The xhci driver will OOPS on resume from S2/S3 if dma_alloc_coherent()
is out of memory. This is a result of two things:
1. xhci_mem_cleanup() in xhci-mem.c free's xhci->lpm_command if
it's not NULL, but doesn't set it to NULL after the free.
2. xhci_mem_cleanup() is called twice on resume, once for normal
restart and once from xhci_mem_init() if dma_alloc_coherent() fails,
resulting in a free of xhci->lpm_command that has already been freed.
The fix is to set xhci->lpm_command to NULL after freeing it.
Signed-off-by: NAl Cooper <alcooperx@gmail.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

If xhci initialization fails before the roothub bandwidth
domains (xhci->rh_bw[i]) are allocated it will oops when
trying to access rh_bw members in xhci_mem_cleanup().
Reported-by: NManuel Reimer <manuel.reimer@gmx.de>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Lists of endpoints are stored for bandwidth calculation for roothub ports.
Make sure we remove all endpoints from the list before the whole device,
containing its endpoints list_head stuctures, is freed.

This used to be done in the wrong order in xhci_mem_cleanup(),
and triggered an oops in resume from S4 (hibernate).

Cc: stable <stable@vger.kernel.org>
Tested-by: NVille Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Use one timer to control command timeout.

start/kick the timer every time a command is completed and a
new command is waiting, or a new command is added to a empty list.

If the timer runs out, then tag the current command as "aborted", and
start the xhci command abortion process.

Previously each function that submitted a command had its own timer.
If that command timed out, a new command structure for the
command was created and it was put on a cancel_cmd_list list,
then a pci write to abort the command ring was issued.

when the ring was aborted, it checked if the current command
was the one to be canceled, later when the ring was stopped the
driver got ownership of the TRBs in the command ring,
compared then to the TRBs in the cancel_cmd_list,
and turned them into No-ops.

Now, instead, at timeout we tag the status of the command in the
command queue to be aborted, and start the ring abortion.
Ring abortion stops the command ring and gives control of the
commands to us.
All the aborted commands are now turned into No-ops.

If the ring is already stopped when the command times outs its not possible
to start the ring abortion, in this case the command is turnd to No-op
right away.

All these changes allows us to remove the entire cancel_cmd_list code.

The functions waiting for a command to finish no longer have their own timeouts.
They will wait either until the command completes normally,
or until the whole command abortion is done.
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Remove the per-device command list and handle_cmd_in_cmd_wait_list()
and use the completion and status variables found in the
command structure in the global command list.
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Create a list to store command structures, add a structure to it every time
a command is submitted, and remove it from the list once we get a
command completion event matching the command.

Callers that wait for completion will free their command structures themselves.
The other command structures are freed in the command completion event handler.

Also add a check that prevents queuing commands if host is dying
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

If we align segment dma pool memory to 64 bytes, then a segment can be located
at 0x10000040 - 0x1000043f, and a segment from another ring at 0x10000440 -
0x1000083f. The last trb in the first segment at 0x10000430 will then translate
to the same radix tree key as the first trb of the second segment, while they
are in different rings!

This patches fixes this by changing the alignment of the dma pool to be 1KB
rather then 64 bytes. An alternative fix would be to reduce the shift used
to calculate the radix tree keys, but that would (slighlty) grow the radix
trees so I believe this is the better fix.

Note this patch is mostly theoretical since in practice I've not seen
the dma_pool actually return not 1KB aligned memory.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

cmd_ring_reserved_trbs gets decremented by xhci_free_stream_info(), so set it
to 0 after freeing all rings, otherwise it wraps around to a very large value
when rings with streams are free-ed.

Before this patch the wrap-around could be triggered when xhci_resume
calls xhci_mem_cleanup if the controller resume fails.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

If we're expanding a stream ring, we want to make sure we can add those
ring segments to the radix tree that maps segments to ring pointers.
Try the radix tree insert after the new ring segments have been allocated
(the last segment in the new ring chunk will point to the first newly
allocated segment), but before the new ring segments are linked into the
old ring.

If insert fails on any one segment, remove each segment from the radix
tree, deallocate the new segments, and return.  Otherwise, link the new
segments into the tree.

HdG: Add a check to only update stream mappings in xhci_ring_expansion when
the ring is a stream ring.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

Before this a device needing ie 32 stream ctxs would end up with an entry from
the small_streams_pool which has 256 bytes entries, where as 32 stream ctxs
need 512 bytes. Things actually keep running for a surprisingly long time
before crashing because of this.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

xhci maintains a radix tree for each stream endpoint because it must
be able to map a trb address to the stream ring.  Each ring segment
must be added to the ring for this to work.  Currently xhci sticks
only the first segment of each stream ring into the radix tree.

Result is that things work initially, but as soon as the first segment
is full xhci can't map the trb address from the completion event to the
stream ring any more -> BOOM.  You'll find this message in the logs:

  ERROR Transfer event for disabled endpoint or incorrect stream ring

This patch adds a helper function to update the radix tree, and a
function to remove ring segments from the tree.  Both functions loop
over the segment list and handles all segments instead of just the
first.

[Note: Sarah changed this patch to add radix_tree_maybe_preload() and
radix_tree_preload_end() calls around the radix tree insert, since we
can now insert entries in interrupt context.  There are now two helper
functions to make the code cleaner, and those functions are moved to
make them static.]
Signed-off-by: NGerd Hoffmann <kraxel@redhat.com>
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This reverts commit e8b37332.  Many xHCI
host controllers can only handle 32-bit addresses, and writing 64-bits
at a time causes them to fail.  Reading 64-bits at a time may also cause
them to return 0xffffffff, so revert this commit as well.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This reverts commit 7dd09a1a.

Many xHCI host controllers can only handle 32-bit addresses, and writing
64-bits at a time causes them to fail.  Rafał reports that USB devices
simply do not enumerate, and reverting this patch helps.  Branimir
reports that his host controller doesn't respond to an Enable Slot
command and dies:

[   75.576160] xhci_hcd 0000:03:00.0: Timeout while waiting for a slot
[   88.991634] xhci_hcd 0000:03:00.0: Stopped the command ring failed, maybe the host is dead
[   88.991748] xhci_hcd 0000:03:00.0: Abort command ring failed
[   88.991845] xhci_hcd 0000:03:00.0: HC died; cleaning up
[   93.985489] xhci_hcd 0000:03:00.0: Timeout while waiting for a slot
[   93.985494] xhci_hcd 0000:03:00.0: Abort the command ring, but the xHCI is dead.
[   98.982586] xhci_hcd 0000:03:00.0: Timeout while waiting for a slot
[   98.982591] xhci_hcd 0000:03:00.0: Abort the command ring, but the xHCI is dead.
[  103.979696] xhci_hcd 0000:03:00.0: Timeout while waiting for a slot
[  103.979702] xhci_hcd 0000:03:00.0: Abort the command ring, but the xHCI is dead
Signed-off-by: NSarah Sharp <sarah.a.sharp@intel.com>
Reported-by: NRafał Miłecki <zajec5@gmail.com>
Reported-by: NBranimir Maksimovic <branimir.maksimovic@gmail.com>
Cc: Xenia Ragiadakou <burzalodowa@gmail.com>

Function xhci_write_64() is used to write 64bit xHC registers residing in MMIO.
On 32bit systems, xHC registers need to be written with 32bit accesses by
writing first the lower 32bits and then the higher 32bits. The header file
asm-generic/io-64-nonatomic-lo-hi.h ensures that on 32bit systems writeq() will
will write 64bit registers in 32bit chunks with low-high order.

Replace all calls to xhci_write_64() with calls to writeq().

This is done to reduce code duplication since 64bit low-high write logic
is already implemented and to take advantage of inherent "atomic" 64bit
write operations on 64bit systems.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

Function xhci_read_64() is used to read 64bit xHC registers residing in MMIO.
On 32bit systems, xHC registers need to be read with 32bit accesses by
reading first the lower 32bits and then the higher 32bits.

Replace all calls to xhci_read_64() with calls to readq() and include
asm-generic/io-64-nonatomic-lo-hi.h header file, so that if the system
is not 64bit, readq() will read registers in 32bit chunks with low-high order.

This is done to reduce code duplication since 64bit low-high read logic
is already implemented and to take advantage of inherent "atomic" 64bit
read operations on 64bit systems.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

Function xhci_writel() is used to write a 32bit value in xHC registers residing
in MMIO address space. It takes as first argument a pointer to the xhci_hcd
although it does not use it. xhci_writel() internally simply calls writel().
This creates an illusion that xhci_writel() is an xhci specific function that
has to be called in a context where a pointer to xhci_hcd is available.

Remove xhci_writel() wrapper function and replace its calls with calls to
writel() to make the code more straight-forward.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

Function xhci_readl() is used to read 32bit xHC registers residing in MMIO
address space. It takes as first argument a pointer to the xhci_hcd although
it does not use it. xhci_readl() internally simply calls readl(). This creates
an illusion that xhci_readl() is an xhci specific function that has to be
called in a context where a pointer to xhci_hcd is available.

Remove the unnecessary xhci_readl() wrapper function and replace its calls to
with calls to readl() to make the code more straightforward.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This patch removes the to_pci_dev() conversion performed to generic struct
device since it is not actually useful (the pointer to the generic device
can be used directly rather through a conversion to pci_dev) and it is pci
bus specific.

This isn't stable material because this code will produce harmless
behavior on non-PCI xHCI hosts.  The pci_device pointer is never
dereferenced, only used to re-calculate the underlying device pointer.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This patch removes the unneccessary check 'if (stream_info)' because
there is already a check few lines above which ensures that stream_info
is not NULL.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This patch converts TRB_CYCLE to le32 to update correctly the Cycle Bit in
'control' field of the link TRB.
This bug was found using sparse.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

How it's supposed to work:
--------------------------

USB 2.0 Link PM is a lower power state that some newer USB 2.0 devices
support.  USB 3.0 devices certified by the USB-IF are required to
support it if they are plugged into a USB 2.0 only port, or a USB 2.0
cable is used.  USB 2.0 Link PM requires both a USB device and a host
controller that supports USB 2.0 hardware-enabled LPM.

USB 2.0 Link PM is designed to be enabled once by software, and the host
hardware handles transitions to the L1 state automatically.  The premise
of USB 2.0 Link PM is to be able to put the device into a lower power
link state when the bus is idle or the device NAKs USB IN transfers for
a specified amount of time.

...but hardware is broken:
--------------------------

It turns out many USB 3.0 devices claim to support USB 2.0 Link PM (by
setting the LPM bit in their USB 2.0 BOS descriptor), but they don't
actually implement it correctly.  This manifests as the USB device
refusing to respond to transfers when it is plugged into a USB 2.0 only
port under the Haswell-ULT/Lynx Point LP xHCI host.

These devices pass the xHCI driver's simple test to enable USB 2.0 Link
PM, wait for the port to enter L1, and then bring it back into L0.  They
only start to break when L1 entry is interleaved with transfers.

Some devices then fail to respond to the next control transfer (usually
a Set Configuration).  This results in devices never enumerating.

Other mass storage devices (such as a later model Western Digital My
Passport USB 3.0 hard drive) respond fine to going into L1 between
control transfers.  They ACK the entry, come out of L1 when the host
needs to send a control transfer, and respond properly to those control
transfers.  However, when the first READ10 SCSI command is sent, the
device NAKs the data phase while it's reading from the spinning disk.
Eventually, the host requests to put the link into L1, and the device
ACKs that request.  Then it never responds to the data phase of the
READ10 command.  This results in not being able to read from the drive.

Some mass storage devices (like the Corsair Survivor USB 3.0 flash
drive) are well behaved.  They ACK the entry into L1 during control
transfers, and when SCSI commands start coming in, they NAK the requests
to go into L1, because they need to be at full power.

Not all USB 3.0 devices advertise USB 2.0 link PM support.  My Point
Grey USB 3.0 webcam advertises itself as a USB 2.1 device, but doesn't
have a USB 2.0 BOS descriptor, so we don't enable USB 2.0 Link PM.  I
suspect that means the device isn't certified.

What do we do about it?
-----------------------

There's really no good way for the kernel to test these devices.
Therefore, the kernel needs to disable USB 2.0 Link PM by default, and
distros will have to enable it by writing 1 to the sysfs file
/sys/bus/usb/devices/../power/usb2_hardware_lpm.  Rip out the xHCI Link
PM test, since it's not sufficient to detect these buggy devices, and
don't automatically enable LPM after the device is addressed.

This patch should be backported to kernels as old as 3.11, that
contain the commit a558ccdc "usb: xhci:
add USB2 Link power management BESL support".  Without this fix, some
USB 3.0 devices will not enumerate or work properly under USB 2.0 ports
on Haswell-ULT systems.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

When a device signals remote wakeup on a roothub, and the suspend change
bit is set, the host controller driver must not give control back to the
USB core until the port goes back into the active state.

EHCI accomplishes this by waiting in the get port status function until
the PORT_RESUME bit is cleared:

                        /* stop resume signaling */
                        temp &= ~(PORT_RWC_BITS | PORT_SUSPEND | PORT_RESUME);
                        ehci_writel(ehci, temp, status_reg);
                        clear_bit(wIndex, &ehci->resuming_ports);
                        retval = ehci_handshake(ehci, status_reg,
                                        PORT_RESUME, 0, 2000 /* 2msec */);

Similarly, the xHCI host should wait until the port goes into U0, before
passing control up to the USB core.  When the port transitions from the
RExit state to U0, the xHCI driver will get a port status change event.
We need to wait for that event before passing control up to the USB
core.

After the port transitions to the active state, the USB core should time
a recovery interval before it talks to the device.  The length of that
recovery interval is TRSMRCY, 10 ms, mentioned in the USB 2.0 spec,
section 7.1.7.7.  The previous xHCI code (which did not wait for the
port to go into U0) would cause the USB core to violate that recovery
interval.

This bug caused numerous USB device disconnects on remote wakeup under
ChromeOS and a Lynx Point LP xHCI host that takes up to 20 ms to move
from RExit to U0.  ChromeOS is very aggressive about power savings, and
sets the autosuspend_delay to 100 ms, and disables USB persist.

I attempted to replicate this bug with Ubuntu 12.04, but could not.  I
used Ubuntu 12.04 on the same platform, with the same BIOS that the bug
was triggered on ChromeOS with.  I also changed the USB sysfs settings
as described above, but still could not reproduce the bug under Ubuntu.
It may be that ChromeOS userspace triggers this bug through additional
settings.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This patch defines a new trace event, which is called xhci_dbg_ring_expansion
and belongs to the event class xhci_log_msg, and adds tracepoints that trace
the debug messages associated with the expansion of endpoint ring when there
is not enough space allocated to hold all pending TRBs.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This patch defines a new trace event, which is called xhci_dbg_init
and belongs to the event class xhci_log_msg, and adds tracepoints that
trace the debug statements in the functions used to start and stop the
xhci-hcd driver.

Also, it removes an unnecessary cast of variable val to unsigned int
in xhci_mem_init(), since val is already declared as unsigned int.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

This patch defines a new trace event, which is called xhci_dbg_context_change
and belongs in the event class xhci_log_msg, and adds tracepoints for tracing
the debug messages related to context updates performed with Configure Endpoint
and Evaluate Context commands.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

CONFIG_USB_XHCI_HCD_DEBUGGING option is used to enable
verbose debugging output for the xHCI host controller
driver.

In the current version of the xhci-hcd driver, this
option must be turned on, in order for the debugging
log messages to be displayed, and users may need to
recompile the linux kernel to obtain debugging
information that will help them track down problems.

This patch removes the above debug option to enable
debugging log messages at all times.
The aim of this is to rely on the debugfs and the
dynamic debugging feature for fine-grained management
of debugging messages and to not force users to set
the debug config option and compile the linux kernel
in order to have access in that information.

This patch, also, removes the XHCI_DEBUG symbol and the
functions dma_to_stream_ring(), xhci_test_radix_tree()
and xhci_event_ring_work() that are not useful anymore.
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

A randconfig build hit the following build errors because xhci.c and
xhci-mem.c use dma mapping functions but don't include
<linux/dma-mapping.h>. Add the missing includes to fix the build errors.

drivers/usb/host/xhci.c In function 'xhci_gen_setup':
drivers/usb/host/xhci.c +4872 : error: implicit declaration of function 'dma_set_mask'
drivers/usb/host/xhci.c +4872 : error: implicit declaration of function 'DMA_BIT_MASK'

drivers/usb/host/xhci-mem.c In function 'xhci_free_stream_ctx':
drivers/usb/host/xhci-mem.c +435 : error: implicit declaration of function 'dma_free_coherent'
drivers/usb/host/xhci-mem.c In function 'xhci_alloc_stream_ctx':
drivers/usb/host/xhci-mem.c +463 : error: implicit declaration of function 'dma_alloc_coherent'
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-usb@vger.kernel.org
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

Fail and free the container context in case dma_pool_alloc() can't allocate
the raw context data part of it

This patch should be backported to kernels as old as 2.6.31, that
contain the commit d115b048 "USB: xhci:
Support for 64-byte contexts".
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: John Youn <johnyoun@synopsys.com>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

If the endpoint type is unknown, set it to 0 and fail gracefully
instead of causing a kernel panic.
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

We may have more speed types in the future, so fail gracefully, rather
than causing the kernel to panic.

BUG() was called if the device speed was unknown when setting max packet
size.  Set the max packet size at the same time as the slot speed and
get rid of one switch statement with BUG() option completely.

[Note: Sarah merged a patch that she wrote that touched the
xhci_setup_addressable_virt_dev function with this patch from Mathias
for clarity.]
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

Fail gracefully, instead of causing the kernel to panic, if the input
control context doesn't have the right type (XHCI_CTX_TYPE_INPUT).  Push
finding the pointer to the input control context up into functions that
can fail.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: John Youn <johnyoun@synopsys.com>

It's horrible coding style to panic the kernel when someone passes you
an argument value you didn't expect.  In the future, we may want to add
additional context types, so it's better to gracefully handle additional
context types instead of panicking.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: John Youn <johnyoun@synopsys.com>

Hardware link powermanagement in usb2 is a per-port capability.
Previously support for hw lpm was enabled for all ports if any usb2 port supported it.

Now instead cache the capability values and check them for each port individually
Signed-off-by: NMathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

If for whatever reason we fall into fail path in xhci_mem_init()
before bw table gets initialized we may access the uninitialized lists
in xhci_mem_cleanup().

Check for bw table before traversing lists in cleanup routine.

This patch should be backported to kernels as old as 3.2, that contain
the commit 839c817c "xhci: Store
information about roothubs and TTs."
Reported-by: NSergey Dyasly <dserrg@gmail.com>
Tested-by: NSergey Dyasly <dserrg@gmail.com>
Signed-off-by: NVladimir Murzin <murzin.v@gmail.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

It is possible that we fail on xhci_mem_init, just before doing
the INIT_LIST_HEAD, and calling xhci_mem_cleanup.

Problem is that, the list_for_each_entry_safe macro, assumes
list heads are initialized (not NULL), and dereferences their 'next'
pointer, causing a kernel panic if this is not yet initialized.

Let's protect from that by moving inits to the beginning.

This patch should be backported to kernels as old as 3.2, that
contain the commit 9574323c "xHCI: test
USB2 software LPM".
Signed-off-by: NSergio Aguirre <sergio.a.aguirre.rodriguez@intel.com>
Acked-by: NDavid Cohen <david.a.cohen@intel.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

This patch shortens the logic in xhci_endpoint_init() by moving common
calculations involving max_packet and max_burst outside the switch
statement, rather than repeating the same code in multiple
case-specific statements.  It also replaces two usages of max_packet
which were clearly intended to be max_burst all along.

More importantly, it compensates for a common bug in high-speed bulk
endpoint descriptors.  In many devices there is a bulk endpoint having
a wMaxPacketSize value smaller than 512, which is forbidden by the USB
spec.  Some xHCI controllers can't handle this and refuse to accept
the endpoint.  This patch changes the max_packet value to 512, which
allows the controller to use the endpoint properly.

In practice the bogus maxpacket size doesn't matter, because none of
the transfers sent via these endpoints are longer than the maxpacket
value anyway.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Reported-and-tested-by: N"Aurélien Leblond" <blablack@gmail.com>
CC: <stable@vger.kernel.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Rename SEGMENT_SIZE and SEGMENT_SHIFT as the former is used in a.out.h.
Signed-off-by: NDavid Howells <dhowells@redhat.com>
cc: Sarah Sharp <sarah.a.sharp@linux.intel.com>
cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
cc: linux-usb@vger.kernel.org
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

xhci driver divides the root hub into two logical hubs which work
respectively for usb 2.0 and usb 3.0 devices. They are independent
devices in the usb core. But in the ACPI table, it's one device node
and all usb2.0 and usb3.0 ports are under it. Binding usb port with
its acpi node needs the raw port number which is reflected in the xhci
extended capabilities table. This patch is to add find_raw_port_number
callback to struct hc_driver(), fill it with xhci_find_raw_port_number()
which will return raw port number and add a wrap usb_hcd_find_raw_port_number().

Otherwise, refactor xhci_find_real_port_number(). Using
xhci_find_raw_port_number() to get real index in the HW port status
registers instead of scanning through the xHCI roothub port array.
This can help to speed up.

All addresses in xhci->usb2_ports and xhci->usb3_ports array are
kown good ports and don't include following bad ports in the extended
capabilities talbe.
     (1) root port that doesn't have an entry
     (2) root port with unknown speed
     (3) root port that is listed twice and with different speeds.

So xhci_find_raw_port_number() will only return port num of good ones
and never touch bad ports above.
Signed-off-by: NLan Tianyu <tianyu.lan@intel.com>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>

A high speed control or bulk endpoint may have bInterval set to zero,
which means it does not NAK.  If bInterval is non-zero, it means the
endpoint NAKs at a rate of 2^(bInterval - 1).

The xHCI code to compute the NAK interval does not handle the special
case of zero properly.  The current code unconditionally subtracts one
from bInterval and uses it as an exponent.  This causes a very large
bInterval to be used, and warning messages like these will be printed:

usb 1-1: ep 0x1 - rounding interval to 32768 microframes, ep desc says 0 microframes

This may cause the xHCI host hardware to reject the Configure Endpoint
command, which means the HS device will be unusable under xHCI ports.

This patch should be backported to kernels as old as 2.6.31, that contain
commit dfa49c4a "USB: xhci - fix math in
xhci_get_endpoint_interval()".
Reported-by: NVincent Pelletier <plr.vincent@gmail.com>
Suggested-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org

xhci_alloc_segments_for_ring() builds a list of xhci_segments and links
the tail to head at the end (forming a ring). When it bails out for OOM
reasons half-way through, it tries to destroy its half-built list with
xhci_free_segments_for_ring(), even though it is not a ring yet. This
causes a null-pointer dereference upon hitting the last element.

Furthermore, one of its callers (xhci_ring_alloc()) mistakenly believes
the output parameters to be valid upon this kind of OOM failure, and
calls xhci_ring_free() on them. Since the (incomplete) list/ring should
already be destroyed in that case, this would lead to a use after free.

This patch fixes those issues by having xhci_alloc_segments_for_ring()
destroy its half-built, non-circular list manually and destroying the
invalid struct xhci_ring in xhci_ring_alloc() with a plain kfree().

This patch should be backported to kernels as old as 2.6.31, that
contains the commit 0ebbab37 "USB: xhci:
Ring allocation and initialization."

A separate patch will need to be developed for kernels older than 3.4,
since the ring allocation code was refactored in that kernel.
Signed-off-by: NJulius Werner <jwerner@chromium.org>
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Cc: stable@vger.kernel.org