If I unplug a device while the UAS driver is loaded, I get an oops
in usb_free_streams(). This is because usb_unbind_interface() calls
usb_disable_interface() which calls usb_disable_endpoint() which sets
ep_out and ep_in to NULL. Then the UAS driver calls usb_pipe_endpoint()
which returns a NULL pointer and passes an array of NULL pointers to
usb_free_streams().
I think the correct fix for this is to check for the NULL pointer
in usb_free_streams() rather than making the driver check for this
situation. My original patch for this checked for dev->state ==
USB_STATE_NOTATTACHED, but the call to usb_disable_interface() is
conditional, so not all drivers would want this check.
Note from Sarah Sharp: This patch does avoid a potential dereference,
but the real fix (which will be implemented later) is to set the
.soft_unbind flag in the usb_driver structure for the UAS driver, and
all drivers that allocate streams. The driver should free any streams
when it is unbound from the interface. This avoids leaking stream rings
in the xHCI driver when usb_disable_interface() is called.
This should be queued for stable trees back to 2.6.35.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@kernel.org
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>
Introduce the notion of a PCI device that may be associated with more than
one USB host controller driver (struct usb_hcd). This patch is the start
of the work to separate the xHCI host controller into two roothubs: a USB
3.0 roothub with SuperSpeed-only ports, and a USB 2.0 roothub with
HS/FS/LS ports.
One usb_hcd structure is designated to be the "primary HCD", and a pointer
is added to the usb_hcd structure to keep track of that. A new function
call, usb_hcd_is_primary_hcd() is added to check whether the USB hcd is
marked as the primary HCD (or if it is not part of a roothub pair). To
allow the USB core and xHCI driver to access either roothub in a pair, a
"shared_hcd" pointer is added to the usb_hcd structure.
Add a new function, usb_create_shared_hcd(), that does roothub allocation
for paired roothubs. It will act just like usb_create_hcd() did if the
primary_hcd pointer argument is NULL. If it is passed a non-NULL
primary_hcd pointer, it sets usb_hcd->shared_hcd and usb_hcd->primary_hcd
fields. It will also skip the bandwidth_mutex allocation, and set the
secondary hcd's bandwidth_mutex pointer to the primary HCD's mutex.
IRQs are only allocated once for the primary roothub.
Introduce a new usb_hcd driver flag that indicates the host controller
driver wants to create two roothubs. If the HCD_SHARED flag is set, then
the USB core PCI probe methods will allocate a second roothub, and make
sure that second roothub gets freed during rmmod and in initialization
error paths.
When usb_hc_died() is called with the primary HCD, make sure that any
roothubs that share that host controller are also marked as being dead.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The xHCI driver essentially has both a USB 2.0 and a USB 3.0 roothub. So
setting the HCD_USB3 bits in the hcd->driver->flags is a bit misleading.
Add a new field to usb_hcd, bcdUSB. Store the result of
hcd->driver->flags & HCD_MASK in it. Later, when we have the xHCI driver
register the two roothubs, we'll set the usb_hcd->bcdUSB field to HCD_USB2
for the USB 2.0 roothub, and HCD_USB3 for the USB 3.0 roothub.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Change the bandwith_mutex in struct usb_hcd to a pointer. This will allow
the pointer to be shared across usb_hcds for the upcoming work to split
the xHCI driver roothub into a USB 2.0/1.1 and a USB 3.0 bus.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Refactor out the code in usb_add_hcd() to request the IRQ line for the
HCD. This will only need to be called once for the two xHCI roothubs, so
it's easier to refactor it into a function, rather than wrapping the long
if-else block into another if statement.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Make the USB 3.0 roothub registered by the USB core have a SuperSpeed
Endpoint Companion Descriptor after the interrupt endpoint. All USB 3.0
devices are required to have this, and the USB 3.0 bus specification
(section 10.13.1) says which values the descriptor should have.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
We would like to allow host controller drivers to stop using hcd->state.
Unfortunately, some host controller drivers use hcd->state as an
implicit way of telling the core that a controller has died. The
roothub registration functions must assume the host died if hcd->state
equals HC_STATE_HALT.
To facilitate drivers that don't want to set hcd->state to
HC_STATE_RUNNING in their initialization routines, we set the state to
running before calling the host controller's start function.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The irq enabling code is going to be refactored into a new function, so
clean up some checkpatch errors before moving it.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
The hcd->state variable is a disaster. It's not clearly owned by
either usbcore or the host controller drivers, and they both change it
from time to time, potentially stepping on each other's toes. It's
not protected by any locks. And there's no mechanism to prevent it
from going through an invalid transition.
This patch (as1451) takes a first step toward fixing these problems.
As it turns out, usbcore uses hcd->state for essentially only two
things: checking whether the controller's root hub is running and
checking whether the controller has died. Therefore the patch adds
two new atomic bitflags to the hcd structure, to store these pieces of
information. The new flags are used only by usbcore, and a private
spinlock prevents invalid combinations (a dead controller's root hub
cannot be running).
The patch does not change the places where usbcore sets hcd->state,
since HCDs may depend on them. Furthermore, there is one place in
usb_hcd_irq() where usbcore still must use hcd->state: An HCD's
interrupt handler can implicitly indicate that the controller died by
setting hcd->state to HC_STATE_HALT. Nevertheless, the new code is a
big improvement over the current code.
The patch makes one other change. The hcd_bus_suspend() and
hcd_bus_resume() routines now check first whether the host controller
has died; if it has then they return immediately without calling the
HCD's bus_suspend or bus_resume methods.
This fixes the major problem reported in Bugzilla #29902: The system
fails to suspend after a host controller dies during system resume.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Alex Terekhov <a.terekhov@gmail.com>
CC: <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is needed to resolve some merge conflicts that were found
in the USB host controller patches, and reported by Stephen Rothwell.
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Provide optional hooks for the host controller driver to override the
default DMA mapping and unmapping routines. In general, these shouldn't
be necessary unless the host controller has special DMA requirements,
such as alignment contraints. If these are not specified, the
general usb_hcd_(un)?map_urb_for_dma functions will be used instead.
Also, pass the status to unmap_urb_for_dma so it can know whether the
DMA buffer has been overwritten.
Finally, add a flag to be used by these implementations if they
allocated a temporary buffer so it can be freed properly when unmapping.
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The convention is to prefix symbols exported from the USB HCD core with
"usb_hcd". This change makes unmap_urb_setup_for_dma() and
unmap_urb_for_dma() consistent with that.
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The USB core keeps track of pending resume requests for root hubs, in
order to resolve races between wakeup requests and suspends. However
the code that does this is subject to another race (between wakeup
requests and resumes) because the WAKEUP_PENDING flag is cleared
before the resume occurs, leaving a window in which another wakeup
request might arrive.
This patch (as1447) fixes the problem by clearing the WAKEUP_PENDING
flag after the resume instead of before it.
This fixes Bugzilla #24952.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Tested-by: Paul Bender <pebender@san.rr.com>
Tested-by: warpme <warpme@o2.pl>
Cc: stable <stable@kernel.org> [.36+]
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* usb-next: (132 commits)
USB: uas: Use GFP_NOIO instead of GFP_KERNEL in I/O submission path
USB: uas: Ensure we only bind to a UAS interface
USB: uas: Rename sense pipe and sense urb to status pipe and status urb
USB: uas: Use kzalloc instead of kmalloc
USB: uas: Fix up the Sense IU
usb: musb: core: kill unneeded #include's
DA8xx: assign name to MUSB IRQ resource
usb: gadget: g_ncm added
usb: gadget: f_ncm.c added
usb: gadget: u_ether: prepare for NCM
usb: pch_udc: Fix setup transfers with data out
usb: pch_udc: Fix compile error, warnings and checkpatch warnings
usb: add ab8500 usb transceiver driver
USB: gadget: Implement runtime PM for MSM bus glue driver
USB: gadget: Implement runtime PM for ci13xxx gadget
USB: gadget: Add USB controller driver for MSM SoC
USB: gadget: Introduce ci13xxx_udc_driver struct
USB: gadget: Initialize ci13xxx gadget device's coherent DMA mask
USB: gadget: Fix "scheduling while atomic" bugs in ci13xxx_udc
USB: gadget: Separate out PCI bus code from ci13xxx_udc
...
The Inventra DMA engine used with the MUSB controller in many
SoCs cannot use DMA for control transfers on EP0, but can use
DMA for all other transfers.
The USB core maps urbs for DMA if hcd->self.uses_dma is true.
(hcd->self.uses_dma is true for MUSB as well).
Split the uses_dma flag into two - one that says if the
controller needs to use PIO for control transfers, and
another which says if the controller uses DMA (for all
other transfers).
Also, populate this flag for all MUSB by default.
(Tested on OMAP3 and OMAP4 boards, with EHCI and MUSB HCDs
simultaneously in use).
Signed-off-by: Maulik Mankad <x0082077@ti.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Anand Gadiyar <gadiyar@ti.com>
Cc: Oliver Neukum <oliver@neukum.org>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Praveena NADAHALLY <praveen.nadahally@stericsson.com>
Cc: Ajay Kumar Gupta <ajay.gupta@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Since the runtime-PM core already defines a .last_busy field in
device.power, this patch uses it to replace the .last_busy field
defined in usb_device and uses pm_runtime_mark_last_busy to implement
usb_mark_last_busy.
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Reviewed-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Split unmap_urb_for_dma() to allow just the setup buffer
to be unmapped. This allows HCDs to use PIO for the setup
buffer if it is not suitable for DMA.
Signed-off-by: Martin Fuzzey <mfuzzey@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The USB stack maps the buffer for DMA if the controller supports DMA.
MUSB controller can perform DMA as well as PIO transfers.
The buffer needs to be unmapped before CPU can perform
PIO data transfers.
Export unmap_urb_for_dma() so that drivers can perform
the DMA unmapping in a sane way.
Signed-off-by: Maulik Mankad <x0082077@ti.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Enable MSI/MSI-X supporting in xhci driver.
Provide the mechanism to fall back using MSI and Legacy IRQs
if MSI-X IRQs register failed.
Signed-off-by: Dong Nguyen <Dong.Nguyen@amd.com>
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>,
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If we use the HCD_LOCAL_MEM flag and dma_declare_coherent_memory() to
enforce the host controller's local memory utilization we also need to
disable native scatter-gather support, otherwise hcd_alloc_coherent() in
map_urb_for_dma() is called with urb->transfer_buffer == NULL, that
triggers a NULL pointer dereference.
We can also consider to add a WARN_ON() and return an error code to
better catch this problem in the future.
At the moment no driver seems to hit this bug, so I should
consider this a low-priority fix.
Signed-off-by: Andrea Righi <arighi@develer.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1395) adds code to hcd_pci_suspend() for handling wakeup
races. This is another general race pattern, similar to the "open
vs. unregister" race we're all familiar with. Here, the race is
between suspending a device and receiving a wakeup request from one of
the device's suspended children.
In particular, if a root-hub wakeup is requested at about the same
time as the corresponding USB controller is suspended, and if the
controller is enabled for wakeup, then the controller should either
fail to suspend or else wake right back up again.
During system sleep this won't happen very much, especially since host
controllers generally aren't enabled for wakeup during sleep. However
it is definitely an issue for runtime PM. Something like this will be
needed to prevent the controller from autosuspending while waiting for
a root-hub resume to take place. (That is, in fact, the common case,
for which there is an extra test.)
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1393) converts several of the single-bit fields in
struct usb_hcd to atomic flags. This is for safety's sake; not all
CPUs can update bitfield values atomically, and these flags are used
in multiple contexts.
The flag fields that are set only during registration or removal can
remain as they are, since non-atomic accesses at those times will not
cause any problems.
(Strictly speaking, the authorized_default flag should become atomic
as well. I didn't bother with it because it gets changed only via
sysfs. It can be done later, if anyone wants.)
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1390) fixes a problem that crops up when a UHCI host
controller is unbound from uhci-hcd while there are still some active
URBs. The URBs have to be unlinked when the root hub is unregistered,
and uhci-hcd relies upon root-hub status polls as part of its
unlinking procedure. But usb_hcd_poll_rh_status() won't make those
status calls if hcd->rh_registered is clear, and the flag is cleared
_before_ the unregistration takes place.
Since hcd->rh_registered is used for other things and needs to be
cleared early, the solution is to add a new flag (rh_pollable) and use
it instead. It gets cleared _after_ the root hub is unregistered.
Now that the status polls don't end too soon, we have to make sure
they also don't occur too late -- after the root hub's usb_device
structure or the HCD's private structures are deallocated. Therefore
the patch adds usb_get_device() and usb_put_device() calls to protect
the root hub structure, and it adds an extra del_timer_sync() to
prevent the root-hub timer from causing an unexpected status poll.
This additional complexity would not be needed if the HCD framework
had provided separate stop() and release() callbacks instead of just
stop(). This lack could be fixed at some future time (although it
would require changes to every host controller driver); when that
happens this patch won't be needed any more.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1389) fixes some errors in the failure pathway of
usb_add_hcd(). The actions it takes ought to be exactly the same as
those taken by usb_remove_hcd(), but they aren't.
In one case (removal of the usb_bus_attr_group), the two routines are
brought into agreement by changing usb_remove_hcd(). All the other
discrepancies are fixed by changing usb_add_hcd().
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Obviously, {} is needed in the branch of
"else if (hcd->driver->flags & HCD_LOCAL_MEM)"
for handling of setup packet mapping.
Signed-off-by: Ming Lei <tom.leiming@gmail.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Cc: stable <stable@kernel.org>
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>
Now that URB_NO_SETUP_DMA_MAP is no longer in use, this patch (as1376)
removes all references to it.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Bulk endpoint streams were added in the USB 3.0 specification. Streams
allow a device driver to overload a bulk endpoint so that multiple
transfers can be queued at once.
The device then decides which transfer it wants to work on first, and can
queue part of a transfer before it switches to a new stream. All this
switching is invisible to the device driver, which just gets a completion
for the URB. Drivers that use streams must be able to handle URBs
completing in a different order than they were submitted to the endpoint.
This requires adding new API to set up xHCI data structures to support
multiple queues ("stream rings") per endpoint. Drivers will allocate a
number of stream IDs before enqueueing URBs to the bulk endpoints of the
device, and free the stream IDs in their disconnect function. See
Documentation/usb/bulk-streams.txt for details.
The new mass storage device class, USB Attached SCSI Protocol (UASP), uses
these streams API.
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 usbcore headers: hcd.h and hub.h are shared between usbcore,
HCDs and a couple of other drivers (e.g. USBIP modules).
So, it makes sense to move them into a more public location and
to cleanup dependency of those modules on kernel internal headers.
This patch moves hcd.h from drivers/usb/core into include/linux/usb/
Signed-of-by: Eric Lescouet <eric@lescouet.org>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1346) changes the idProduct value for USB-3.0 root hubs
from 0x0002 (which we already use for USB-2.0 root hubs) to 0x0003.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
CC: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1329) converts the USB stack over to the PM core's
runtime PM framework. This involves numerous changes throughout
usbcore, especially to hub.c and driver.c. Perhaps the most notable
change is that CONFIG_USB_SUSPEND now depends on CONFIG_PM_RUNTIME
instead of CONFIG_PM.
Several fields in the usb_device and usb_interface structures are no
longer needed. Some code which used to depend on CONFIG_USB_PM now
depends on CONFIG_USB_SUSPEND (requiring some rearrangement of header
files).
The only visible change in behavior should be that following a system
sleep (resume from RAM or resume from hibernation), autosuspended USB
devices will be resumed just like everything else. They won't remain
suspended. But if they aren't in use then they will naturally
autosuspend again in a few seconds.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch (as1324) makes a small change to the code used for remote
wakeup of root hubs. hcd_resume_work() now calls the hub driver's
remote-wakeup routine instead of implementing its own version.
The patch is complicated by the need to rename remote_wakeup() to
usb_remote_wakeup(), make it non-static, and declare it in a header
file. There's also the additional complication required to make
everything work when CONFIG_PM isn't set; the do-nothing inline
routine had to be moved into the header file.
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
USB devices do not have to sort interfaces in their descriptors based on
the interface number, and they may choose to skip interface numbers. The
USB bandwidth allocation code for installing a new configuration assumes
the for loop variable will match the interface number. Make it use the
interface number (bInterfaceNumber) in the descriptor instead.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Borislav Petkov reports issues with duplicate sysfs endpoint files after a
resume from a hibernate. It turns out that the code to support alternate
settings under xHCI has issues when a device with a non-default alternate
setting is reset during the hibernate:
[ 427.681810] Restarting tasks ...
[ 427.681995] hub 1-0:1.0: state 7 ports 6 chg 0004 evt 0000
[ 427.682019] usb usb3: usb resume
[ 427.682030] ohci_hcd 0000:00:12.0: wakeup root hub
[ 427.682191] hub 1-0:1.0: port 2, status 0501, change 0000, 480 Mb/s
[ 427.682205] usb 1-2: usb wakeup-resume
[ 427.682226] usb 1-2: finish reset-resume
[ 427.682886] done.
[ 427.734658] ehci_hcd 0000:00:12.2: port 2 high speed
[ 427.734663] ehci_hcd 0000:00:12.2: GetStatus port 2 status 001005 POWER sig=se0 PE CONNECT
[ 427.746682] hub 3-0:1.0: hub_reset_resume
[ 427.746693] hub 3-0:1.0: trying to enable port power on non-switchable hub
[ 427.786715] usb 1-2: reset high speed USB device using ehci_hcd and address 2
[ 427.839653] ehci_hcd 0000:00:12.2: port 2 high speed
[ 427.839666] ehci_hcd 0000:00:12.2: GetStatus port 2 status 001005 POWER sig=se0 PE CONNECT
[ 427.847717] ohci_hcd 0000:00:12.0: GetStatus roothub.portstatus [1] = 0x00010100 CSC PPS
[ 427.915497] hub 1-2:1.0: remove_intf_ep_devs: if: ffff88022f9e8800 ->ep_devs_created: 1
[ 427.915774] hub 1-2:1.0: remove_intf_ep_devs: bNumEndpoints: 1
[ 427.915934] hub 1-2:1.0: if: ffff88022f9e8800: endpoint devs removed.
[ 427.916158] hub 1-2:1.0: create_intf_ep_devs: if: ffff88022f9e8800 ->ep_devs_created: 0, ->unregistering: 0
[ 427.916434] hub 1-2:1.0: create_intf_ep_devs: bNumEndpoints: 1
[ 427.916609] ep_81: create, parent hub
[ 427.916632] ------------[ cut here ]------------
[ 427.916644] WARNING: at fs/sysfs/dir.c:477 sysfs_add_one+0x82/0x96()
[ 427.916649] Hardware name: System Product Name
[ 427.916653] sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:12.2/usb1/1-2/1-2:1.0/ep_81'
[ 427.916658] Modules linked in: binfmt_misc kvm_amd kvm powernow_k8 cpufreq_ondemand cpufreq_powersave cpufreq_userspace freq_table cpufreq_conservative ipv6 vfat fat
+8250_pnp 8250 pcspkr ohci_hcd serial_core k10temp edac_core
[ 427.916694] Pid: 278, comm: khubd Not tainted 2.6.33-rc2-00187-g08d869a-dirty #13
[ 427.916699] Call Trace:
The problem is caused by a mismatch between the USB core's view of the
device state and the USB device and xHCI host's view of the device state.
After the device reset and re-configuration, the device and the xHCI host
think they are using alternate setting 0 of all interfaces. However, the
USB core keeps track of the old state, which may include non-zero
alternate settings. It uses intf->cur_altsetting to keep the endpoint
sysfs files for the old state across the reset.
The bandwidth allocation functions need to know what the xHCI host thinks
the current alternate settings are, so original patch set
intf->cur_altsetting to the alternate setting 0. This caused duplicate
endpoint files to be created.
The solution is to not set intf->cur_altsetting before calling
usb_set_interface() in usb_reset_and_verify_device(). Instead, we add a
new flag to struct usb_interface to tell usb_hcd_alloc_bandwidth() to use
alternate setting 0 as the currently installed alternate setting.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: Borislav Petkov <petkovbb@googlemail.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix new kernel-doc warnings in usb core:
Warning(drivers/usb/core/usb.c:79): No description found for parameter 'config'
Warning(drivers/usb/core/usb.c:79): No description found for parameter 'iface_num'
Warning(drivers/usb/core/usb.c:79): No description found for parameter 'alt_num'
Warning(drivers/usb/core/hcd.c:1622): No description found for parameter 'udev'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make the USB core check the bandwidth when switching from one
interface alternate setting to another. Also check the bandwidth
when resetting a configuration (so that alt setting 0 is used). If
this check fails, the device's state is unchanged. If the device
refuses the new alt setting, re-instate the old alt setting in the
host controller hardware.
If a USB device doesn't have an alternate interface setting 0, install
the first alt setting in its descriptors when a new configuration is
requested, or the device is reset.
Add a mutex per root hub to protect bandwidth operations:
adding/reseting/changing configurations, and changing alternate interface
settings. We want to ensure that the xHCI host controller and the USB
device are set up for the same configurations and alternate settings.
There are two (possibly three) steps to do this:
1. The host controller needs to check that bandwidth is available for a
different setting, by issuing and waiting for a configure endpoint
command.
2. Once that returns successfully, a control message is sent to the
device.
3. If that fails, the host controller must be notified through another
configure endpoint command.
The mutex is used to make these three operations seem atomic, to prevent
another driver from using more bandwidth for a different device while
we're in the middle of these operations.
While we're touching the bandwidth code, rename usb_hcd_check_bandwidth()
to usb_hcd_alloc_bandwidth(). This function does more than just check
that the bandwidth change won't exceed the bus bandwidth; it actually
changes the bandwidth configuration in the xHCI host controller.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Refactor out the code to find alternate interface settings into
usb_find_alt_setting(). Print a debugging message and return null if the
alt setting is not found.
While we're at it, correct a bug in the refactored code. The interfaces
in the configuration's interface cache are not necessarily in numerical
order, so we can't just use the interface number as an array index. Loop
through the interface caches, looking for the correct interface.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Quiet the following sparse noise:
warning: Using plain integer as NULL pointer
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In map_urb_for_dma(), the DMA address returned by dma_map_single()
is not checked to determine if it is legal. This lack of checking
contributed to a problem with the libertas wireless driver
(http://marc.info/?l=linux-wireless&m=125695331205062&w=2). The
difficulty was not detected until the buffer was unmapped. By this time
memory corruption had occurred.
The situation is fixed by testing the returned DMA address, and
returning -EAGAIN if the address is invalid.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The previous code had a bug that would add a trailing null byte to
the returned descriptor.
Signed-off-by: George Spelvin <linux@horizon.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Commit 5fd29d6ccb ("printk: clean up
handling of log-levels and newlines") changed printk semantics. printk
lines with multiple KERN_<level> prefixes are no longer emitted as
before the patch.
<level> is now included in the output on each additional use.
Remove all uses of multiple KERN_<level>s in formats.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Replace if-elseif-else with switch-case
to keep the code consistent which is semantically same
Switch-case is used here,
http://www.spinics.net/lists/linux-usb/msg17201.html
Making consistent at other places in usb/core
Also easier to read and maintain when USB4.0, 5.0, ... comes
Signed-off-by: Viral Mehta <viral.mehta@einfochips.com>
Tested-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is the original patch I created before David Vrabel posted a better
patch (http://marc.info/?l=linux-usb&m=123377477209109&w=2) that does
basically the same thing. This patch will get replaced with his
(modified) patch later.
Allow USB device drivers that use usb_sg_init() and usb_sg_wait() to push
bulk endpoint scatter gather lists down to the host controller drivers.
This allows host controller drivers to more efficiently enqueue these
transfers, and allows the xHCI host controller to better take advantage of
USB 3.0 "bursts" for bulk endpoints.
This patch currently only enables scatter gather lists for bulk endpoints.
Other endpoint types that use the usb_sg_* functions will not have their
scatter gather lists pushed down to the host controller. For periodic
endpoints, we want each scatterlist entry to be a separate transfer.
Eventually, HCDs could parse these scatter-gather lists for periodic
endpoints also. For now, we use the old code and call usb_submit_urb()
for each scatterlist entry.
The caller of usb_sg_init() can request that all bytes in the scatter
gather list be transferred by passing in a length of zero. Handle that
request for a bulk endpoint under xHCI by walking the scatter gather list
and calculating the length. We could let the HCD handle a zero length in
this case, but I'm not sure if the core layers in between will get
confused by this.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Originally, the USB core had no support for allocating bandwidth when a
particular configuration or alternate setting for an interface was
selected. Instead, the device driver's URB submission would fail if
there was not enough bandwidth for a periodic endpoint. Drivers could
work around this, by using the scatter-gather list API to guarantee
bandwidth.
This patch adds host controller API to allow the USB core to allocate or
deallocate bandwidth for an endpoint. Endpoints are added to or dropped
from a copy of the current schedule by calling add_endpoint() or
drop_endpoint(), and then the schedule is atomically evaluated with a
call to check_bandwidth(). This allows all the endpoints for a new
configuration or alternate setting to be added at the same time that the
endpoints from the old configuration or alt setting are dropped.
Endpoints must be added to the schedule before any URBs are submitted to
them. The HCD must be allowed to reject a new configuration or alt
setting before the control transfer is sent to the device requesting the
change. It may reject the change because there is not enough bandwidth,
not enough internal resources (such as memory on an embedded host
controller), or perhaps even for security reasons in a virtualized
environment.
If the call to check_bandwidth() fails, the USB core must call
reset_bandwidth(). This causes the schedule to be reverted back to the
state it was in just after the last successful check_bandwidth() call.
If the call succeeds, the host controller driver (and hardware) will have
changed its internal state to match the new configuration or alternate
setting. The USB core can then issue a control transfer to the device to
change the configuration or alt setting. This allows the core to test new
configurations or alternate settings before unbinding drivers bound to
interfaces in the old configuration.
WIP:
The USB core must add endpoints from all interfaces in a configuration
to the schedule, because a driver may claim that interface at any time.
A slight optimization might be to add the endpoints to the schedule once
a driver claims that interface. FIXME
This patch does not cover changing alternate settings, but it does
handle a configuration change or de-configuration. FIXME
The code for managing the schedule is currently HCD specific. A generic
scheduling algorithm could be added for host controllers without
built-in scheduling support. For now, if a host controller does not
define the check_bandwidth() function, the call to
usb_hcd_check_bandwidth() will always succeed.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add USB 3.0 root hub descriptors. This is a kludge because I reused the old
USB 2.0 descriptors, instead of using the new USB 3.0 hub descriptors with
endpoint companion descriptors and other descriptors. I did this because I
wasn't ready to add USB 3.0 hub changes to khubd. For now, a USB 3.0 roothub
looks like a USB 2.0 roothub, with a higher speed.
USB 3.0 hubs have no transaction translator (TT).
Make USB core debugging handle super speed ports.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Modify the USB core to handle the new USB 3.0 speed, "SuperSpeed". This
is 5.0 Gbps (wire speed). There are probably more places that check for
speed that I've missed.
SuperSpeed devices have a 512 byte endpoint 0 max packet size. This shows
up as a bMaxPacketSize0 set to 0x09 (see table 9-8 of the USB 3.0 bus
spec).
xHCI spec says that the xHC can handle intervals up to 2^15 microframes. That
might change when real silicon becomes available.
Add FIXME note for SuperSpeed isochronous endpoints. They can transmit up
to 16 packets in one "burst" before they wait for an acknowledgment of the
packets. They can do up to 3 bursts per microframe (determined by the
mult value in the endpoint companion descriptor). The xHCI driver doesn't
have support for isoc yet, so fix this later.
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
