Skip to content

cloud-kernel
cloud-kernel

openanolis / cloud-kernel
1 年多 前同步成功

通知 161
Star 36
Fork 7
cloud-kernel
cloud-kernel
切换分支/标签
  1. 11 11月, 2016 3 次提交
  3. 08 11月, 2016 6 次提交
  5. 05 11月, 2016 2 次提交
    • N
      PCI: designware: Check for iATU unroll support after initializing host · 416379f9
      由 Niklas Cassel 提交于 
      dw_pcie_iatu_unroll_enabled() reads a dbi_base register.  Reading any
dbi_base register before pp->ops->host_init has been called causes
"imprecise external abort" on platforms like ARTPEC-6, where the PCIe
module is disabled at boot and first enabled in pp->ops->host_init.  Move
dw_pcie_iatu_unroll_enabled() to dw_pcie_setup_rc(), since it is after
pp->ops->host_init, but before pp->iatu_unroll_enabled is actually used.

Fixes: a0601a47 ("PCI: designware: Add iATU Unroll feature")
Tested-by: NJames Le Cuirot <chewi@gentoo.org>
Signed-off-by: NNiklas Cassel <niklas.cassel@axis.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Acked-by: NJoao Pinto <jpinto@synopsys.com>
Acked-by: NOlof Johansson <olof@lixom.net>
      416379f9
    • V
      i2c: core: fix NULL pointer dereference under race condition · 147b36d5
      由 Vladimir Zapolskiy 提交于 
      Race condition between registering an I2C device driver and
deregistering an I2C adapter device which is assumed to manage that
I2C device may lead to a NULL pointer dereference due to the
uninitialized list head of driver clients.

The root cause of the issue is that the I2C bus may know about the
registered device driver and thus it is matched by bus_for_each_drv(),
but the list of clients is not initialized and commonly it is NULL,
because I2C device drivers define struct i2c_driver as static and
clients field is expected to be initialized by I2C core:

  i2c_register_driver()             i2c_del_adapter()
    driver_register()                 ...
      bus_add_driver()                ...
        ...                           bus_for_each_drv(..., __process_removed_adapter)
      ...                               i2c_do_del_adapter()
    ...                                   list_for_each_entry_safe(..., &driver->clients, ...)
    INIT_LIST_HEAD(&driver->clients);

To solve the problem it is sufficient to do clients list head
initialization before calling driver_register().

The problem was found while using an I2C device driver with a sluggish
registration routine on a bus provided by a physically detachable I2C
master controller, but practically the oops may be reproduced under
the race between arbitraty I2C device driver registration and managing
I2C bus device removal e.g. by unbinding the latter over sysfs:

% echo 21a4000.i2c > /sys/bus/platform/drivers/imx-i2c/unbind
  Unable to handle kernel NULL pointer dereference at virtual address 00000000
  Internal error: Oops: 17 [#1] SMP ARM
  CPU: 2 PID: 533 Comm: sh Not tainted 4.9.0-rc3+ #61
  Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
  task: e5ada400 task.stack: e4936000
  PC is at i2c_do_del_adapter+0x20/0xcc
  LR is at __process_removed_adapter+0x14/0x1c
  Flags: NzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment none
  Control: 10c5387d  Table: 35bd004a  DAC: 00000051
  Process sh (pid: 533, stack limit = 0xe4936210)
  Stack: (0xe4937d28 to 0xe4938000)
  Backtrace:
  [<c0667be0>] (i2c_do_del_adapter) from [<c0667cc0>] (__process_removed_adapter+0x14/0x1c)
  [<c0667cac>] (__process_removed_adapter) from [<c0516998>] (bus_for_each_drv+0x6c/0xa0)
  [<c051692c>] (bus_for_each_drv) from [<c06685ec>] (i2c_del_adapter+0xbc/0x284)
  [<c0668530>] (i2c_del_adapter) from [<bf0110ec>] (i2c_imx_remove+0x44/0x164 [i2c_imx])
  [<bf0110a8>] (i2c_imx_remove [i2c_imx]) from [<c051a838>] (platform_drv_remove+0x2c/0x44)
  [<c051a80c>] (platform_drv_remove) from [<c05183d8>] (__device_release_driver+0x90/0x12c)
  [<c0518348>] (__device_release_driver) from [<c051849c>] (device_release_driver+0x28/0x34)
  [<c0518474>] (device_release_driver) from [<c0517150>] (unbind_store+0x80/0x104)
  [<c05170d0>] (unbind_store) from [<c0516520>] (drv_attr_store+0x28/0x34)
  [<c05164f8>] (drv_attr_store) from [<c0298acc>] (sysfs_kf_write+0x50/0x54)
  [<c0298a7c>] (sysfs_kf_write) from [<c029801c>] (kernfs_fop_write+0x100/0x214)
  [<c0297f1c>] (kernfs_fop_write) from [<c0220130>] (__vfs_write+0x34/0x120)
  [<c02200fc>] (__vfs_write) from [<c0221088>] (vfs_write+0xa8/0x170)
  [<c0220fe0>] (vfs_write) from [<c0221e74>] (SyS_write+0x4c/0xa8)
  [<c0221e28>] (SyS_write) from [<c0108a20>] (ret_fast_syscall+0x0/0x1c)
Signed-off-by: NVladimir Zapolskiy <vladimir_zapolskiy@mentor.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>
Cc: stable@kernel.org
      147b36d5
  7. 03 11月, 2016 3 次提交
    • S
      firewire: net: fix fragmented datagram_size off-by-one · e9300a4b
      由 Stefan Richter 提交于 
      RFC 2734 defines the datagram_size field in fragment encapsulation
headers thus:

    datagram_size:  The encoded size of the entire IP datagram.  The
    value of datagram_size [...] SHALL be one less than the value of
    Total Length in the datagram's IP header (see STD 5, RFC 791).

Accordingly, the eth1394 driver of Linux 2.6.36 and older set and got
this field with a -/+1 offset:

    ether1394_tx() /* transmit */
        ether1394_encapsulate_prep()
            hdr->ff.dg_size = dg_size - 1;

    ether1394_data_handler() /* receive */
        if (hdr->common.lf == ETH1394_HDR_LF_FF)
            dg_size = hdr->ff.dg_size + 1;
        else
            dg_size = hdr->sf.dg_size + 1;

Likewise, I observe OS X 10.4 and Windows XP Pro SP3 to transmit 1500
byte sized datagrams in fragments with datagram_size=1499 if link
fragmentation is required.

Only firewire-net sets and gets datagram_size without this offset.  The
result is lacking interoperability of firewire-net with OS X, Windows
XP, and presumably Linux' eth1394.  (I did not test with the latter.)
For example, FTP data transfers to a Linux firewire-net box with max_rec
smaller than the 1500 bytes MTU
  - from OS X fail entirely,
  - from Win XP start out with a bunch of fragmented datagrams which
    time out, then continue with unfragmented datagrams because Win XP
    temporarily reduces the MTU to 576 bytes.

So let's fix firewire-net's datagram_size accessors.

Note that firewire-net thereby loses interoperability with unpatched
firewire-net, but only if link fragmentation is employed.  (This happens
with large broadcast datagrams, and with large datagrams on several
FireWire CardBus cards with smaller max_rec than equivalent PCI cards,
and it can be worked around by setting a small enough MTU.)

Cc: stable@vger.kernel.org
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
      e9300a4b
    • S
      firewire: net: guard against rx buffer overflows · 667121ac
      由 Stefan Richter 提交于 
      The IP-over-1394 driver firewire-net lacked input validation when
handling incoming fragmented datagrams.  A maliciously formed fragment
with a respectively large datagram_offset would cause a memcpy past the
datagram buffer.

So, drop any packets carrying a fragment with offset + length larger
than datagram_size.

In addition, ensure that
  - GASP header, unfragmented encapsulation header, or fragment
    encapsulation header actually exists before we access it,
  - the encapsulated datagram or fragment is of nonzero size.
Reported-by: NEyal Itkin <eyal.itkin@gmail.com>
Reviewed-by: NEyal Itkin <eyal.itkin@gmail.com>
Fixes: CVE 2016-8633
Cc: stable@vger.kernel.org
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>
      667121ac
    • G
      virtio-gpu: fix vblank events · eed6f0ed
      由 Gerd Hoffmann 提交于 
      virtio-gpu sends vblank events in virtio_gpu_crtc_atomic_flush, and
because of that it must be called for disabled planes too.  Ask
drm_atomic_helper_commit_planes to do that.

v2: update to use new drm_atomic_helper_commit_planes() API.
Signed-off-by: NGerd Hoffmann <kraxel@redhat.com>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>
Signed-off-by: NDave Airlie <airlied@redhat.com>
      eed6f0ed
  9. 02 11月, 2016 1 次提交
    • J
      gpio/mvebu: Use irq_domain_add_linear · 812d4788
      由 Jason Gunthorpe 提交于 
      This fixes the irq allocation in this driver to not print:
 irq: Cannot allocate irq_descs @ IRQ34, assuming pre-allocated
 irq: Cannot allocate irq_descs @ IRQ66, assuming pre-allocated

Which happens because the driver already called irq_alloc_descs()
and so the change to use irq_domain_add_simple resulted in calling
irq_alloc_descs() twice.

Modernize the irq allocation in this driver to use the
irq_domain_add_linear flow directly and eliminate the use of
irq_domain_add_simple/legacy

Fixes: ce931f57 ("gpio/mvebu: convert to use irq_domain_add_simple()")
Signed-off-by: NJason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
      812d4788
  11. 01 11月, 2016 3 次提交
    • P
      drm/nouveau/acpi: fix check for power resources support · b0a6af8b
      由 Peter Wu 提交于 
      Check whether the kernel really supports power resources for a device,
otherwise the power might not be removed when the device is runtime
suspended (DSM should still work in these cases where PR does not).

This is a workaround for a problem where ACPICA and Windows 10 differ in
behavior. ACPICA does not correctly enumerate power resources within a
conditional block (due to delayed execution of such blocks) and as a
result power_resources is set to false even if _PR3 exists.

Fixes: 692a17dc ("drm/nouveau/acpi: fix lockup with PCIe runtime PM")
Link: https://bugs.freedesktop.org/show_bug.cgi?id=98398Reported-and-tested-by: NRick Kerkhof <rick.2889@gmail.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Cc: stable@vger.kernel.org # v4.8+
Signed-off-by: NPeter Wu <peter@lekensteyn.nl>
Signed-off-by: NDave Airlie <airlied@redhat.com>
      b0a6af8b
    • M
      gpio: of: fix GPIO drivers with multiple gpio_chip for a single node · c7e9d398
      由 Masahiro Yamada 提交于 
      Sylvain Lemieux reports the LPC32xx GPIO driver is broken since
commit 762c2e46 ("gpio: of: remove of_gpiochip_and_xlate() and
struct gg_data").  Probably, gpio-etraxfs.c and gpio-davinci.c are
broken too.

Those drivers register multiple gpio_chip that are associated to a
single OF node, and their own .of_xlate() checks if the passed
gpio_chip is valid.

Now, the problem is of_find_gpiochip_by_node() returns the first
gpio_chip found to match the given node.  So, .of_xlate() fails,
except for the first GPIO bank.

Reverting the commit could be a solution, but I do not want to go
back to the mess of struct gg_data.  Another solution here is to
take the match by a node pointer and the success of .of_xlate().
It is a bit clumsy to call .of_xlate twice; for gpio_chip matching
and for really getting the gpio_desc index.  Perhaps, our long-term
goal might be to convert the drivers to single chip registration,
but this commit will solve the problem until then.

Fixes: 762c2e46 ("gpio: of: remove of_gpiochip_and_xlate() and struct gg_data")
Signed-off-by: NMasahiro Yamada <yamada.masahiro@socionext.com>
Reported-by: NSylvain Lemieux <slemieux.tyco@gmail.com>
Tested-by: NDavid Lechner <david@lechnology.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
      c7e9d398
    • L
      gpio: GPIO_GET_LINE{HANDLE,EVENT}_IOCTL: Fix file descriptor leak · 953b956a
      由 Lars-Peter Clausen 提交于 
      When allocating a new line handle or event a file is allocated that it is
associated to. The file is attached to a file descriptor of the current
process and the file descriptor is returned to userspace using
copy_to_user(). If this copy operation fails the line handle or event
allocation is aborted, all acquired resources are freed and an error is
returned.

But the file struct is not freed and left attached to the userspace
application and even though the file descriptor number was not copied it is
trivial to guess. If a userspace application performs a IOCTL on such a
left over file descriptor it will trigger a use-after-free and if the file
descriptor is closed (latest when the application exits) a double-free is
triggered.

anon_inode_getfd() performs 3 tasks, allocate a file struct, allocate a
file descriptor for the current process and install the file struct in the
file descriptor. As soon as the file struct is installed in the file
descriptor it is accessible by userspace (even if the IOCTL itself hasn't
completed yet), this means uninstalling the fd on the error path is not an
option, since userspace might already got a reference to the file.

Instead anon_inode_getfd() needs to be broken into its individual steps.
The allocation of the file struct and file descriptor is done first, then
the copy_to_user() is executed and only if it succeeds the file is
installed.

Since the file struct is reference counted it can not be just freed, but
its reference needs to be dropped, which will also call the release()
callback, which will free the state attached to the file. So in this case
the normal error cleanup path should not be taken.

Cc: stable@vger.kernel.org
Fixes: d932cd49 ("gpio: free handles in fringe cases")
Signed-off-by: NLars-Peter Clausen <lars@metafoo.de>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
      953b956a
  13. 31 10月, 2016 9 次提交
  15. 30 10月, 2016 13 次提交