a non-SuperSpeed port.
Once it figures out the port speed, it kicks the proper roothub.
The function to find a slot ID based on the port index is updated to take
into account that the two roothubs will have over-lapping port indexes.
It checks that the virtual device with a matching port index is the same
speed as the passed in roothub.
There's also changes to the driver initialization and shutdown paths:
1. Make sure that the xhci_hcd pointer is shared across the two
usb_hcd structures. The xhci_hcd pointer is allocated and the
registers are mapped in when xhci_pci_setup() is called with the
primary HCD. When xhci_pci_setup() is called with the non-primary
HCD, the xhci_hcd pointer is stored.
2. Make sure to set the sg_tablesize for both usb_hcd structures. Set
the PCI DMA mask for the non-primary HCD to allow for 64-bit or 32-bit
DMA. (The PCI DMA mask is set from the primary HCD further down in
the xhci_pci_setup() function.)
3. Ensure that the host controller doesn't start kicking khubd in
response to port status changes before both usb_hcd structures are
registered. xhci_run() only starts the xHC running once it has been
called with the non-primary roothub. Similarly, the xhci_stop()
function only halts the host controller when it is called with the
non-primary HCD. Then on the second call, it resets and cleans up the
MSI-X irqs.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
xhci_find_slot_id_by_port() tries to map the port index to the slot ID for
the USB device. In the future, there will be two xHCI roothubs, and their
port indices will overlap. Therefore, xhci_find_slot_id_by_port() will
need to use information in the roothub's usb_hcd structure to map the port
index and roothub speed to the right slot ID.
Add a new parameter to xhci_find_slot_id_by_port(), in order to pass in
the roothub's usb_hcd structure.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
There are several variables in the xhci_hcd structure that are related to
bus suspend and resume state. There are a couple different port status
arrays that are accessed by port index. Move those variables into a
separate structure, xhci_bus_state. Stash that structure in xhci_hcd.
When we have two roothhubs that can be suspended and resumed separately,
we can have two xhci_bus_states, and index into the port arrays in each
structure with the fake roothub port index (not the real hardware port
index).
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
In the upcoming patches, the roothub emulation code will need to return
port status and port change buffers based on whether they are called with
the xHCI USB 2.0 or USB 3.0 roothub. To facilitate that, make the roothub
code index into an array of port addresses with wIndex, rather than
calculating the address using the offset and the address of the PORTSC
registers. Later we can set the port array to be the array of USB 3.0
port addresses, or the USB 2.0 port addresses, depending on the roothub
passed in.
Create a temporary (statically sized) port array and fill it in with both
USB 3.0 and USB 2.0 port addresses. This is inefficient to do for every
roothub call, but this is needed for git bisect compatibility. The
temporary port array will be deleted in a subsequent patch.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The hcd->flags are in a sorry state. Some of them are clearly specific to
the particular roothub (HCD_POLL_RH, HCD_POLL_PENDING, and
HCD_WAKEUP_PENDING), but some flags are related to PCI device state
(HCD_HW_ACCESSIBLE and HCD_SAW_IRQ). This is an issue when one PCI device
can have two roothubs that share the same IRQ line and hardware.
Make sure to set HCD_FLAG_SAW_IRQ for both roothubs when an interrupt is
serviced, or an URB is unlinked without an interrupt. (We can't tell if
the host actually serviced an interrupt for a particular bus, but we can
tell it serviced some interrupt.)
HCD_HW_ACCESSIBLE is set once by usb_add_hcd(), which is set for both
roothubs as they are added, so it doesn't need to be modified.
HCD_POLL_RH and HCD_POLL_PENDING are only checked by the USB core, and
they are never set by the xHCI driver, since the roothub never needs to be
polled.
The usb_hcd's state field is a similar mess. Sometimes the state applies
to the underlying hardware: HC_STATE_HALT, HC_STATE_RUNNING, and
HC_STATE_QUIESCING. But sometimes the state refers to the roothub state:
HC_STATE_RESUMING and HC_STATE_SUSPENDED.
Alan Stern recently made the USB core not rely on the hcd->state variable.
Internally, the xHCI driver still checks for HC_STATE_SUSPENDED, so leave
that code in. Remove all references to HC_STATE_HALT, since the xHCI
driver only sets and doesn't test those variables. We still have to set
HC_STATE_RUNNING, since Alan's patch has a bug that means the roothub
won't get registered if we don't set that.
Alan's patch made the USB core check a different variable when trying to
determine whether to suspend a roothub. The xHCI host has a split
roothub, where two buses are registered for one PCI device. Each bus in
the xHCI split roothub can be suspended separately, but both buses must be
suspended before the PCI device can be suspended. Therefore, make sure
that the USB core checks HCD_RH_RUNNING() for both roothubs before
suspending the PCI host.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Make sure to call into the USB core's link, unlink, and giveback URB
functions with the usb_hcd pointer found by using urb->dev->bus. This
will avoid confusion later, when the xHCI driver will deal with URBs from
two separate buses (the USB 3.0 roothub and the faked USB 2.0 roothub).
Assume xhci_urb_dequeue() will be called with the proper usb_hcd.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xHCI driver doesn't ever test hcd->state for HC_STATE_HALT. The USB
core recently stopped using it internally, so there's no point in setting
it in the driver. We still need to set HC_STATE_RUNNING in order to make
it past the USB core's hcd->state check in register_roothub().
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xhci_bus_suspend() and xhci_bus_resume() functions are a bit hard to
read, because they have an ambiguously named variable "port". Rename it
to "port_index". Introduce a new temporary variable, "max_ports" that
holds the maximum number of roothub ports the host controller supports.
This will reduce the number of register reads, and make it easy to change
the maximum number of ports when there are two roothubs.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The test of placing a number of command no-ops on the command ring and
counting the number of no-op events that were generated was only used
during the initial xHCI driver bring up. This test is no longer used, so
delete it.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The expression
while (running_total < sg_dma_len(sg))
does not take into account that the remaining data length can be less
than sg_dma_len(sg). In that case, running_total can end up being
greater than the total data length, so an extra TRB is counted.
Changing the expression to
while (running_total < sg_dma_len(sg) && running_total < temp)
fixes that.
This patch should be queued for stable kernels back to 2.6.31.
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@kernel.org
Calculations like
running_total = TRB_MAX_BUFF_SIZE -
(sg_dma_address(sg) & (TRB_MAX_BUFF_SIZE - 1));
if (running_total != 0)
num_trbs++;
are incorrect, because running_total can never be zero, so the if()
expression will never be true. I think the intention was that
running_total be in the range of 0 to TRB_MAX_BUFF_SIZE-1, not 1
to TRB_MAX_BUFF_SIZE. So adding a
running_total &= TRB_MAX_BUFF_SIZE - 1;
fixes the problem.
This patch should be queued for stable kernels back to 2.6.31.
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@kernel.org
This makes it easier to spot some problems, which will be fixed by the
next patch in the series. Also change dev_dbg to dev_err in
check_trb_math(), so any math errors will be visible even when running
with debug disabled.
Note: This patch changes the expressions containing
"((1 << TRB_MAX_BUFF_SHIFT) - 1)" to use the equivalent
"(TRB_MAX_BUFF_SIZE - 1)". No change in behavior is intended for
those expressions.
This patch should be queued for stable kernels back to 2.6.31.
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@kernel.org
Change the BUGs in xhci_find_new_dequeue_state() to WARN_ONs, to avoid
bringing down the box if one of them is hit
This patch should be queued for stable kernels back to 2.6.31.
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@kernel.org
The unused space in the doorbell is now marked as RsvdZ, not RsvdP, so
we can avoid reading the doorbell before writing it.
Update the doorbell-related defines to produce the entire doorbell value
from a single macro. Document the doorbell format in a comment.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
dev_dbg() is used to print ordinary transfer messages in xhci-ring.c.
System log messages will be flushed if CONFIG_USB_DEBUG is set. Replace the
dev_dbg() with xhci_dbg().
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
giveback_first_trb() controls the cycle bit set of the start_trb, to ensure
that the start_trb is written last and the host controller will receive a
whole td at a time.
However, if the ring is wrapped and cycle bit is toggled to zero, then
giveback_first_trb() will be of no effect. In this case, set the cycle bit of
start_trb to 1 at the beginning and clear it in giveback_first_trb().
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Parameter *td is not used in giveback_first_trb(). Remove it.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Fix the more_trbs_coming field of queue_trb() in isoc transfer.
Signed-off-by: Andiry Xu <andiry.xu@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The original code that resumed the USB bus on a port status change would
only do so when there was a device connected to the port. If a device was
just disconnected, the event would be queued for khubd, but khubd wouldn't
run. That would leave the connect status change (CSC) bit set.
If a USB device was plugged into that same port, the xHCI host controller
would set the current connect status (CCS) bit. But since the CSC bit was
already set, it would not generate an interrupt for a port status change
event. That would mean the user could "Safely Remove" a device, have the
bus suspend, disconnect the device, re-plug it in, and then the device
would never be enumerated.
Plugging in a different device on another port would cause the bus to
resume, and khubd would notice the re-connected device. Running lsusb
would also resume the bus, leading users to report the problem "went away"
when using diagnostic tools.
The solution is to resume the bus when a port status change event is
received, regardless of the port status.
Thank you very much to Maddog for helping me track down this Heisenbug.
This patch should be queued for the 2.6.37 stable tree.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reported-by: Jon 'maddog' Hall <maddog@li.org>
Tested-by: Andiry Xu <andiry.xu@amd.com>
Cc: stable@kernel.org
If the xHCI host controller shares an interrupt line with another device,
the xHCI driver needs to check if the interrupt was generated by its
hardware. Unfortunately, the user will see a ton of "Spurious interrupt."
lines if the other hardware interrupts often. Lawrence found his dmesg
output cluttered with this output when the xHCI host shared an interrupt
with his i915 hardware.
Remove the warning, as sharing an interrupt is a normal thing.
This should be applied to the 2.6.36 stable tree.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Reported-by: Lawrence Rust <lvr@softsystem.co.uk>
Cc: stable@kernel.org
This commit implements port remote wakeup.
When a port is in U3 state and resume signaling is detected from a device,
the port transitions to the Resume state, and the xHC generates a Port Status
Change Event.
For USB3 port, software write a '0' to the PLS field to complete the resume
signaling. For USB2 port, the resume should be signaling for at least 20ms,
irq handler set a timer for port remote wakeup, and then finishes process in
hub_control GetPortStatus.
Some codes are borrowed from EHCI code.
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>
Add software trigger USB device suspend resume function hook.
Do port suspend & resume in terms of xHCI spec.
Port Suspend:
Stop all endpoints via Stop Endpoint Command with Suspend (SP) flag set.
Place individual ports into suspend mode by writing '3' for Port Link State
(PLS) field into PORTSC register. This can only be done when the port is in
Enabled state. When writing, the Port Link State Write Strobe (LWS) bit shall
be set to '1'.
Allocate an xhci_command and stash it in xhci_virt_device to wait completion for
the last Stop Endpoint Command. Use the Suspend bit in TRB to indicate the Stop
Endpoint Command is for port suspend. Based on Sarah's suggestion.
Port Resume:
Write '0' in PLS field, device will transition to running state.
Ring an endpoints' doorbell to restart it.
Ref: USB device remote wake need another patch to implement. For details of
how USB subsystem do power management, please see:
Documentation/usb/power-management.txt
Signed-off-by: Crane Cai <crane.cai@amd.com>
Signed-off-by: Libin Yang <libin.yang@amd.com>
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>
This patch fixes a isoc transfer bug reported by Sander Eikelenboom.
When ep->skip is set, endpoint ring dequeue pointer should be updated
when processed every missed td. Although ring dequeue pointer will also
be updated when ep->skip is clear, leave it intact during missed tds
processing may cause two issues:
1). If the very next valid transfer following missed tds is a short
transfer, its actual_length will be miscalculated;
2). If there are too many missed tds during transfer, new inserted tds
may found the transfer ring full and urb enqueue fails.
Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Tested-by: Sander Eikelenboom <linux@eikelenboom.it>
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>
The code to increment the TRB pointer has a slight ambiguity that could
lead to a bug on different compilers. The ANSI C specification does not
specify the precedence of the assignment operator over the postfix
operator. gcc 4.4 produced the correct code (increment the pointer and
assign the value), but a MIPS compiler that one of John's clients used
assigned the old (unincremented) value.
Remove the unnecessary assignment to make all compilers produce the
correct assembly.
Signed-off-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
To tell the host controller that there are transfers on the endpoint
rings, we need to ring the endpoint doorbell. This is a PCI MMIO write,
which can be delayed until another register read is queued.
The previous code would flush the doorbell write by reading the doorbell
register after the write. This may take time, and it's not necessary to
force the host controller to know about the transfers right away. Don't
flush the doorbell register writes.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The interrupter register set includes a register that says whether interrupts
are pending for each event ring (the IP bit). Each MSI-X vector will get its
own interrupter set with separate IP bits. The status register includes an
"Event Interrupt (EINT)" bit that is set when an IP bit is set in any of the
interrupters.
When PCI interrupts are used, the EINT bit exactly mirrors the IP bit in the
single interrupter set, and it is a waste of time to check both registers when
trying to figure out if the xHC interrupted or another device on the shared IRQ
line interrupted. Only check the IP bit to reduce register reads.
The IP bit is automatically cleared by the xHC when MSI or MSI-X is enabled. It
doesn't make sense to read that register to check for shared interrupts (since
MSI and MSI-X aren't shared). It also doesn't make sense to write to that
register to clear the IP bit, since it is cleared by the hardware.
We can tell whether MSI or MSI-X is enabled by looking at the irq number in
hcd->irq. If it's -1, we know MSI or MSI-X is enabled.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Now that the event handler functions no longer use xhci_set_hc_event_deq()
to update the event ring dequeue pointer, that function is not used by
anything in xhci-ring.c. Move that function into xhci-mem.c and make it
static.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The xHCI specification suggests that writing the hardware event ring dequeue
pointer register too often can be an expensive operation for the xHCI hardware
to manage. It suggests minimizing the number of writes to that register.
Originally, the driver wrote the event ring dequeue pointer after each
event was processed. Depending on how the event ring moderation register
is set up and how fast the transfers are completing, there may be several
events processed for each interrupt. This patch makes the hardware event
ring dequeue pointer be written only once per interrupt.
Make the transfer event handler and port status event handler only write
the software event ring dequeue pointer. Move the updating of the
hardware event ring dequeue pointer into the interrupt function. Move the
contents of xhci_set_hc_event_deq() into the interrupt handler. The
interrupt handler must clear the event handler busy flag, so it might as
well also write the dequeue pointer to the same register. This eliminates
two 32-bit PCI reads and two 32-bit PCI writes.
Reported-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>
xhci_handle_event() is now only called from within xhci-ring.c, so make it
static.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Remove a duplicate register read of the interrupt pending register from
xhci_irq(). Also, remove waiting on the posted write of that register.
The host will see it eventually. It will probably read the register
itself before deciding whether to interrupt the system again, forcing the
posted write to complete.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When we move xhci_work() into xhci_irq(), we don't need to read the operational
register status field twice.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Most of the work for interrupt handling is done in xhci-ring.c, so it makes
sense to move the functions that are first called when an interrupt happens
(xhci_irq() or xhci_msi_irq()) into xhci-ring.c, so that the compiler can better
optimize them.
Shorten some lines to make it pass checkpatch.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
I've been using perf to measure the top symbols while transferring 1GB of data
on a USB 3.0 drive with dd. This is using the raw disk with /dev/sdb, with a
block size of 1K.
During performance testing, the top symbol was xhci_triad_to_transfer_ring(), a
function that should return immediately if streams are not enabled for an
endpoint. It turned out that the functions to find the endpoint ring was
defined in xhci-mem.c and used in xhci-ring.c and xhci-hcd.c. I moved a copy of
xhci_triad_to_transfer_ring() and xhci_urb_to_transfer_ring() into xhci-ring.c
and declared them static. I also made a static version of
xhci_urb_to_transfer_ring() in xhci.c.
This improved throughput on a 1GB read of the raw disk with dd from
186MB/s to 195MB/s, and perf reported sampling the xhci_triad_to_transfer_ring()
0.06% of the time, rather than 9.26% of the time.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch implements isochronous urb enqueue and interrupt handler part.
When an isochronous urb is passed to xHCI driver, first check the transfer
ring to guarantee there is enough room for the whole urb. Then update the
start_frame and interval field of the urb. Always assume URB_ISO_ASAP
is set, and never use urb->start_frame as input.
The number of isoc TDs is equal to urb->number_of_packets. One isoc TD is
consumed every Interval. Each isoc TD consists of an Isoch TRB chained to
zero or more Normal TRBs.
Call prepare_transfer for each TD to do initialization; then calculate the
number of TRBs needed for each TD. If the data required by an isoc TD is
physically contiguous (not crosses a page boundary), then only one isoc TRB
is needed; otherwise one or more additional normal TRB shall be chained to
the isoc TRB by the host.
Set TRB_IOC to the last TRB of each isoc TD. Do not ring endpoint doorbell
to start xHC procession until all the TDs are inserted to the endpoint
transer ring.
In irq handler, update urb status and actual_length, increase
urb_priv->td_cnt. When all the TDs are completed(td_cnt is equal to
urb_priv->length), giveback the urb to usbcore.
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>
Add urb_priv data structure to xHCI driver. This structure allows multiple
xhci TDs to be linked to one urb, which is essential for isochronous
transfer. For non-isochronous urb, only one TD is needed for one urb;
for isochronous urb, the TD number for the urb is equal to
urb->number_of_packets.
The length field of urb_priv indicates the number of TDs in the urb.
The td_cnt field indicates the number of TDs already processed by xHC.
When td_cnt matches length, the urb can be given back to usbcore.
When an urb is dequeued or cancelled, add all the unprocessed TDs to the
endpoint's cancelled_td_list. When process a cancelled TD, increase
td_cnt field. When td_cnt equals urb_priv->length, giveback the
cancelled urb.
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>
This patch adds mechanism to process Missed Service Error Event.
Sometimes the xHC is unable to process the isoc TDs in time, it will
generate Missed Service Error Event. In this case some TDs on the ring are
not processed and missed. When encounter a Missed Servce Error Event, set
the skip flag of the ep, and process the missed TDs until reach the next
processed TD, then clear the skip flag.
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>
This patch adds new cases to trb_comp_code switch, and moves
the switch judgment ahead of fetching td.
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>
This patch moves the bulk and interrupt td processing part in
handle_tx_event() into a separate function process_bulk_intr_td().
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>
This patch moves the ctrl td processing part in handle_tx_event()
into a separate function process_ctrl_td().
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>
This patch moves the td universal processing part in handle_tx_event()
into a separate function finish_td().
if finish_td() returns 1, it indicates the urb can be given back.
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>
Commit 6c12db90f1 also seems to have
introduced a bug that is triggered when the command ring is about to wrap.
The inc_enq() function will not have moved the enqueue pointer past the
link TRB. It is supposed to be moved past the link TRB in prepare_ring(),
which should be called before a TD is enqueued. However, the
queue_command() function never calls the prepare_ring() function because
prepare_ring() is only supposed to be used for endpoint rings. That means
the enqueue pointer will not be moved past the link TRB, and will get
overwritten.
The fix is to make queue_command() call prepare_ring() with a fake
endpoint status (set to running). Then the enqueue pointer will get moved
past the link TRB.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Commit 6c12db90f1 introduced a bug for
control transfers. The patch was supposed to change when the link TRBs at
the end of each ring segment were given to the hardware. If a transfer
descriptor (TD) ended just before the link TRB, the code wouldn't give
back the link TRB to the hardware; instead it would be given back in
prepare_ring() just before the next TD was enqueued at the top of the
ring.
Unfortunately, the code relied on checking the chain bit of the TRB to
determine whether the TD ended just before the link TRB. It assumed that
the ring enqueuing code would call prepare_ring() before enqueuing the
next TD. However, control transfers are made of multiple TDs, and
prepare_ring() is only called once before enqueuing two or three TDs.
If the first or second TD of the control transfer ended just before the
link TRB, then the code in inc_enq() would not move the enqueue pointer
past the link TRB, and the link TRB would get overwritten. This would
cause the xHCI driver to start writing to memory past the ring segment,
and eventually the system would crash or hang.
The fix is to add a flag to inc_enq() that says whether the caller will
enqueue more TDs before calling prepare_ring(). If the chain bit is
cleared (meaning this is the last TRB in a TD), and the caller will not
enqueue more TDs, then we defer giving back the link TRB.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The NEC xHCI host controller firmware version can be found by putting a
vendor-specific command on the command ring and extracting the BCD
encoded-version out of the vendor-specific event TRB.
The firmware version debug line in dmesg will look like:
xhci_hcd 0000:05:00.0: NEC firmware version 30.21
(NEC merged with Renesas Technologies and became Renesas Electronics on
April 1, 2010. I have their OK to merge this vendor-specific code.)
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: Satoshi Otani <satoshi.otani.xm@renesas.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Change transfer ring behavior to not follow/activate link TRBs
until active TRBs are queued after it. This change affects
the behavior when a TD ends just before a link TRB.
Signed-off-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
After using state stored in xhci_virt_ep to clean up a stalled endpoint,
be sure to set the stalled stream ID back to 0.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Change the type of the URB's 'sg' pointer from a usb_sg_request to
a scatterlist. This allows drivers to submit scatter-gather lists
without using the usb_sg_wait() interface. It has the added benefit
of removing the typecasts that were added as part of patch as1368 (and
slightly decreasing the number of pointer dereferences).
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Reviewed-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix usb sparse warnings:
drivers/usb/host/isp1362-hcd.c:2220:50: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:43:24: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:49:24: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:161:24: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:198:16: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:319:31: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:1231:33: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-pci.c:177:23: warning: non-ANSI function declaration of function 'xhci_register_pci'
drivers/usb/host/xhci-pci.c:182:26: warning: non-ANSI function declaration of function 'xhci_unregister_pci'
drivers/usb/host/xhci-ring.c:342:32: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-ring.c:525:34: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-ring.c:1009:32: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-ring.c:1031:32: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-ring.c:1041:16: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-ring.c:1096:30: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-ring.c:1100:27: warning: Using plain integer as NULL pointer
drivers/usb/host/xhci-mem.c:224:27: warning: symbol 'xhci_alloc_container_ctx' was not declared. Should it be static?
drivers/usb/host/xhci-mem.c:242:6: warning: symbol 'xhci_free_container_ctx' was not declared. Should it be static?
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Lothar Wassmann <LW@KARO-electronics.de>
Signed-off By: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Much of the xHCI driver code assumes that endpoints only have one ring.
Now an endpoint can have one ring per enabled stream ID, so correct that
assumption. Use functions that translate the stream_id field in the URB
or the DMA address of a TRB into the correct stream ring.
Correct the polling loop to print out all enabled stream rings. Make the
URB cancellation routine find the correct stream ring if the URB has
stream_id set. Make sure the URB enqueueing routine does the same. Also
correct the code that handles stalled/halted endpoints.
Check that commands and registers that can take stream IDs handle them
properly. That includes ringing an endpoint doorbell, resetting a
stalled/halted endpoint, and setting a transfer ring dequeue pointer
(since that command can set the dequeue pointer in a stream context or an
endpoint context).
Correct the transfer event handler to translate a TRB DMA address into the
stream ring it was enqueued to. Make the code to allocate and prepare TD
structures adds the TD to the right td_list for the stream ring. Make
sure the code to give the first TRB in a TD to the hardware manipulates
the correct stream ring.
When an endpoint stalls, store the stream ID of the stream ring that
stalled in the xhci_virt_ep structure. Use that instead of the stream ID
in the URB, since an URB may be re-used after it is given back after a
non-control endpoint stall.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add support for allocating streams for USB 3.0 bulk endpoints. See
Documentation/usb/bulk-streams.txt for more information about how and why
you would use streams.
When an endpoint has streams enabled, instead of having one ring where all
transfers are enqueued to the hardware, it has several rings. The ring
dequeue pointer in the endpoint context is changed to point to a "Stream
Context Array". This is basically an array of pointers to transfer rings,
one for each stream ID that the driver wants to use.
The Stream Context Array size must be a power of two, and host controllers
can place a limit on the size of the array (4 to 2^16 entries). These
two facts make calculating the size of the Stream Context Array and the
number of entries actually used by the driver a bit tricky.
Besides the Stream Context Array and rings for all the stream IDs, we need
one more data structure. The xHCI hardware will not tell us which stream
ID a transfer event was for, but it will give us the slot ID, endpoint
index, and physical address for the TRB that caused the event. For every
endpoint on a device, add a radix tree to map physical TRB addresses to
virtual segments within a stream ring.
Keep track of whether an endpoint is transitioning to using streams, and
don't enqueue any URBs while that's taking place. Refuse to transition an
endpoint to streams if there are already URBs enqueued for that endpoint.
We need to make sure that freeing streams does not fail, since a driver's
disconnect() function may attempt to do this, and it cannot fail.
Pre-allocate the command structure used to issue the Configure Endpoint
command, and reserve space on the command ring for each stream endpoint.
This may be a bit overkill, but it is permissible for the driver to
allocate all streams in one call and free them in multiple calls. (It is
not advised, however, since it is a waste of resources and time.)
Even with the memory and ring room pre-allocated, freeing streams can
still fail because the xHC rejects the configure endpoint command. It is
valid (by the xHCI 0.96 spec) to return a "Bandwidth Error" or a "Resource
Error" for a configure endpoint command. We should never see a Bandwidth
Error, since bulk endpoints do not effect the reserved bandwidth. The
host controller can still return a Resource Error, but it's improbable
since the xHC would be going from a more resource-intensive configuration
(streams) to a less resource-intensive configuration (no streams).
If the xHC returns a Resource Error, the endpoint will be stuck with
streams and will be unusable for drivers. It's an unavoidable consequence
of broken host controller hardware.
Includes bug fixes from the original patch, contributed by
John Youn <John.Youn@synopsys.com> and Andy Green <AGreen@PLXTech.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1368) fixes a rather obscure bug in usbmon: When tracing
URBs sent by the scatter-gather library, it accesses the data buffers
while they are still mapped for DMA.
The solution is to move the mapping and unmapping out of the s-g
library and into the usual place in hcd.c. This requires the addition
of new URB flag bits to describe the kind of mapping needed, since we
have to call dma_map_sg() if the HCD supports native scatter-gather
operation and dma_map_page() if it doesn't. The nice thing about
having the new flags is that they simplify the testing for unmapping.
The patch removes the only caller of usb_buffer_[un]map_sg(), so those
functions are #if'ed out. A later patch will remove them entirely.
As a result of this change, urb->sg will be set in situations where
it wasn't set previously. Hence the xhci and whci drivers are
adjusted to test urb->num_sgs instead, which retains its original
meaning and is nonzero only when the HCD has to handle a scatterlist.
Finally, even when a submission error occurs we don't want to hand
URBs to usbmon before they are unmapped. The submission path is
rearranged so that map_urb_for_dma() is called only for non-root-hub
URBs and unmap_urb_for_dma() is called immediately after a submission
error. This simplifies the error handling.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The length of the scatter gather list a driver can enqueue is limited by
the bus' sg_tablesize to 62 entries. Each entry will be described by at
least one transfer request block (TRB). If the entry's buffer crosses a
64KB boundary, then that entry will have to be described by two or more
TRBs. So even if the USB device driver respects sg_tablesize, the whole
scatter list may take more than 62 TRBs to describe, and won't fit on
the ring.
Don't assume that an empty ring means there is enough room on the
transfer ring. The old code would unconditionally queue this too-large
transfer, and over write the beginning of the transfer. This would mean
the cycle bit was unchanged in those overwritten transfers, causing the
hardware to think it didn't own the TRBs, and the host would seem to
hang.
Now drivers may see submit_urb() fail with -ENOMEM if the transfers are
too big to fit on the ring.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the USB core installs a new interface, it unconditionally clears the
halts on all the endpoints on the new interface. Usually the xHCI host
needs to know when an endpoint is reset, so it can change its internal
endpoint state. In this case, it doesn't care, because the endpoints were
never halted in the first place.
To avoid issuing a redundant Reset Endpoint command, the xHCI driver looks
at xhci_virt_ep->stopped_td to determine if the endpoint was actually
halted. However, the functions that handle the stall never set that
variable to NULL after it dealt with the stall. So if an endpoint stalled
and a Reset Endpoint command completed, and then the class driver tried to
install a new alternate setting, the xHCI driver would access the old
xhci_virt_ep->stopped_td pointer. A similar problem occurs if the
endpoint has been stopped to cancel a transfer.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
When a USB device is reset, the xHCI hardware must know, in order to match
the device state and disable all endpoints except control endpoint 0.
Issue a Reset Device command after a USB device is successfully reset.
Wait on the command to finish, and then cache or free the disabled
endpoint rings.
There are four different USB device states that the xHCI hardware tracks:
- disabled/enabled - device connection has just been detected,
- default - the device has been reset and has an address of 0,
- addressed - the device has a non-zero address but no configuration has
been set,
- configured - a set configuration succeeded.
The USB core may issue a port reset when a device is in any state, but the
Reset Device command will fail for a 0.96 xHC if the device is not in the
addressed or configured state. Don't consider this failure as an error,
but don't free any endpoint rings if this command fails.
A storage driver may request that the USB device be reset during error
handling, so use GPF_NOIO instead of GPF_KERNEL while allocating memory
for the Reset Device command.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
All commands that can be issued to the xHCI hardware can come back with
vendor-specific "informational" completion codes. These are to be treated
like a successful completion code. Refactor out the code to test for the
range of these codes and print debugging messages.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The xHCI driver issues a Configure Endpoint command for two reasons:
- a new configuration or alternate interface setting is selected
- a quirky Fresco Logic prototype requires the command after a Reset
Endpoint command.
The xHCI driver only waits on the command in the first case.
When a configure endpoint command completes, the driver needs to know why
the command was generated. When the driver only supported selecting an
initial configuration, the check was simple. Unfortunately that check
doesn't work now that the driver supports alternate interfaces. If an
endpoint must be dropped (because it's not in the new alternate setting)
and no new endpoints are added, the math involving
xhci_last_valid_endpoint() will assign -1 to an unsigned integer and cause
an out-of-bounds array access.
Move the check for the quirky hardware sooner and avoid the bad array
access.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The xHCI 0.95 and 0.96 specification defines several transfer buffer
request completion codes that indicate a USB transaction error occurred.
When a stall, babble, transaction, or split transaction error completion code
is set, the xHCI has halted that endpoint ring. Software must issue a
Reset Endpoint command and a Set Transfer Ring Dequeue Pointer command
to clean up the halted ring.
The USB device driver is supposed to call into usb_reset_endpoint() when
an endpoint stalls. That calls into the xHCI driver to issue the proper
commands. However, drivers don't call that function for the other
errors that cause the xHC to halt the endpoint ring. If a babble,
transaction, or split transaction error occurs, check if the endpoint
context reports a halted condition, and clean up the endpoint ring if it
does.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
An xHCI host controller manufacturer can choose to implement several
vendor-specific informational completion codes. These are all to be
treated like a successful transfer completion.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the xHCI hardware says a transfer completed with a split
transaction error, set the URB status to -EPROTO.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The transfer descriptor (TD) is a series of transfer request buffers
(TRBs) that describe the buffer pointer, length, and other
characteristics. The xHCI controllers want to know an estimate of how
long the TD is, for caching reasons. In each TRB, there is a "TD size"
field that provides a rough estimate of the remaining buffers to be
transmitted, including the buffer pointed to by that TRB.
The TD size is 5 bits long, and contains the remaining size in bytes,
right shifted by 10 bits. So a remaining TD size less than 1024 would get
a zero in the TD size field, and a remaining size greater than 32767 would
get 31 in the field.
This patches fixes a bug in the TD_REMAINDER macro that is triggered when
the URB has a scatter gather list with a size bigger than 32767 bytes.
Not all host controllers pay attention to the TD size field, so the bug
will not appear on all USB 3.0 hosts.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It's not surprising that the transfer request buffer (TRB) physical to
virtual address translation function has bugs in it, since I wrote most of
it at 4am last October. Add a test suite to check the TRB math. This
runs at memory initialization time, and causes the driver to fail to load
if the TRB math fails.
Please excuse the excessively long lines in the test vectors; they can't
really be made shorter and still be readable.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
CONFIG_USB_HCD_STAT was used in an abandoned patch to track host
controller throughput statistics. Since CONFIG_USB_HCD_STAT will never be
defined, remove code that can never run.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In order to giveback a canceled URB, we must ensure that the xHCI
hardware will not access the buffer in an URB. We can't modify the
buffer pointers on endpoint rings without issuing and waiting for a stop
endpoint command. Since URBs can be canceled in interrupt context, we
can't wait on that command. The old code trusted that the host
controller would respond to the command, and would giveback the URBs in
the event handler. If the hardware never responds to the stop endpoint
command, the URBs will never be completed, and we might hang the USB
subsystem.
Implement a watchdog timer that is spawned whenever a stop endpoint
command is queued. If a stop endpoint command event is found on the
event ring during an interrupt, we need to stop the watchdog timer with
del_timer(). Since del_timer() can fail if the timer is running and
waiting on the xHCI lock, we need a way to signal to the timer that
everything is fine and it should exit. If we simply clear
EP_HALT_PENDING, a new stop endpoint command could sneak in and set it
before the watchdog timer can grab the lock.
Instead we use a combination of the EP_HALT_PENDING flag and a counter
for the number of pending stop endpoint commands
(xhci_virt_ep->stop_cmds_pending). If we need to cancel the watchdog
timer and del_timer() succeeds, we decrement the number of pending stop
endpoint commands. If del_timer() fails, we leave the number of pending
stop endpoint commands alone. In either case, we clear the
EP_HALT_PENDING flag.
The timer will decrement the number of pending stop endpoint commands
once it obtains the lock. If the timer is the tail end of the last stop
endpoint command (xhci_virt_ep->stop_cmds_pending == 0), and the
endpoint's command is still pending (EP_HALT_PENDING is set), we assume
the host is dying. The watchdog timer will set XHCI_STATE_DYING, try to
halt the xHCI host, and give back all pending URBs.
Various other places in the driver need to check whether the xHCI host
is dying. If the interrupt handler ever notices, it should immediately
stop processing events. The URB enqueue function should also return
-ESHUTDOWN. The URB dequeue function should simply return the value
of usb_hcd_check_unlink_urb() and the watchdog timer will take care of
giving the URB back. When a device is disconnected, the xHCI hardware
structures should be freed without issuing a disable slot command (since
the hardware probably won't respond to it anyway). The debugging
polling loop should stop polling if the host is dying.
When a device is disconnected, any pending watchdog timers are killed
with del_timer_sync(). It must be synchronous so that the watchdog
timer doesn't attempt to access the freed endpoint structures.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In the old code, there was a race condition between the stop endpoint
command and the URB submission process. When the stop endpoint command is
handled by the event handler, the endpoint ring is assumed to be stopped.
When a stop endpoint command is queued, URB submissions are to not ring
the doorbell. The old code would check the number of pending URBs to be
canceled, and would not ring the doorbell if it was non-zero.
However, the following race condition could occur with the old code:
1. Cancel an URB, add it to the list of URBs to be canceled, queue the stop
endpoint command, and increment ep->cancels_pending to 1.
2. The URB finishes on the HW, and an event is enqueued to the event ring
(at the same time as 1).
3. The stop endpoint command finishes, and the endpoint is halted. An
event is queued to the event ring.
4. The event handler sees the finished URB, notices it was to be
canceled, decrements ep->cancels_pending to 0, and removes it from the to
be canceled list.
5. The event handler drops the lock and gives back the URB. The
completion handler requeues the URB (or a different driver enqueues a new
URB). This causes the endpoint's doorbell to be rung, since
ep->cancels_pending == 0. The endpoint is now running.
6. A second URB is canceled, and it's added to the canceled list.
Since ep->cancels_pending == 0, a new stop endpoint command is queued, and
ep->cancels_pending is incremented to 1.
7. The event handler then sees the completed stop endpoint command. The
handler assumes the endpoint is stopped, but it isn't. It attempts to
move the dequeue pointer or change TDs to cancel the second URB, while the
hardware is actively accessing the endpoint ring.
To eliminate this race condition, a new endpoint state bit is introduced,
EP_HALT_PENDING. When this bit is set, a stop endpoint command has been
queued, and the command handler has not begun to process the URB
cancellation list yet. The endpoint doorbell should not be rung when this
is set. Set this when a stop endpoint command is queued, clear it when
the handler for that command runs, and check if it's set before ringing a
doorbell. ep->cancels_pending is eliminated, because it is no longer
used.
Make sure to ring the doorbell for an endpoint when the stop endpoint
command handler runs, even if the canceled URB list is empty. All
canceled URBs could have completed and new URBs could have been enqueued
without the doorbell being rung before the command was handled.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The trb_in_td() function in the xHCI driver is supposed to translate a
physical transfer buffer request (TRB) into a virtual pointer to the ring
segment that TRB is in.
Unfortunately, a mistake in this function may cause endless loops as the
driver searches through the linked list of ring segments over and over
again. Fix a couple bugs that may lead to loops or bad output:
1. Bail out if we get a NULL pointer when translating the segment's
private structure and the starting DMA address of the segment chunk. If
this happens, we've been handed a starting TRB pointer from a different
ring.
2. Make sure the function works when there's multiple segments in the
ring. In the while loop to search through the ring segments, use the
current segment variable (cur_seg), rather than the starting segment
variable (start_seg) that is passed in.
3. Stop searching the ring if we've run through all the segments in the
ring.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
For a USB hub to work under an xHCI host controller, the xHC's internal
scheduler must be made aware of the hub's characteristics. Add an xHCI
hook that the USB core will call after it fetches the hub descriptor.
This hook will add hub information to the slot context for that device,
including whether it has multiple TTs or a single TT, the number of ports
on the hub, and TT think time.
Setting up the slot context for the device is different for 0.95 and 0.96
xHCI host controllers.
Some of the slot context reserved fields in the 0.95 specification were
changed into hub fields in the 0.96 specification. Don't set the TT think
time or number of ports for a hub if we're dealing with a 0.95-compliant
xHCI host controller.
The 0.95 xHCI specification says that to modify the hub flag, we need to
issue an evaluate context command. The 0.96 specification says that flag
can be set with a configure endpoint command. Issue the correct command
based on the version reported by the hardware.
This patch does not add support for multi-TT hubs. Multi-TT hubs expose
a single TT on alt setting 0, and multi-TT on alt setting 1. The xHCI
driver can't handle setting alternate interfaces yet.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In the xHCI driver, configure endpoint commands that are submitted to the
hardware may involve one of two data structures. If the configure
endpoint command is setting up a new configuration or modifying max packet
sizes, the data structures and completions are statically allocated in the
xhci_virt_device structure. If the command is being used to set up
streams or add hub information, then the data structures are dynamically
allocated, and placed on a device command waiting list.
Break out the code to check whether a completed command is in the device
command waiting list. Fix a subtle bug in the old code: continue
processing the command if the command isn't in the wait list. In the old
code, if there was a command in the wait list, but it didn't match the
completed command, the completed command event would be dropped.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Some commands to the xHCI hardware cannot be allowed to fail due to out of
memory issues or the command ring being full.
Add a way to reserve a TRB on the command ring, and make all command
queueing functions indicate whether they are using a reserved TRB.
Add a way to pre-allocate all the memory a command might need. A command
needs an input context, a variable to store the status, and (optionally) a
completion for the caller to wait on. Change all code that assumes the
input device context, status, and completion for a command is stored in
the xhci virtual USB device structure (xhci_virt_device).
Store pending completions in a FIFO in xhci_virt_device. Make the event
handler for a configure endpoint command check to see whether a pending
command in the list has completed. We need to use separate input device
contexts for some configure endpoint commands, since multiple drivers can
submit requests at the same time that require a configure endpoint
command.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The xhci_ring structure contained information that is really related to an
endpoint, not a ring. This will cause problems later when endpoint
streams are supported and there are multiple rings per endpoint.
Move the endpoint state and cancellation information into a new virtual
endpoint structure, xhci_virt_ep. The list of TRBs to be cancelled should
be per endpoint, not per ring, for easy access. There can be only one TRB
that the endpoint stopped on after a stop endpoint command (even with
streams enabled); move the stopped TRB information into the new virtual
endpoint structure. Also move the 31 endpoint rings and temporary ring
storage from the virtual device structure (xhci_virt_device) into the
virtual endpoint structure (xhci_virt_ep).
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Interrupt transfers are submitted to the xHCI hardware using the same TRB
type as bulk transfers. Re-use the bulk transfer enqueueing code to
enqueue interrupt transfers.
Interrupt transfers are a bit different than bulk transfers. When the
interrupt endpoint is to be serviced, the xHC will consume (at most) one
TD. A TD (comprised of sg list entries) can take several service
intervals to transmit. The important thing for device drivers to note is
that if they use the scatter gather interface to submit interrupt
requests, they will not get data sent from two different scatter gather
lists in the same service interval.
For now, the xHCI driver will use the service interval from the endpoint's
descriptor (bInterval). Drivers will need a hook to poll at a more
frequent interval. Set urb->interval to the interval that the xHCI
hardware will use.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Sarah Sharp