linux-toradex.git/drivers/usb/usbip, branch v4.9.52

usb: usbip: set buffer pointers to NULL after free

2017-07-12T13:01:03+00:00

commit b3b51417d0af63fb9a06662dc292200aed9ea53f upstream.

The usbip stack dynamically allocates the transfer_buffer and
setup_packet of each urb that got generated by the tcp to usb stub code.
As these pointers are always used only once we will set them to NULL
after use. This is done likewise to the free_urb code in vudc_dev.c.
This patch fixes double kfree situations where the usbip remote side
added the URB_FREE_BUFFER.

Signed-off-by: Michael Grzeschik 
Acked-by: Shuah Khan 
Signed-off-by: Greg Kroah-Hartman

USB: usbip: fix nonconforming hub descriptor

2017-06-24T05:11:15+00:00

commit ec963b412a54aac8e527708ecad06a6988a86fb4 upstream.

Fix up the root-hub descriptor to accommodate the variable-length
DeviceRemovable and PortPwrCtrlMask fields, while marking all ports as
removable (and leaving the reserved bit zero unset).

Also add a build-time constraint on VHCI_HC_PORTS which must never be
greater than USB_MAXCHILDREN (but this was only enforced through a
KConfig constant).

This specifically fixes the descriptor layout whenever VHCI_HC_PORTS is
greater than seven (default is 8).

Fixes: 04679b3489e0 ("Staging: USB/IP: add client driver")
Cc: Takahiro Hirofuchi 
Cc: Valentina Manea 
Signed-off-by: Johan Hovold 
Acked-by: Shuah Khan 
Signed-off-by: Greg Kroah-Hartman

usbip: vudc: fix: Clear already_seen flag also for ep0

2017-01-06T09:40:11+00:00

commit 3e448e13a662fb20145916636127995cbf37eb83 upstream.

ep_list inside gadget structure doesn't contain ep0.
It is stored separately in ep0 field.

This causes an urb hang if gadget driver decides to
delay setup handling. On host side this is visible as
timeout error when setting configuration.

This bug can be reproduced using for example any gadget
with mass storage function.

Fixes: abdb29574322 ("usbip: vudc: Add vudc_transfer")
Signed-off-by: Krzysztof Opasiak 
Acked-by: Shuah Khan 
Signed-off-by: Greg Kroah-Hartman

usb: Kconfig: using select for USB_COMMON dependency

2016-09-27T10:20:17+00:00

According to (badf6d47f8a9 "usb: common: rework CONFIG_USB_COMMON logic")
we should select USB_COMMON at Kconfig when usb common stuffs are needed,
but some of Kconfig enties have not followed it, update them.

Cc: Felipe Balbi 
Cc: Heikki Krogerus 
Signed-off-by: Peter Chen 
Acked-by: Arnd Bergmann 
Signed-off-by: Greg Kroah-Hartman

usb: usbip: vudc: fix left shift overflow

2016-08-30T20:28:52+00:00

In v_recv_cmd_submit(), urb_p->urb->pipe has the type unsigned int
(which is 32-bit long on x86_64) but 11<<30 results in a 34-bit integer.
Therefore the 2 leading bits are truncated and

    urb_p->urb->pipe &= ~(11 << 30);

has the same meaning as

    urb_p->urb->pipe &= ~(3 << 30);

This second statement seems to be how the code was intended to be
written, as PIPE_ constants have values between 0 and 3.

The overflow has been detected with a clang warning:

    drivers/usb/usbip/vudc_rx.c:145:27: warning: signed shift result
    (0x2C0000000) requires 35 bits to represent, but 'int' only has 32
    bits [-Wshift-overflow]
            urb_p->urb->pipe &= ~(11 << 30);
                                  ~~ ^  ~~

Fixes: 79c02cb1fd5c ("usbip: vudc: Add vudc_rx")
Signed-off-by: Nicolas Iooss 
Signed-off-by: Greg Kroah-Hartman

usbip: vudc: Fix apparent cut-n-paste error

2016-08-30T17:06:43+00:00

Coverity picked up that this looks like a cut-n-paste from an almost
identical sequence below that didn't get its variable renamed.

Signed-off-by: Dave Jones 
Reviewed-by: Krzysztof Opasiak 
Signed-off-by: Greg Kroah-Hartman

usbip: vhci_hcd: fix return value check in add_platform_device()

2016-08-15T14:25:59+00:00

In case of error, the function platform_device_register_simple() returns
ERR_PTR() and never returns NULL. The NULL test in the return value
check should be replaced with IS_ERR().

Signed-off-by: Wei Yongjun 
Signed-off-by: Greg Kroah-Hartman

usb: usbip: stub_rx: don't print error when allocating urb fails

2016-08-15T13:54:27+00:00

kmalloc will print enough information in case of failure.

Signed-off-by: Wolfram Sang 
Signed-off-by: Greg Kroah-Hartman

usbip: vhci extension: modifications to vhci driver

2016-08-09T14:21:05+00:00

Modification to Kconfig, vhci_hc.c, vhci.h and vhci_sysfs.c.

1. kernel config

Followings are added.

USBIP_VHCI_HC_PORTS: Number of ports per USB/IP virtual host
controller. The default is 8 - same as current VHCI_NPORTS.
USBIP_VHCI_NR_HCS: Number of USB/IP virtual host controllers. The
default is 1. This paratmeter is replaced with USBIP_VHCI_INIT_HCS and
USBIP_VHCI_MAX_HCS included in succeeding dynamic extension patch.

2. the_controller to controllers

the_controller is changed to vhci_pdevs: array of struct
platform_device.

3. vhci_sysfs.c

Sysfs structure is changed as following.

BEFORE:
    /sys/devices/platform
        +-- vhci
            +-- status
            +-- attach
            +-- detach
            +-- usbip_debug

AFTER: example for CONFIG_USBIP_NR_HCS=4
    /sys/devices/platform
        +-- vhci
        |   +-- nports
        |   +-- status
        |   +-- status.1
        |   +-- status.2
        |   +-- status.3
        |   +-- attach
        |   +-- detach
        |   +-- usbip_debug
        +-- vhci.1
        +-- vhci.2
        +-- vhci.3

vhci[.N] is shown for each host controller kobj. vhch.1, vhci.2, ...
are shown only when CONFIG_USBIP_NR_HCS is more than 1. Only 'vhci'
(without number) has user space interfaces. 'nports' is newly added to
give ports-per-controller and number of controlles. Before that, number
of ports is acquired by reading status lines. Status is divided for
each controller to avoid page size (4KB) limitation.

Old userspace tool binaries work with the first status within the first
controller.

Inconsistency between status header and content is fixed.
4th and 5th column are
header:          "dev bus"
content(unused): "000 000"
content(used):   "%08x", devid
Only 1st and 2nd column are used by program. In old version, sscanf()
in parse_status expect no bus column. And bus_id string is shown in the
last column. Then bus in the header is removed and unused content is
replaced with 8 zeros. The sscanf() expects more than 5 columns and new
has 6 columns so there's no compatibility issue in this change.

Signed-off-by: Nobuo Iwata 
Signed-off-by: Greg Kroah-Hartman

Merge 4.7-rc4 into usb-next

2016-06-20T14:40:51+00:00

We need the 4.7-rc4 fixes in here as well.

Signed-off-by: Greg Kroah-Hartman