openwrt/staging/blogic.git
5 years agostaging: bcm2835-camera: Provide more specific probe error messages
Stefan Wahren [Thu, 6 Dec 2018 18:28:59 +0000 (19:28 +0100)]
staging: bcm2835-camera: Provide more specific probe error messages

Currently there is only a catch-all info message which print the
relevant error code without any context. So add more specific error
messages in order to narrow down possible issues.

Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: bcm2835-audio: Drop DT dependency
Stefan Wahren [Thu, 6 Dec 2018 18:28:58 +0000 (19:28 +0100)]
staging: bcm2835-audio: Drop DT dependency

Just like the bcm2835-video make this a platform driver which is probed
by vchiq. In order to change the number of channels use a module
parameter instead, but use the maximum as default.

Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: bcm2835-audio: use module_platform_driver() macro
Stefan Wahren [Thu, 6 Dec 2018 18:28:57 +0000 (19:28 +0100)]
staging: bcm2835-audio: use module_platform_driver() macro

There is not much value behind this boilerplate, so use
module_platform_driver() instead.

Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: bcm2835-audio: Enable compile test
Stefan Wahren [Thu, 6 Dec 2018 18:28:56 +0000 (19:28 +0100)]
staging: bcm2835-audio: Enable compile test

Enable the compilation test for bcm2835-audio to gain more build coverage.

Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: vchiq_arm: Register a platform device for audio
Stefan Wahren [Thu, 6 Dec 2018 18:28:55 +0000 (19:28 +0100)]
staging: vchiq_arm: Register a platform device for audio

Following Eric's commit 37b7b3087a2f ("staging/vc04_services: Register a
platform device for the camera driver.") this register the audio driver as
a platform device, too.

Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: vchiq_arm: Fix camera device registration
Stefan Wahren [Thu, 6 Dec 2018 18:28:54 +0000 (19:28 +0100)]
staging: vchiq_arm: Fix camera device registration

Since the camera driver isn't probed via DT, we need to properly setup DMA.

Fixes: 37b7b3087a2f ("staging/vc04_services: Register a platform device for the camera driver.")
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: bcm2835-camera: Move module info to the end
Stefan Wahren [Thu, 6 Dec 2018 18:28:53 +0000 (19:28 +0100)]
staging: bcm2835-camera: Move module info to the end

In order to have this more consistent between the vc04 services move
the module information to the end of the file.

Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: bcm2835-camera: fix module autoloading
Stefan Wahren [Thu, 6 Dec 2018 18:28:52 +0000 (19:28 +0100)]
staging: bcm2835-camera: fix module autoloading

In order to make the module bcm2835-camera load automatically, we need to
add a module alias.

Fixes: 4bebb0312ea9 ("staging/bcm2835-camera: Set ourselves up as a platform driver.")
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: bcm2835-camera: Abort probe if there is no camera
Stefan Wahren [Thu, 6 Dec 2018 18:28:51 +0000 (19:28 +0100)]
staging: bcm2835-camera: Abort probe if there is no camera

Abort the probing of the camera driver in case there isn't a camera
actually connected to the Raspberry Pi. This solution also avoids a
NULL ptr dereference of mmal instance on driver unload.

Fixes: 7b3ad5abf027 ("staging: Import the BCM2835 MMAL-based V4L2 camera driver.")
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-dts: avoid warnings in pinctrl definitions
NeilBrown [Sun, 9 Dec 2018 05:20:33 +0000 (16:20 +1100)]
staging: mt7621-dts: avoid warnings in pinctrl definitions

The device-tree checking code sees node names  "i2c" and "spi" in the
pinctrl definition and thinks these are defining i2c or spi devices,
and complains that they look wrong.

So add a '0' to the end of each name (much like "uart" and "rgmii"
have numbers at the end) to avoid the warning.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-mmc: fix compile warnging: cmd_buf
NeilBrown [Sun, 9 Dec 2018 05:20:33 +0000 (16:20 +1100)]
staging: mt7621-mmc: fix compile warnging: cmd_buf

cmd_buf is only used when MT6575_SD_DEBUG is defined.
So only declare it in that case.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-mmc: add missing header.
NeilBrown [Sun, 9 Dec 2018 05:20:33 +0000 (16:20 +1100)]
staging: mt7621-mmc: add missing header.

<linux/interrupt.h> is sometimes included by other include files,
and sometimes not, depending on config, particularly CONFIG_HIGHMEM.

So include it explicitly rather than relying on implicit inclusion.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-eth: fix compile warning.
NeilBrown [Sun, 9 Dec 2018 05:20:32 +0000 (16:20 +1100)]
staging: mt7621-eth: fix compile warning.

This code generates a waring as PHY_GBIT_FEATURES is "long" but
->supported in "int".

It looks likely that "PHY_1000BT_FEATURES" is the correct define to
use - it is intended to be used with the ->features field.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-dts: allow gnubee to reboot cleanly.
NeilBrown [Sun, 9 Dec 2018 05:20:32 +0000 (16:20 +1100)]
staging: mt7621-dts: allow gnubee to reboot cleanly.

Since commit bb276262e88d ("mtd: spi-nor: only apply reset hacks to
broken hardware"), we need to mark the spi-nor as "broken" for reboot
to work.
Note that nothing is actually broken here.  The hardware-watchdog in
the SoC isn't wired in a way that works, but then the board doesn't
claim to support a hardware watchdog - and the SPI certain isn't
"broken".

This causes an annoying warning on every boot, but that is better than
failing on ever reboot.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-mmc: set correct dma mask.
NeilBrown [Sun, 9 Dec 2018 05:20:32 +0000 (16:20 +1100)]
staging: mt7621-mmc: set correct dma mask.

Since commit f8c55dc6e828 ("MIPS: use generic dma noncoherent ops for
simple noncoherent platforms") changed MIPS dma handling, the mmc
driver fails because it doesn't have a dma mask is set.

So set the correct dma mask.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-eth: set correct dma mask.
NeilBrown [Sun, 9 Dec 2018 05:20:32 +0000 (16:20 +1100)]
staging: mt7621-eth: set correct dma mask.

Since commit f8c55dc6e828 ("MIPS: use generic dma noncoherent ops for
simple noncoherent platforms") changed MIPS dma handling, the eth
driver fails because the dma mask is set on the wrong 'struct device'.

Move the setting to the correct struct device.

Signed-off-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: vchiq: rework remove_event handling
Arnd Bergmann [Mon, 10 Dec 2018 21:11:58 +0000 (22:11 +0100)]
staging: vchiq: rework remove_event handling

I had started the removal of semaphores in this driver without knowing
that Nicolas Saenz Julienne also worked on this. In case of the "remote
event" infrastructure, my solution seemed significantly better, so I'm
proposing this as a change on top.

The problem with using either semaphores or completions here is that
it's an overly complex way of waking up a thread, and it looks like the
'count' of the semaphore can easily get out of sync, even though I found
it hard to come up with a specific example.

Changing it to a 'wait_queue_head_t' instead of a completion simplifies
this by letting us wait directly on the 'event->fired' variable that is
set by the videocore.

Another simplification is passing the wait queue directly into the helper
functions instead of going through the fragile logic of recording the
offset inside of a structure as part of a shared memory variable. This
also avoids one uncached memory read and should be faster.

Note that I'm changing it back to 'killable' after the previous patch
changed 'killable' to 'interruptible', apparently based on a misunderstanding
of the subtle down_interruptible() macro override in vchiq_killable.h.

Fixes: f27e47bc6b8b ("staging: vchiq: use completions instead of semaphores")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8723bs: remove semaphore remnants
Arnd Bergmann [Mon, 10 Dec 2018 21:40:32 +0000 (22:40 +0100)]
staging: rtl8723bs: remove semaphore remnants

Nothing uses the semaphores any more in this driver, so remove
all references to that type.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8723bs: change pwrctrl lock to a mutex
Arnd Bergmann [Mon, 10 Dec 2018 21:40:31 +0000 (22:40 +0100)]
staging: rtl8723bs: change pwrctrl lock to a mutex

This semaphore is used like a mutex, so it should use the regular
mutex API, as we do in the other copies of this driver.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8723bs: change semaphores to completions
Arnd Bergmann [Mon, 10 Dec 2018 21:40:30 +0000 (22:40 +0100)]
staging: rtl8723bs: change semaphores to completions

This driver uses many semaphores, most of them are equivalent to
completions. The other copies of this driver got moved over to
completions a while ago, so do the same here.

In this usage scenario, the two are equivalent, so the behavior
should not change.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8723bs: fix typo in the TODO file
Michael Straube [Mon, 10 Dec 2018 09:52:02 +0000 (10:52 +0100)]
staging: rtl8723bs: fix typo in the TODO file

HGz -> GHz

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8723bs: remove dead code, rtw_port_switch_chk()
Michael Straube [Mon, 10 Dec 2018 09:52:01 +0000 (10:52 +0100)]
staging: rtl8723bs: remove dead code, rtw_port_switch_chk()

The function rtw_port_switch_chk() returns always false.
Remove the function and related dead code.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8188eu: reuse Hal_GetChnlGroup88E()
Michael Straube [Fri, 7 Dec 2018 20:55:15 +0000 (21:55 +0100)]
staging: rtl8188eu: reuse Hal_GetChnlGroup88E()

Use Hal_GetChnlGroup88E() instead of duplicating it's code
in get_rx_power_val_by_reg().

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: rtl8188eu: simplify loop in rtl88eu_phy_iq_calibrate()
Michael Straube [Fri, 7 Dec 2018 20:55:14 +0000 (21:55 +0100)]
staging: rtl8188eu: simplify loop in rtl88eu_phy_iq_calibrate()

Zeroing the array result[m][n] and setting only the values at
even 'n's simplifies the code and slightly reduces object file
size.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: comedi: change do_insn*_ioctl to allow more samples
Spencer E. Olson [Tue, 4 Dec 2018 19:07:50 +0000 (12:07 -0700)]
staging: comedi: change do_insn*_ioctl to allow more samples

Changes do_insn*_ioctl functions to allow for data lengths for each
comedi_insn of up to 2^16.  This patch also changes these functions to only
allocate as much memory as is necessary for each comedi_insn, rather than
allocating a fixed-sized scratch space.

In testing some user-space code for the new INSN_DEVICE_CONFIG_GET_ROUTES
facility with some newer hardware, I discovered that do_insn_ioctl and
do_insnlist_ioctl limited the amount of data that can be passed into the
kernel for insn's to a length of 256.  For some newer hardware, the number
of routes can be greater than 1000.  Working around the old limits (256)
would complicate the user-space/kernel interaction.

The new upper limit is reasonable with current memory available and does
not otherwise impact the memory footprint for any current or otherwise
typical configuration.

Signed-off-by: Spencer E. Olson <olsonse@umich.edu>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-spi: drop support for SPI mode 1/2/3
Chuanhong Guo [Thu, 6 Dec 2018 13:15:09 +0000 (21:15 +0800)]
staging: mt7621-spi: drop support for SPI mode 1/2/3

As explained in previous patch, this SPI controller seems to be
tested on SPI flash only before mass production and some bits are
swizzled under other SPI modes probably due to incorrect wiring
inside the silicon. Drop implementation of SPI mode 1/2/3 since
they are broken.

Also drop RT2880_SPI_MODE_BITS macro because we now have only
SPI_LSB_FIRST implemented and the mode_bits is so short that we
don't need a macro there.

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Reviewed-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: mt7621-spi: drop the broken full-duplex mode
Chuanhong Guo [Thu, 6 Dec 2018 13:15:08 +0000 (21:15 +0800)]
staging: mt7621-spi: drop the broken full-duplex mode

According to John Crispin (aka blogic) on IRC on Nov 26 2018:
  so basically i made cs1 work for MTK/labs when i built
  the linkit smart for them. the req-sheet said that cs1 should be proper
  duplex spi. however ....
   1) the core will always send 1 byte before any transfer, this is the
      m25p80 command.
   2) mode 3 is broken and bit reversed (?)
   3) some bit are incorrectly wired in hw for mode2/3
  we wrote a test script and test for [0-0xffff] on all modes and certain
  bits are swizzled under certain conditions and it was not possible to
  fix this even using a hack.
  we then decided to use spi-gpio and i never removed the errornous code
  basically the spi is fecked for anything but half duplex spi mode0
  running a sflash on it

The controller will always send some data from OPCODE register under half
duplex mode before starting a full-duplex transfer, so the full-duplex
mode is broken.
This piece of code also make CS1 unavailable since it forces the
broken full-duplex mode to be used on CS1.

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Reviewed-by: NeilBrown <neil@brown.name>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: speakup: change semaphore to completion
Arnd Bergmann [Mon, 10 Dec 2018 21:41:50 +0000 (22:41 +0100)]
staging: speakup: change semaphore to completion

In this driver, both function the same way, but we want to eventually
kill off semaphores, so a completion is the better choice here.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: Fix spelling issue
Aaron Strahlberger [Tue, 11 Dec 2018 10:54:41 +0000 (11:54 +0100)]
staging: erofs: Fix spelling issue

Changed "stoped" to "stopped".

Signed-off-by: Aaron Strahlberger <aaron.strahlberger@posteo.de>
Signed-off-by: Julius Wiedmann <julius.wiedmann@fau.de>
Signed-off-by: Dominik Huber <domi250@gmx.de>
Reviewed-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: remove __EROFS_BIT
Gao Xiang [Tue, 11 Dec 2018 17:57:30 +0000 (01:57 +0800)]
staging: erofs: remove __EROFS_BIT

It's better to use pre-calculated values to make
on-disk definition more straight-forward and human-readable.

Since there is the only one user, let's remove
__EROFS_BIT entirely.

Suggested-by: Joe Perches <joe@perches.com>
Signed-off-by: Gao Xiang <hsiangkao@aol.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: unzip_vle_lz4.c,utils.c: rectify BUG_ONs
Gao Xiang [Tue, 11 Dec 2018 07:17:50 +0000 (15:17 +0800)]
staging: erofs: unzip_vle_lz4.c,utils.c: rectify BUG_ONs

remove all redundant BUG_ONs, and turn the rest
useful usages to DBG_BUGONs.

Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: unzip_{pagevec.h,vle.c}: rectify BUG_ONs
Gao Xiang [Tue, 11 Dec 2018 07:17:49 +0000 (15:17 +0800)]
staging: erofs: unzip_{pagevec.h,vle.c}: rectify BUG_ONs

remove all redundant BUG_ONs, and turn the rest
useful usages to DBG_BUGONs.

Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoMerge 4.20-rc6 into staging-next
Greg Kroah-Hartman [Mon, 10 Dec 2018 08:23:50 +0000 (09:23 +0100)]
Merge 4.20-rc6 into staging-next

We want the staging fixes in here as well.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoLinux 4.20-rc6
Linus Torvalds [Sun, 9 Dec 2018 23:31:00 +0000 (15:31 -0800)]
Linux 4.20-rc6

5 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Sun, 9 Dec 2018 23:12:33 +0000 (15:12 -0800)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:
 "A decent batch of fixes here. I'd say about half are for problems that
  have existed for a while, and half are for new regressions added in
  the 4.20 merge window.

   1) Fix 10G SFP phy module detection in mvpp2, from Baruch Siach.

   2) Revert bogus emac driver change, from Benjamin Herrenschmidt.

   3) Handle BPF exported data structure with pointers when building
      32-bit userland, from Daniel Borkmann.

   4) Memory leak fix in act_police, from Davide Caratti.

   5) Check RX checksum offload in RX descriptors properly in aquantia
      driver, from Dmitry Bogdanov.

   6) SKB unlink fix in various spots, from Edward Cree.

   7) ndo_dflt_fdb_dump() only works with ethernet, enforce this, from
      Eric Dumazet.

   8) Fix FID leak in mlxsw driver, from Ido Schimmel.

   9) IOTLB locking fix in vhost, from Jean-Philippe Brucker.

  10) Fix SKB truesize accounting in ipv4/ipv6/netfilter frag memory
      limits otherwise namespace exit can hang. From Jiri Wiesner.

  11) Address block parsing length fixes in x25 from Martin Schiller.

  12) IRQ and ring accounting fixes in bnxt_en, from Michael Chan.

  13) For tun interfaces, only iface delete works with rtnl ops, enforce
      this by disallowing add. From Nicolas Dichtel.

  14) Use after free in liquidio, from Pan Bian.

  15) Fix SKB use after passing to netif_receive_skb(), from Prashant
      Bhole.

  16) Static key accounting and other fixes in XPS from Sabrina Dubroca.

  17) Partially initialized flow key passed to ip6_route_output(), from
      Shmulik Ladkani.

  18) Fix RTNL deadlock during reset in ibmvnic driver, from Thomas
      Falcon.

  19) Several small TCP fixes (off-by-one on window probe abort, NULL
      deref in tail loss probe, SNMP mis-estimations) from Yuchung
      Cheng"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (93 commits)
  net/sched: cls_flower: Reject duplicated rules also under skip_sw
  bnxt_en: Fix _bnxt_get_max_rings() for 57500 chips.
  bnxt_en: Fix NQ/CP rings accounting on the new 57500 chips.
  bnxt_en: Keep track of reserved IRQs.
  bnxt_en: Fix CNP CoS queue regression.
  net/mlx4_core: Correctly set PFC param if global pause is turned off.
  Revert "net/ibm/emac: wrong bit is used for STA control"
  neighbour: Avoid writing before skb->head in neigh_hh_output()
  ipv6: Check available headroom in ip6_xmit() even without options
  tcp: lack of available data can also cause TSO defer
  ipv6: sr: properly initialize flowi6 prior passing to ip6_route_output
  mlxsw: spectrum_switchdev: Fix VLAN device deletion via ioctl
  mlxsw: spectrum_router: Relax GRE decap matching check
  mlxsw: spectrum_switchdev: Avoid leaking FID's reference count
  mlxsw: spectrum_nve: Remove easily triggerable warnings
  ipv4: ipv6: netfilter: Adjust the frag mem limit when truesize changes
  sctp: frag_point sanity check
  tcp: fix NULL ref in tail loss probe
  tcp: Do not underestimate rwnd_limited
  net: use skb_list_del_init() to remove from RX sublists
  ...

5 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 9 Dec 2018 23:09:55 +0000 (15:09 -0800)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Ingo Molnar:
 "Three fixes: a boot parameter re-(re-)fix, a retpoline build artifact
  fix and an LLVM workaround"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/vdso: Drop implicit common-page-size linker flag
  x86/build: Fix compiler support check for CONFIG_RETPOLINE
  x86/boot: Clear RSDP address in boot_params for broken loaders

5 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 9 Dec 2018 22:21:33 +0000 (14:21 -0800)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull kprobes fixes from Ingo Molnar:
 "Two kprobes fixes: a blacklist fix and an instruction patching related
  corruption fix"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  kprobes/x86: Blacklist non-attachable interrupt functions
  kprobes/x86: Fix instruction patching corruption when copying more than one RIP-relative instruction

5 years agoMerge branch 'efi-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 9 Dec 2018 22:03:56 +0000 (14:03 -0800)]
Merge branch 'efi-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull EFI fixes from Ingo Molnar:
 "Two fixes: a large-system fix and an earlyprintk fix with certain
  resolutions"

* 'efi-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/earlyprintk/efi: Fix infinite loop on some screen widths
  x86/efi: Allocate e820 buffer before calling efi_exit_boot_service

5 years agonet/sched: cls_flower: Reject duplicated rules also under skip_sw
Or Gerlitz [Sun, 9 Dec 2018 16:10:24 +0000 (18:10 +0200)]
net/sched: cls_flower: Reject duplicated rules also under skip_sw

Currently, duplicated rules are rejected only for skip_hw or "none",
hence allowing users to push duplicates into HW for no reason.

Use the flower tables to protect for that.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Reported-by: Chris Mi <chrism@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'bnxt_en-Bug-fixes'
David S. Miller [Sun, 9 Dec 2018 19:46:59 +0000 (11:46 -0800)]
Merge branch 'bnxt_en-Bug-fixes'

Michael Chan says:

====================
bnxt_en: Bug fixes.

The first patch fixes a regression on CoS queue setup, introduced
recently by the 57500 new chip support patches.  The rest are
fixes related to ring and resource accounting on the new 57500 chips.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Fix _bnxt_get_max_rings() for 57500 chips.
Michael Chan [Sun, 9 Dec 2018 12:01:02 +0000 (07:01 -0500)]
bnxt_en: Fix _bnxt_get_max_rings() for 57500 chips.

The CP rings are accounted differently on the new 57500 chips.  There
must be enough CP rings for the sum of RX and TX rings on the new
chips.  The current logic may be over-estimating the RX and TX rings.

The output parameter max_cp should be the maximum NQs capped by
MSIX vectors available for networking in the context of 57500 chips.
The existing code which uses CMPL rings capped by the MSIX vectors
works most of the time but is not always correct.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Fix NQ/CP rings accounting on the new 57500 chips.
Michael Chan [Sun, 9 Dec 2018 12:01:01 +0000 (07:01 -0500)]
bnxt_en: Fix NQ/CP rings accounting on the new 57500 chips.

The new 57500 chips have introduced the NQ structure in addition to
the existing CP rings in all chips.  We need to introduce a new
bnxt_nq_rings_in_use().  On legacy chips, the 2 functions are the
same and one will just call the other.  On the new chips, they
refer to the 2 separate ring structures.  The new function is now
called to determine the resource (NQ or CP rings) associated with
MSIX that are in use.

On 57500 chips, the RDMA driver does not use the CP rings so
we don't need to do the subtraction adjustment.

Fixes: 41e8d7983752 ("bnxt_en: Modify the ring reservation functions for 57500 series chips.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Keep track of reserved IRQs.
Michael Chan [Sun, 9 Dec 2018 12:01:00 +0000 (07:01 -0500)]
bnxt_en: Keep track of reserved IRQs.

The new 57500 chips use 1 NQ per MSIX vector, whereas legacy chips use
1 CP ring per MSIX vector.  To better unify this, add a resv_irqs
field to struct bnxt_hw_resc.  On legacy chips, we initialize resv_irqs
with resv_cp_rings.  On new chips, we initialize it with the allocated
MSIX resources.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agobnxt_en: Fix CNP CoS queue regression.
Michael Chan [Sun, 9 Dec 2018 12:00:59 +0000 (07:00 -0500)]
bnxt_en: Fix CNP CoS queue regression.

Recent changes to support the 57500 devices have created this
regression.  The bnxt_hwrm_queue_qportcfg() call was moved to be
called earlier before the RDMA support was determined, causing
the CoS queues configuration to be set before knowing whether RDMA
was supported or not.  Fix it by moving it to the right place right
after RDMA support is determined.

Fixes: 98f04cf0f1fc ("bnxt_en: Check context memory requirements from firmware.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'char-misc-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregk...
Linus Torvalds [Sun, 9 Dec 2018 18:43:17 +0000 (10:43 -0800)]
Merge tag 'char-misc-4.20-rc6' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are some small driver fixes for 4.20-rc6.

  There is a hyperv fix that for some reaon took forever to get into a
  shape that could be applied to the tree properly, but resolves a much
  reported issue. The others are some gnss patches, one a bugfix and the
  two others updates to the MAINTAINERS file to properly match the gnss
  files in the tree.

  All have been in linux-next for a while with no reported issues"

* tag 'char-misc-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  MAINTAINERS: exclude gnss from SIRFPRIMA2 regex matching
  MAINTAINERS: add gnss scm tree
  gnss: sirf: fix activation retry handling
  Drivers: hv: vmbus: Offload the handling of channels to two workqueues

5 years agoMerge tag 'staging-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 9 Dec 2018 18:35:33 +0000 (10:35 -0800)]
Merge tag 'staging-4.20-rc6' of git://git./linux/kernel/git/gregkh/staging

Pull staging fixes from Greg KH:
 "Here are two staging driver bugfixes for 4.20-rc6.

  One is a revert of a previously incorrect patch that was merged a
  while ago, and the other resolves a possible buffer overrun that was
  found by code inspection.

  Both of these have been in the linux-next tree with no reported
  issues"

* tag 'staging-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  Revert commit ef9209b642f "staging: rtl8723bs: Fix indenting errors and an off-by-one mistake in core/rtw_mlme_ext.c"
  staging: rtl8712: Fix possible buffer overrun

5 years agoMerge tag 'tty-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 9 Dec 2018 18:24:29 +0000 (10:24 -0800)]
Merge tag 'tty-4.20-rc6' of git://git./linux/kernel/git/gregkh/tty

Pull tty driver fixes from Greg KH:
 "Here are three small tty driver fixes for 4.20-rc6

  Nothing major, just some bug fixes for reported issues. Full details
  are in the shortlog.

  All of these have been in linux-next for a while with no reported
  issues"

* tag 'tty-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  kgdboc: fix KASAN global-out-of-bounds bug in param_set_kgdboc_var()
  tty: serial: 8250_mtk: always resume the device in probe.
  tty: do not set TTY_IO_ERROR flag if console port

5 years agoMerge tag 'usb-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Sun, 9 Dec 2018 18:18:24 +0000 (10:18 -0800)]
Merge tag 'usb-4.20-rc6' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB fixes for 4.20-rc6

  The "largest" here are some xhci fixes for reported issues. Also here
  is a USB core fix, some quirk additions, and a usb-serial fix which
  required the export of one of the tty layer's functions to prevent
  code duplication. The tty maintainer agreed with this change.

  All of these have been in linux-next with no reported issues"

* tag 'usb-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  xhci: Prevent U1/U2 link pm states if exit latency is too long
  xhci: workaround CSS timeout on AMD SNPS 3.0 xHC
  USB: check usb_get_extra_descriptor for proper size
  USB: serial: console: fix reported terminal settings
  usb: quirk: add no-LPM quirk on SanDisk Ultra Flair device
  USB: Fix invalid-free bug in port_over_current_notify()
  usb: appledisplay: Add 27" Apple Cinema Display

5 years agoMerge tag '4.20-rc5-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 9 Dec 2018 18:15:13 +0000 (10:15 -0800)]
Merge tag '4.20-rc5-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "Three small fixes: a fix for smb3 direct i/o, a fix for CIFS DFS for
  stable and a minor cifs Kconfig fix"

* tag '4.20-rc5-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  CIFS: Avoid returning EBUSY to upper layer VFS
  cifs: Fix separator when building path from dentry
  cifs: In Kconfig CONFIG_CIFS_POSIX needs depends on legacy (insecure cifs)

5 years agoMerge tag 'dax-fixes-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdim...
Linus Torvalds [Sun, 9 Dec 2018 17:54:04 +0000 (09:54 -0800)]
Merge tag 'dax-fixes-4.20-rc6' of git://git./linux/kernel/git/nvdimm/nvdimm

Pull dax fixes from Dan Williams:
 "The last of the known regression fixes and fallout from the Xarray
  conversion of the filesystem-dax implementation.

  On the path to debugging why the dax memory-failure injection test
  started failing after the Xarray conversion a couple more fixes for
  the dax_lock_mapping_entry(), now called dax_lock_page(), surfaced.
  Those plus the bug that started the hunt are now addressed. These
  patches have appeared in a -next release with no issues reported.

  Note the touches to mm/memory-failure.c are just the conversion to the
  new function signature for dax_lock_page().

  Summary:

   - Fix the Xarray conversion of fsdax to properly handle
     dax_lock_mapping_entry() in the presense of pmd entries

   - Fix inode destruction racing a new lock request"

* tag 'dax-fixes-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  dax: Fix unlock mismatch with updated API
  dax: Don't access a freed inode
  dax: Check page->mapping isn't NULL

5 years agoMerge tag 'libnvdimm-fixes-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 9 Dec 2018 17:46:54 +0000 (09:46 -0800)]
Merge tag 'libnvdimm-fixes-4.20-rc6' of git://git./linux/kernel/git/nvdimm/nvdimm

Pull libnvdimm fixes from Dan Williams:
 "A regression fix for the Address Range Scrub implementation, yes
  another one, and support for platforms that misalign persistent memory
  relative to the Linux memory hotplug section constraint. Longer term,
  support for sub-section memory hotplug would alleviate alignment
  waste, but until then this hack allows a 'struct page' memmap to be
  established for these misaligned memory regions.

  These have all appeared in a -next release, and thanks to Patrick for
  reporting and testing the alignment padding fix.

  Summary:

   - Unless and until the core mm handles memory hotplug units smaller
     than a section (128M), persistent memory namespaces must be padded
     to section alignment.

     The libnvdimm core already handled section collision with "System
     RAM", but some configurations overlap independent "Persistent
     Memory" ranges within a section, so additional padding injection is
     added for that case.

   - The recent reworks of the ARS (address range scrub) state machine
     to reduce the number of state flags inadvertantly missed a
     conversion of acpi_nfit_ars_rescan() call sites. Fix the regression
     whereby user-requested ARS results in a "short" scrub rather than a
     "long" scrub.

   - Fixup the unit tests to handle / test the 128M section alignment of
     mocked test resources.

* tag 'libnvdimm-fixes-4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  acpi/nfit: Fix user-initiated ARS to be "ARS-long" rather than "ARS-short"
  libnvdimm, pfn: Pad pfn namespaces relative to other regions
  tools/testing/nvdimm: Align test resources to 128M

5 years agonet/mlx4_core: Correctly set PFC param if global pause is turned off.
Tarick Bedeir [Fri, 7 Dec 2018 08:30:26 +0000 (00:30 -0800)]
net/mlx4_core: Correctly set PFC param if global pause is turned off.

rx_ppp and tx_ppp can be set between 0 and 255, so don't clamp to 1.

Fixes: 6e8814ceb7e8 ("net/mlx4_en: Fix mixed PFC and Global pause user control requests")
Signed-off-by: Tarick Bedeir <tarick@google.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux...
Linus Torvalds [Sun, 9 Dec 2018 01:45:20 +0000 (17:45 -0800)]
Merge branch 'fixes' of git://git./linux/kernel/git/evalenti/linux-soc-thermal

Pull thermal SoC fixes from Eduardo Valentin:
 "Fixes for armada and broadcom thermal drivers"

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/evalenti/linux-soc-thermal:
  thermal: broadcom: constify thermal_zone_of_device_ops structure
  thermal: armada: constify thermal_zone_of_device_ops structure
  thermal: bcm2835: Switch to SPDX identifier
  thermal: armada: fix legacy resource fixup
  thermal: armada: fix legacy validity test sense

5 years agoMerge tag 'asm-generic-4.20' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd...
Linus Torvalds [Sat, 8 Dec 2018 19:44:04 +0000 (11:44 -0800)]
Merge tag 'asm-generic-4.20' of git://git./linux/kernel/git/arnd/asm-generic

Pull asm-generic fix from Arnd Bergmann:
 "Multiple people reported a bug I introduced in asm-generic/unistd.h in
  4.20, this is the obvious bugfix to get glibc and others to correctly
  build again on new architectures that no longer provide the old
  fstatat64() family of system calls"

* tag 'asm-generic-4.20' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  asm-generic: unistd.h: fixup broken macro include.

5 years agoMerge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 8 Dec 2018 19:33:26 +0000 (11:33 -0800)]
Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "A few clk driver fixes this time:

   - Introduce protected-clock DT binding to fix breakage on qcom
     sdm845-mtp boards where the qspi clks introduced this merge window
     cause the firmware on those boards to take down the system if we
     try to read the clk registers

   - Fix a couple off-by-one errors found by Dan Carpenter

   - Handle failure in zynq fixed factor clk driver to avoid using
     uninitialized data"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: zynqmp: Off by one in zynqmp_is_valid_clock()
  clk: mmp: Off by one in mmp_clk_add()
  clk: mvebu: Off by one bugs in cp110_of_clk_get()
  arm64: dts: qcom: sdm845-mtp: Mark protected gcc clocks
  clk: qcom: Support 'protected-clocks' property
  dt-bindings: clk: Introduce 'protected-clocks' property
  clk: zynqmp: handle fixed factor param query error

5 years agoMerge tag 'xfs-4.20-fixes-3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Sat, 8 Dec 2018 19:25:02 +0000 (11:25 -0800)]
Merge tag 'xfs-4.20-fixes-3' of git://git./fs/xfs/xfs-linux

Pull xfs fixes from Darrick Wong:
 "Here are hopefully the last set of fixes for 4.20.

  There's a fix for a longstanding statfs reporting problem with project
  quotas, a correction for page cache invalidation behaviors when
  fallocating near EOF, and a fix for a broken metadata verifier return
  code.

  Finally, the most important fix is to the pipe splicing code (aka the
  generic copy_file_range fallback) to avoid pointless short directio
  reads by only asking the filesystem for as much data as there are
  available pages in the pipe buffer. Our previous fix (simulated short
  directio reads because the number of pages didn't match the length of
  the read requested) caused subtle problems on overlayfs, so that part
  is reverted.

  Anyhow, this series passes fstests -g all on xfs and overlay+xfs, and
  has passed 17 billion fsx operations problem-free since I started
  testing

  Summary:

   - Fix broken project quota inode counts

   - Fix incorrect PAGE_MASK/PAGE_SIZE usage

   - Fix incorrect return value in btree verifier

   - Fix WARN_ON remap flags false positive

   - Fix splice read overflows"

* tag 'xfs-4.20-fixes-3' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  iomap: partially revert 4721a601099 (simulated directio short read on EFAULT)
  splice: don't read more than available pipe space
  vfs: allow some remap flags to be passed to vfs_clone_file_range
  xfs: fix inverted return from xfs_btree_sblock_verify_crc
  xfs: fix PAGE_MASK usage in xfs_free_file_space
  fs/xfs: fix f_ffree value for statfs when project quota is set

5 years agoRevert "mm, thp: consolidate THP gfp handling into alloc_hugepage_direct_gfpmask"
David Rientjes [Fri, 7 Dec 2018 22:50:16 +0000 (14:50 -0800)]
Revert "mm, thp: consolidate THP gfp handling into alloc_hugepage_direct_gfpmask"

This reverts commit 89c83fb539f95491be80cdd5158e6f0ce329e317.

This should have been done as part of 2f0799a0ffc0 ("mm, thp: restore
node-local hugepage allocations").  The movement of the thp allocation
policy from alloc_pages_vma() to alloc_hugepage_direct_gfpmask() was
intended to only set __GFP_THISNODE for mempolicies that are not
MPOL_BIND whereas the revert could set this regardless of mempolicy.

While the check for MPOL_BIND between alloc_hugepage_direct_gfpmask()
and alloc_pages_vma() was racy, that has since been removed since the
revert.  What is left is the possibility to use __GFP_THISNODE in
policy_node() when it is unexpected because the special handling for
hugepages in alloc_pages_vma()  was removed as part of the consolidation.

Secondly, prior to 89c83fb539f9, alloc_pages_vma() implemented a somewhat
different policy for hugepage allocations, which were allocated through
alloc_hugepage_vma().  For hugepage allocations, if the allocating
process's node is in the set of allowed nodes, allocate with
__GFP_THISNODE for that node (for MPOL_PREFERRED, use that node with
__GFP_THISNODE instead).  This was changed for shmem_alloc_hugepage() to
allow fallback to other nodes in 89c83fb539f9 as it did for new_page() in
mm/mempolicy.c which is functionally different behavior and removes the
requirement to only allocate hugepages locally.

So this commit does a full revert of 89c83fb539f9 instead of the partial
revert that was done in 2f0799a0ffc0.  The result is the same thp
allocation policy for 4.20 that was in 4.19.

Fixes: 89c83fb539f9 ("mm, thp: consolidate THP gfp handling into alloc_hugepage_direct_gfpmask")
Fixes: 2f0799a0ffc0 ("mm, thp: restore node-local hugepage allocations")
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoRevert "net/ibm/emac: wrong bit is used for STA control"
Benjamin Herrenschmidt [Fri, 7 Dec 2018 04:05:04 +0000 (15:05 +1100)]
Revert "net/ibm/emac: wrong bit is used for STA control"

This reverts commit 624ca9c33c8a853a4a589836e310d776620f4ab9.

This commit is completely bogus. The STACR register has two formats, old
and new, depending on the version of the IP block used. There's a pair of
device-tree properties that can be used to specify the format used:

has-inverted-stacr-oc
has-new-stacr-staopc

What this commit did was to change the bit definition used with the old
parts to match the new parts. This of course breaks the driver on all
the old ones.

Instead, the author should have set the appropriate properties in the
device-tree for the variant used on his board.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge branch 'skb-headroom-slab-out-of-bounds'
David S. Miller [Sat, 8 Dec 2018 00:24:40 +0000 (16:24 -0800)]
Merge branch 'skb-headroom-slab-out-of-bounds'

Stefano Brivio says:

====================
Fix slab out-of-bounds on insufficient headroom for IPv6 packets

Patch 1/2 fixes a slab out-of-bounds occurring with short SCTP packets over
IPv4 over L2TP over IPv6 on a configuration with relatively low HEADER_MAX.

Patch 2/2 makes sure we avoid writing before the allocated buffer in
neigh_hh_output() in case the headroom is enough for the unaligned hardware
header size, but not enough for the aligned one, and that we warn if we hit
this condition.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoneighbour: Avoid writing before skb->head in neigh_hh_output()
Stefano Brivio [Thu, 6 Dec 2018 18:30:37 +0000 (19:30 +0100)]
neighbour: Avoid writing before skb->head in neigh_hh_output()

While skb_push() makes the kernel panic if the skb headroom is less than
the unaligned hardware header size, it will proceed normally in case we
copy more than that because of alignment, and we'll silently corrupt
adjacent slabs.

In the case fixed by the previous patch,
"ipv6: Check available headroom in ip6_xmit() even without options", we
end up in neigh_hh_output() with 14 bytes headroom, 14 bytes hardware
header and write 16 bytes, starting 2 bytes before the allocated buffer.

Always check we're not writing before skb->head and, if the headroom is
not enough, warn and drop the packet.

v2:
 - instead of panicking with BUG_ON(), WARN_ON_ONCE() and drop the packet
   (Eric Dumazet)
 - if we avoid the panic, though, we need to explicitly check the headroom
   before the memcpy(), otherwise we'll have corrupted slabs on a running
   kernel, after we warn
 - use __skb_push() instead of skb_push(), as the headroom check is
   already implemented here explicitly (Eric Dumazet)

Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoipv6: Check available headroom in ip6_xmit() even without options
Stefano Brivio [Thu, 6 Dec 2018 18:30:36 +0000 (19:30 +0100)]
ipv6: Check available headroom in ip6_xmit() even without options

Even if we send an IPv6 packet without options, MAX_HEADER might not be
enough to account for the additional headroom required by alignment of
hardware headers.

On a configuration without HYPERV_NET, WLAN, AX25, and with IPV6_TUNNEL,
sending short SCTP packets over IPv4 over L2TP over IPv6, we start with
100 bytes of allocated headroom in sctp_packet_transmit(), end up with 54
bytes after l2tp_xmit_skb(), and 14 bytes in ip6_finish_output2().

Those would be enough to append our 14 bytes header, but we're going to
align that to 16 bytes, and write 2 bytes out of the allocated slab in
neigh_hh_output().

KASan says:

[  264.967848] ==================================================================
[  264.967861] BUG: KASAN: slab-out-of-bounds in ip6_finish_output2+0x1aec/0x1c70
[  264.967866] Write of size 16 at addr 000000006af1c7fe by task netperf/6201
[  264.967870]
[  264.967876] CPU: 0 PID: 6201 Comm: netperf Not tainted 4.20.0-rc4+ #1
[  264.967881] Hardware name: IBM 2827 H43 400 (z/VM 6.4.0)
[  264.967887] Call Trace:
[  264.967896] ([<00000000001347d6>] show_stack+0x56/0xa0)
[  264.967903]  [<00000000017e379c>] dump_stack+0x23c/0x290
[  264.967912]  [<00000000007bc594>] print_address_description+0xf4/0x290
[  264.967919]  [<00000000007bc8fc>] kasan_report+0x13c/0x240
[  264.967927]  [<000000000162f5e4>] ip6_finish_output2+0x1aec/0x1c70
[  264.967935]  [<000000000163f890>] ip6_finish_output+0x430/0x7f0
[  264.967943]  [<000000000163fe44>] ip6_output+0x1f4/0x580
[  264.967953]  [<000000000163882a>] ip6_xmit+0xfea/0x1ce8
[  264.967963]  [<00000000017396e2>] inet6_csk_xmit+0x282/0x3f8
[  264.968033]  [<000003ff805fb0ba>] l2tp_xmit_skb+0xe02/0x13e0 [l2tp_core]
[  264.968037]  [<000003ff80631192>] l2tp_eth_dev_xmit+0xda/0x150 [l2tp_eth]
[  264.968041]  [<0000000001220020>] dev_hard_start_xmit+0x268/0x928
[  264.968069]  [<0000000001330e8e>] sch_direct_xmit+0x7ae/0x1350
[  264.968071]  [<000000000122359c>] __dev_queue_xmit+0x2b7c/0x3478
[  264.968075]  [<00000000013d2862>] ip_finish_output2+0xce2/0x11a0
[  264.968078]  [<00000000013d9b14>] ip_finish_output+0x56c/0x8c8
[  264.968081]  [<00000000013ddd1e>] ip_output+0x226/0x4c0
[  264.968083]  [<00000000013dbd6c>] __ip_queue_xmit+0x894/0x1938
[  264.968100]  [<000003ff80bc3a5c>] sctp_packet_transmit+0x29d4/0x3648 [sctp]
[  264.968116]  [<000003ff80b7bf68>] sctp_outq_flush_ctrl.constprop.5+0x8d0/0xe50 [sctp]
[  264.968131]  [<000003ff80b7c716>] sctp_outq_flush+0x22e/0x7d8 [sctp]
[  264.968146]  [<000003ff80b35c68>] sctp_cmd_interpreter.isra.16+0x530/0x6800 [sctp]
[  264.968161]  [<000003ff80b3410a>] sctp_do_sm+0x222/0x648 [sctp]
[  264.968177]  [<000003ff80bbddac>] sctp_primitive_ASSOCIATE+0xbc/0xf8 [sctp]
[  264.968192]  [<000003ff80b93328>] __sctp_connect+0x830/0xc20 [sctp]
[  264.968208]  [<000003ff80bb11ce>] sctp_inet_connect+0x2e6/0x378 [sctp]
[  264.968212]  [<0000000001197942>] __sys_connect+0x21a/0x450
[  264.968215]  [<000000000119aff8>] sys_socketcall+0x3d0/0xb08
[  264.968218]  [<000000000184ea7a>] system_call+0x2a2/0x2c0

[...]

Just like ip_finish_output2() does for IPv4, check that we have enough
headroom in ip6_xmit(), and reallocate it if we don't.

This issue is older than git history.

Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agotcp: lack of available data can also cause TSO defer
Eric Dumazet [Thu, 6 Dec 2018 17:58:24 +0000 (09:58 -0800)]
tcp: lack of available data can also cause TSO defer

tcp_tso_should_defer() can return true in three different cases :

 1) We are cwnd-limited
 2) We are rwnd-limited
 3) We are application limited.

Neal pointed out that my recent fix went too far, since
it assumed that if we were not in 1) case, we must be rwnd-limited

Fix this by properly populating the is_cwnd_limited and
is_rwnd_limited booleans.

After this change, we can finally move the silly check for FIN
flag only for the application-limited case.

The same move for EOR bit will be handled in net-next,
since commit 1c09f7d073b1 ("tcp: do not try to defer skbs
with eor mark (MSG_EOR)") is scheduled for linux-4.21

Tested by running 200 concurrent netperf -t TCP_RR -- -r 60000,100
and checking none of them was rwnd_limited in the chrono_stat
output from "ss -ti" command.

Fixes: 41727549de3e ("tcp: Do not underestimate rwnd_limited")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Suggested-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Linus Torvalds [Fri, 7 Dec 2018 22:34:10 +0000 (14:34 -0800)]
Merge tag 'for_linus' of git://git./linux/kernel/git/mst/vhost

Pull vhost/virtio fixes from Michael Tsirkin:
 "A couple of last-minute fixes"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  vhost/vsock: fix use-after-free in network stack callers
  virtio/s390: fix race in ccw_io_helper()
  virtio/s390: avoid race on vcdev->config
  vhost/vsock: fix reset orphans race with close timeout

5 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 7 Dec 2018 22:18:49 +0000 (14:18 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fix from Catalin Marinas:
 "Avoid sending IPIs with interrupts disabled"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: hibernate: Avoid sending cross-calling with interrupts disabled

5 years agoMerge tag 'gcc-plugins-v4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 7 Dec 2018 21:13:07 +0000 (13:13 -0800)]
Merge tag 'gcc-plugins-v4.20-rc6' of git://git./linux/kernel/git/kees/linux

Pull gcc stackleak plugin fixes from Kees Cook:

 - Remove tracing for inserted stack depth marking function (Anders
   Roxell)

 - Move gcc-plugin pass location to avoid objtool warnings (Alexander
   Popov)

* tag 'gcc-plugins-v4.20-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  stackleak: Register the 'stackleak_cleanup' pass before the '*free_cfg' pass
  stackleak: Mark stackleak_track_stack() as notrace

5 years agoMerge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Fri, 7 Dec 2018 21:07:10 +0000 (13:07 -0800)]
Merge branch 'linus' of git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fixes from Herbert Xu:

 - Disable the new crypto stats interface as it's still being changed

 - Fix potential uses-after-free in cbc/cfb/pcbc.

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: user - Disable statistics interface
  crypto: do not free algorithm before using

5 years agoMerge tag 'pci-v4.20-fixes-3' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaa...
Linus Torvalds [Fri, 7 Dec 2018 20:58:34 +0000 (12:58 -0800)]
Merge tag 'pci-v4.20-fixes-3' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:
 "Revert ASPM change that caused a regression"

* tag 'pci-v4.20-fixes-3' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  Revert "PCI/ASPM: Do not initialize link state when aspm_disabled is set"

5 years agoipv6: sr: properly initialize flowi6 prior passing to ip6_route_output
Shmulik Ladkani [Fri, 7 Dec 2018 07:50:17 +0000 (09:50 +0200)]
ipv6: sr: properly initialize flowi6 prior passing to ip6_route_output

In 'seg6_output', stack variable 'struct flowi6 fl6' was missing
initialization.

Fixes: 6c8702c60b88 ("ipv6: sr: add support for SRH encapsulation and injection with lwtunnels")
Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agoMerge tag 'for-linus-20181207' of git://git.kernel.dk/linux-block
Linus Torvalds [Fri, 7 Dec 2018 18:40:37 +0000 (10:40 -0800)]
Merge tag 'for-linus-20181207' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:
 "Let's try this again...

  We're finally happy with the DM livelock issue, and it's also passed
  overnight testing and the corruption regression test. The end result
  is much nicer now too, which is great.

  Outside of that fix, there's a pull request for NVMe with two small
  fixes, and a regression fix for BFQ from this merge window. The BFQ
  fix looks bigger than it is, it's 90% comment updates"

* tag 'for-linus-20181207' of git://git.kernel.dk/linux-block:
  blk-mq: punt failed direct issue to dispatch list
  nvmet-rdma: fix response use after free
  nvme: validate controller state before rescheduling keep alive
  block, bfq: fix decrement of num_active_groups

5 years agoMerge branch 'i2c/for-current-fixed' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 7 Dec 2018 18:31:31 +0000 (10:31 -0800)]
Merge branch 'i2c/for-current-fixed' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "A set of driver bugfixes for the I2C subsystem"

* 'i2c/for-current-fixed' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: uniphier-f: fix violation of tLOW requirement for Fast-mode
  i2c: uniphier: fix violation of tLOW requirement for Fast-mode
  i2c: uniphier-f: fill TX-FIFO only in IRQ handler for repeated START
  i2c: uniphier-f: fix timeout error after reading 8 bytes
  i2c: scmi: Fix probe error on devices with an empty SMB0001 ACPI device node
  i2c: axxia: properly handle master timeout
  i2c: rcar: check bus state before reinitializing
  i2c: nvidia-gpu: limit reads also for combined messages
  i2c: nvidia-gpu: adhere to I2C fault codes

5 years agoMerge tag 'dmaengine-fix-4.20-rc6' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Fri, 7 Dec 2018 17:58:34 +0000 (09:58 -0800)]
Merge tag 'dmaengine-fix-4.20-rc6' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine fixes from Vinod Koul:
 "Another pull request for dmaengine. We got bunch of fixes early this
  week and all are tagged to stable. Hope this is last fix for this
  cycle:

   - Fix imx-sdma handling of channel terminations, this involves
     reverting two commits and implement async termination

   - Fix cppi dma channel deletion from pending list on stop

   - Fix FIFO size for dw controller in Intel Merrifield"

* tag 'dmaengine-fix-4.20-rc6' of git://git.infradead.org/users/vkoul/slave-dma:
  dmaengine: dw: Fix FIFO size for Intel Merrifield
  dmaengine: cppi41: delete channel from pending list when stop channel
  dmaengine: imx-sdma: use GFP_NOWAIT for dma descriptor allocations
  dmaengine: imx-sdma: implement channel termination via worker
  Revert "dmaengine: imx-sdma: alloclate bd memory from dma pool"
  Revert "dmaengine: imx-sdma: Use GFP_NOWAIT for dma allocations"

5 years agox86/vdso: Drop implicit common-page-size linker flag
Nick Desaulniers [Thu, 6 Dec 2018 19:12:31 +0000 (11:12 -0800)]
x86/vdso: Drop implicit common-page-size linker flag

GNU linker's -z common-page-size's default value is based on the target
architecture. arch/x86/entry/vdso/Makefile sets it to the architecture
default, which is implicit and redundant. Drop it.

Fixes: 2aae950b21e4 ("x86_64: Add vDSO for x86-64 with gettimeofday/clock_gettime/getcpu")
Reported-by: Dmitry Golovin <dima@golovin.in>
Reported-by: Bill Wendling <morbo@google.com>
Suggested-by: Dmitry Golovin <dima@golovin.in>
Suggested-by: Rui Ueyama <ruiu@google.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Andy Lutomirski <luto@kernel.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Fangrui Song <maskray@google.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: x86-ml <x86@kernel.org>
Link: https://lkml.kernel.org/r/20181206191231.192355-1-ndesaulniers@google.com
Link: https://bugs.llvm.org/show_bug.cgi?id=38774
Link: https://github.com/ClangBuiltLinux/linux/issues/31
5 years agostaging: erofs: simplify `z_erofs_vle_submit_all'
Gao Xiang [Fri, 7 Dec 2018 16:19:18 +0000 (00:19 +0800)]
staging: erofs: simplify `z_erofs_vle_submit_all'

Previously, there are too many hacked stuffs such as `__FSIO_1',
`lstgrp_noio', `lstgrp_io' out there in `z_erofs_vle_submit_all'.

Revisit the whole process by properly introducing jobqueue to
represent each type of queued workgroups, furthermore hide all of
crazyness behind independent separated functions.

After this patch, 2 independent jobqueues exist if managed cache
is enabled, or 1 jobqueue if disabled.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: redefine where `owned_workgrp_t' points
Gao Xiang [Fri, 7 Dec 2018 16:19:17 +0000 (00:19 +0800)]
staging: erofs: redefine where `owned_workgrp_t' points

By design, workgroups are queued in the form of linked lists.

Previously, it points to the next `z_erofs_vle_workgroup',
which isn't flexible enough to simplify `z_erofs_vle_submit_all'.

Let's fix it by pointing to the next `owned_workgrp_t' and use
container_of to get its coresponding `z_erofs_vle_workgroup'.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: refine compressed pages preload flow
Gao Xiang [Fri, 7 Dec 2018 16:19:16 +0000 (00:19 +0800)]
staging: erofs: refine compressed pages preload flow

Currently, there are two kinds of compressed pages in erofs:
  1) file pages for the in-place decompression and
  2) managed pages for cached decompression.
Both are all stored in grp->compressed_pages[].

For managed pages, they could already exist or could be preloaded
in this round, including the following cases in detail:
  1) Already valid (loaded in some previous round);
  2) PAGE_UNALLOCATED, should be allocated at the time of submission;
  3) Just found in the managed cache, and with an extra page ref.
Currently, 1) and 3) can be distinguishable by lock_page and
checking its PG_private, which is guaranteed by the reclaim path,
but it's better to do a double check by using an extra tag.

This patch reworks the preload flow by introducing such the tag
by using tagged pointer, too many #ifdefs are removed as well.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: revisit the page submission flow
Gao Xiang [Fri, 7 Dec 2018 16:19:15 +0000 (00:19 +0800)]
staging: erofs: revisit the page submission flow

Previously, the submission flow works with cached compressed pages
reclaim path in a tricky way, and it could be buggy if the reclaim
path changes later without such tricky restrictions. For example,
currently one PagePrivate(page) is evaluated without taking page
lock (it only follows a wait_for_page_locked which closes such race)
and no handling solves the potential page truncation case.

In addition, it's also full of #ifdefs in the function, which
is hard to understand and maintain. this patch fixes them all.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: localize UNALLOCATED_CACHED_PAGE placeholder
Gao Xiang [Fri, 7 Dec 2018 16:19:14 +0000 (00:19 +0800)]
staging: erofs: localize UNALLOCATED_CACHED_PAGE placeholder

In practice, in order to do cached decompression rather than reuse
them for in-place decompression and make full use of pages in
page_pool instead of allocating as much as possible, an unallocated
placeholder was introduce to mark all in compressed_pages[] and
they will be replaced at the time of submission.

Previously EROFS_UNALLOCATED_CACHED_PAGE was included in internal.h,
which is unnecessary since it's only internally used in decompression
subsystem, move it to unzip_vle.c and rename it to PAGE_UNALLOCATED.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: introduce MNGD_MAPPING helper
Gao Xiang [Fri, 7 Dec 2018 16:19:13 +0000 (00:19 +0800)]
staging: erofs: introduce MNGD_MAPPING helper

This patch introduces MNGD_MAPPING to wrap up
sbi->managed_cache->i_mapping, which will be used
to solve too many #ifdefs in a single function.

No logic changes.

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agostaging: erofs: fix use-after-free of on-stack `z_erofs_vle_unzip_io'
Gao Xiang [Fri, 7 Dec 2018 16:19:12 +0000 (00:19 +0800)]
staging: erofs: fix use-after-free of on-stack `z_erofs_vle_unzip_io'

The root cause is the race as follows:
 Thread #0                         Thread #1

 z_erofs_vle_unzip_kickoff         z_erofs_submit_and_unzip

                                    struct z_erofs_vle_unzip_io io[]
   atomic_add_return()
                                    wait_event()
                                    [end of function]
   wake_up()

Fix it by taking the waitqueue lock between atomic_add_return and
wake_up to close such the race.

kernel message:

Unable to handle kernel paging request at virtual address 97f7052caa1303dc
...
Workqueue: kverityd verity_work
task: ffffffe32bcb8000 task.stack: ffffffe3298a0000
PC is at __wake_up_common+0x48/0xa8
LR is at __wake_up+0x3c/0x58
...
Call trace:
...
[<ffffff94a08ff648>] __wake_up_common+0x48/0xa8
[<ffffff94a08ff8b8>] __wake_up+0x3c/0x58
[<ffffff94a0c11b60>] z_erofs_vle_unzip_kickoff+0x40/0x64
[<ffffff94a0c118e4>] z_erofs_vle_read_endio+0x94/0x134
[<ffffff94a0c83c9c>] bio_endio+0xe4/0xf8
[<ffffff94a1076540>] dec_pending+0x134/0x32c
[<ffffff94a1076f28>] clone_endio+0x90/0xf4
[<ffffff94a0c83c9c>] bio_endio+0xe4/0xf8
[<ffffff94a1095024>] verity_work+0x210/0x368
[<ffffff94a08c4150>] process_one_work+0x188/0x4b4
[<ffffff94a08c45bc>] worker_thread+0x140/0x458
[<ffffff94a08cad48>] kthread+0xec/0x108
[<ffffff94a0883ab4>] ret_from_fork+0x10/0x1c
Code: d1006273 54000260 f9400804 b9400019 (b85fc081)
---[ end trace be9dde154f677cd1 ]---

Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
5 years agoarm64: hibernate: Avoid sending cross-calling with interrupts disabled
Will Deacon [Fri, 7 Dec 2018 12:47:10 +0000 (12:47 +0000)]
arm64: hibernate: Avoid sending cross-calling with interrupts disabled

Since commit 3b8c9f1cdfc50 ("arm64: IPI each CPU after invalidating the
I-cache for kernel mappings"), a call to flush_icache_range() will use
an IPI to cross-call other online CPUs so that any stale instructions
are flushed from their pipelines. This triggers a WARN during the
hibernation resume path, where flush_icache_range() is called with
interrupts disabled and is therefore prone to deadlock:

  | Disabling non-boot CPUs ...
  | CPU1: shutdown
  | psci: CPU1 killed.
  | CPU2: shutdown
  | psci: CPU2 killed.
  | CPU3: shutdown
  | psci: CPU3 killed.
  | WARNING: CPU: 0 PID: 1 at ../kernel/smp.c:416 smp_call_function_many+0xd4/0x350
  | Modules linked in:
  | CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.20.0-rc4 #1

Since all secondary CPUs have been taken offline prior to invalidating
the I-cache, there's actually no need for an IPI and we can simply call
__flush_icache_range() instead.

Cc: <stable@vger.kernel.org>
Fixes: 3b8c9f1cdfc50 ("arm64: IPI each CPU after invalidating the I-cache for kernel mappings")
Reported-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Tested-by: Kunihiko Hayashi <hayashi.kunihiko@socionext.com>
Tested-by: James Morse <james.morse@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
5 years agoMerge branch 'nvme-4.20' of git://git.infradead.org/nvme into for-linus
Jens Axboe [Fri, 7 Dec 2018 15:40:13 +0000 (08:40 -0700)]
Merge branch 'nvme-4.20' of git://git.infradead.org/nvme into for-linus

Pull NVMe fixes from Christoph.

* 'nvme-4.20' of git://git.infradead.org/nvme:
  nvmet-rdma: fix response use after free
  nvme: validate controller state before rescheduling keep alive

5 years agoblk-mq: punt failed direct issue to dispatch list
Jens Axboe [Fri, 7 Dec 2018 05:17:44 +0000 (22:17 -0700)]
blk-mq: punt failed direct issue to dispatch list

After the direct dispatch corruption fix, we permanently disallow direct
dispatch of non read/write requests. This works fine off the normal IO
path, as they will be retried like any other failed direct dispatch
request. But for the blk_insert_cloned_request() that only DM uses to
bypass the bottom level scheduler, we always first attempt direct
dispatch. For some types of requests, that's now a permanent failure,
and no amount of retrying will make that succeed. This results in a
livelock.

Instead of making special cases for what we can direct issue, and now
having to deal with DM solving the livelock while still retaining a BUSY
condition feedback loop, always just add a request that has been through
->queue_rq() to the hardware queue dispatch list. These are safe to use
as no merging can take place there. Additionally, if requests do have
prepped data from drivers, we aren't dependent on them not sharing space
in the request structure to safely add them to the IO scheduler lists.

This basically reverts ffe81d45322c and is based on a patch from Ming,
but with the list insert case covered as well.

Fixes: ffe81d45322c ("blk-mq: fix corruption with direct issue")
Cc: stable@vger.kernel.org
Suggested-by: Ming Lei <ming.lei@redhat.com>
Reported-by: Bart Van Assche <bvanassche@acm.org>
Tested-by: Ming Lei <ming.lei@redhat.com>
Acked-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 years agonvmet-rdma: fix response use after free
Israel Rukshin [Wed, 5 Dec 2018 16:54:57 +0000 (16:54 +0000)]
nvmet-rdma: fix response use after free

nvmet_rdma_release_rsp() may free the response before using it at error
flow.

Fixes: 8407879 ("nvmet-rdma: fix possible bogus dereference under heavy load")
Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agonvme: validate controller state before rescheduling keep alive
James Smart [Wed, 28 Nov 2018 01:04:44 +0000 (17:04 -0800)]
nvme: validate controller state before rescheduling keep alive

Delete operations are seeing NULL pointer references in call_timer_fn.
Tracking these back, the timer appears to be the keep alive timer.

nvme_keep_alive_work() which is tied to the timer that is cancelled
by nvme_stop_keep_alive(), simply starts the keep alive io but doesn't
wait for it's completion. So nvme_stop_keep_alive() only stops a timer
when it's pending. When a keep alive is in flight, there is no timer
running and the nvme_stop_keep_alive() will have no affect on the keep
alive io. Thus, if the io completes successfully, the keep alive timer
will be rescheduled.   In the failure case, delete is called, the
controller state is changed, the nvme_stop_keep_alive() is called while
the io is outstanding, and the delete path continues on. The keep
alive happens to successfully complete before the delete paths mark it
as aborted as part of the queue termination, so the timer is restarted.
The delete paths then tear down the controller, and later on the timer
code fires and the timer entry is now corrupt.

Fix by validating the controller state before rescheduling the keep
alive. Testing with the fix has confirmed the condition above was hit.

Signed-off-by: James Smart <jsmart2021@gmail.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoblock, bfq: fix decrement of num_active_groups
Paolo Valente [Thu, 6 Dec 2018 18:18:18 +0000 (19:18 +0100)]
block, bfq: fix decrement of num_active_groups

Since commit '2d29c9f89fcd ("block, bfq: improve asymmetric scenarios
detection")', if there are process groups with I/O requests waiting for
completion, then BFQ tags the scenario as 'asymmetric'. This detection
is needed for preserving service guarantees (for details, see comments
on the computation * of the variable asymmetric_scenario in the
function bfq_better_to_idle).

Unfortunately, commit '2d29c9f89fcd ("block, bfq: improve asymmetric
scenarios detection")' contains an error exactly in the updating of
the number of groups with I/O requests waiting for completion: if a
group has more than one descendant process, then the above number of
groups, which is renamed from num_active_groups to a more appropriate
num_groups_with_pending_reqs by this commit, may happen to be wrongly
decremented multiple times, namely every time one of the descendant
processes gets all its pending I/O requests completed.

A correct, complete solution should work as follows. Consider a group
that is inactive, i.e., that has no descendant process with pending
I/O inside BFQ queues. Then suppose that num_groups_with_pending_reqs
is still accounting for this group, because the group still has some
descendant process with some I/O request still in
flight. num_groups_with_pending_reqs should be decremented when the
in-flight request of the last descendant process is finally completed
(assuming that nothing else has changed for the group in the meantime,
in terms of composition of the group and active/inactive state of
child groups and processes). To accomplish this, an additional
pending-request counter must be added to entities, and must be
updated correctly.

To avoid this additional field and operations, this commit resorts to
the following tradeoff between simplicity and accuracy: for an
inactive group that is still counted in num_groups_with_pending_reqs,
this commit decrements num_groups_with_pending_reqs when the first
descendant process of the group remains with no request waiting for
completion.

This simplified scheme provides a fix to the unbalanced decrements
introduced by 2d29c9f89fcd. Since this error was also caused by lack
of comments on this non-trivial issue, this commit also adds related
comments.

Fixes: 2d29c9f89fcd ("block, bfq: improve asymmetric scenarios detection")
Reported-by: Steven Barrett <steven@liquorix.net>
Tested-by: Steven Barrett <steven@liquorix.net>
Tested-by: Lucjan Lucjanov <lucjan.lucjanov@gmail.com>
Reviewed-by: Federico Motta <federico@willer.it>
Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 years agoMerge tag 'gnss-4.20-rc6' of https://git.kernel.org/pub/scm/linux/kernel/git/johan...
Greg Kroah-Hartman [Fri, 7 Dec 2018 13:05:28 +0000 (14:05 +0100)]
Merge tag 'gnss-4.20-rc6' of https://git./linux/kernel/git/johan/gnss into char-misc-linus

Johan writes:

GNSS fixes for 4.20-rc6

Here's a fix for a broken activation retry loop in the sirf driver.

Included are also two MAINTAINERS updates.

All have been in linux-next with no reported issues.

Signed-off-by: Johan Hovold <johan@kernel.org>
* tag 'gnss-4.20-rc6' of https://git.kernel.org/pub/scm/linux/kernel/git/johan/gnss:
  MAINTAINERS: exclude gnss from SIRFPRIMA2 regex matching
  MAINTAINERS: add gnss scm tree
  gnss: sirf: fix activation retry handling

5 years agoCIFS: Avoid returning EBUSY to upper layer VFS
Long Li [Thu, 6 Dec 2018 04:51:06 +0000 (04:51 +0000)]
CIFS: Avoid returning EBUSY to upper layer VFS

EBUSY is not handled by VFS, and will be passed to user-mode. This is not
correct as we need to wait for more credits.

This patch also fixes a bug where rsize or wsize is used uninitialized when
the call to server->ops->wait_mtu_credits() fails.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Long Li <longli@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Reviewed-by: Pavel Shilovsky <pshilov@microsoft.com>
5 years agocrypto: user - Disable statistics interface
Herbert Xu [Fri, 7 Dec 2018 05:56:08 +0000 (13:56 +0800)]
crypto: user - Disable statistics interface

Since this user-space API is still undergoing significant changes,
this patch disables it for the current merge window.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
5 years agoMerge tag 'drm-fixes-2018-12-07' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 7 Dec 2018 03:35:50 +0000 (19:35 -0800)]
Merge tag 'drm-fixes-2018-12-07' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "There's a bit more in here than I'd like, and I'm hoping things calm
  down when I'm out.

  msm:
   - a bunch of display fixes for the new DPU
   - a couple of command submission fixes

  omap:
   - some DSI fixes

  ast:
   - driver unload crash fix

  core:
   - fix the lease uevent so userspace can distinguish it

  amd:
   - fix a bpc regression
   - fix lru handling regression
   - fixed firmware support for new GPUs
   - power management fixes for vega20"

* tag 'drm-fixes-2018-12-07' of git://anongit.freedesktop.org/drm/drm: (37 commits)
  drm/ast: Fix connector leak during driver unload
  drm/amdgpu/vcn: Update vcn.cur_state during suspend
  drm/amd/display: Fix overflow/truncation from strncpy.
  drm/amd/powerplay: improve OD code robustness
  drm/amdgpu: enlarge maximum waiting time of KIQ
  drm/fb-helper: Fix typo in parameter description
  drm/amd/powerplay: support SoftMin/Max setting for some specific DPM
  drm/amd/powerplay: issue pre-display settings for display change event
  drm/amd/powerplay: support new pptable upload on Vega20
  drm/amdgpu/gmc8: always load MC firmware in the driver
  drm/amdgpu/gmc8: update MC firmware for polaris
  drm/amdgpu: update mc firmware image for polaris12 variants
  drm/msm: Fix error return checking
  drm/msm/dpu: Ignore alpha for XBGR8888 format
  drm/msm: dpu: Fix "WARNING: invalid free of devm_ allocated data"
  drm/msm/hdmi: Drop pointless static qualifier in msm_hdmi_bind()
  drm/msm: Move fence put to where failure occurs
  drm/msm: dpu: Don't set legacy plane->crtc pointer
  drm/msm/gpu: Don't map command buffers with nr_relocs equal to 0
  drm/msm/hdmi: Enable HPD after HDMI IRQ is set up
  ...

5 years agoMerge tag 'nfs-for-4.20-5' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Fri, 7 Dec 2018 02:57:04 +0000 (18:57 -0800)]
Merge tag 'nfs-for-4.20-5' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 "This is mainly fallout from the updates to the SUNRPC code that is
  being triggered from less common combinations of NFS mount options.

  Highlights include:

  Stable fixes:
   - Fix a page leak when using RPCSEC_GSS/krb5p to encrypt data.

  Bugfixes:
   - Fix a regression that causes the RPC receive code to hang
   - Fix call_connect_status() so that it handles tasks that got
     transmitted while queued waiting for the socket lock.
   - Fix a memory leak in call_encode()
   - Fix several other connect races.
   - Fix receive code error handling.
   - Use the discard iterator rather than MSG_TRUNC for compatibility
     with AF_UNIX/AF_LOCAL sockets.
   - nfs: don't dirty kernel pages read by direct-io
   - pnfs/Flexfiles fix to enforce per-mirror stateid only for NFSv4
     data servers"

* tag 'nfs-for-4.20-5' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  SUNRPC: Don't force a redundant disconnection in xs_read_stream()
  SUNRPC: Fix up socket polling
  SUNRPC: Use the discard iterator rather than MSG_TRUNC
  SUNRPC: Treat EFAULT as a truncated message in xs_read_stream_request()
  SUNRPC: Fix up handling of the XDRBUF_SPARSE_PAGES flag
  SUNRPC: Fix RPC receive hangs
  SUNRPC: Fix a potential race in xprt_connect()
  SUNRPC: Fix a memory leak in call_encode()
  SUNRPC: Fix leak of krb5p encode pages
  SUNRPC: call_connect_status() must handle tasks that got transmitted
  nfs: don't dirty kernel pages read by direct-io
  flexfiles: enforce per-mirror stateid only for v4 DSes

5 years agoMerge branch 'spectre' of git://git.armlinux.org.uk/~rmk/linux-arm
Linus Torvalds [Fri, 7 Dec 2018 00:45:36 +0000 (16:45 -0800)]
Merge branch 'spectre' of git://git.armlinux.org.uk/~rmk/linux-arm

Pull ARM spectre fix from Russell King:
 "Exynos folk noticed that CPU hotplug wasn't working with their kernel
  configuration, and have tested this as fixing the problem"

* 'spectre' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: ensure that processor vtables is not lost after boot

5 years agoMerge branch 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm
Linus Torvalds [Fri, 7 Dec 2018 00:39:44 +0000 (16:39 -0800)]
Merge branch 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm

Pull ARM fixes from Russell King:
 "Some small fixes that have been accumulated:

   - Chris Cole noticed that in a SMP environment, the DMA cache
     coherence handling can produce undesirable results in a corner
     case

   - Propagate that fix for ARMv7M as well

   - Fix a false positive with source fortification

   - Fix an uninitialised return that Nathan Jones spotted"

* 'fixes' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: 8816/1: dma-mapping: fix potential uninitialized return
  ARM: 8815/1: V7M: align v7m_dma_inv_range() with v7 counterpart
  ARM: 8814/1: mm: improve/fix ARM v7_dma_inv_range() unaligned address handling
  ARM: 8806/1: kprobes: Fix false positive with FORTIFY_SOURCE

5 years agoi2c: uniphier-f: fix violation of tLOW requirement for Fast-mode
Masahiro Yamada [Thu, 6 Dec 2018 03:55:28 +0000 (12:55 +0900)]
i2c: uniphier-f: fix violation of tLOW requirement for Fast-mode

Currently, the clock duty is set as tLOW/tHIGH = 1/1. For Fast-mode,
tLOW is set to 1.25 us while the I2C spec requires tLOW >= 1.3 us.

tLOW/tHIGH = 5/4 would meet both Standard-mode and Fast-mode:
  Standard-mode: tLOW = 5.56 us, tHIGH = 4.44 us
  Fast-mode:     tLOW = 1.39 us, tHIGH = 1.11 us

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoi2c: uniphier: fix violation of tLOW requirement for Fast-mode
Masahiro Yamada [Thu, 6 Dec 2018 03:55:27 +0000 (12:55 +0900)]
i2c: uniphier: fix violation of tLOW requirement for Fast-mode

Currently, the clock duty is set as tLOW/tHIGH = 1/1. For Fast-mode,
tLOW is set to 1.25 us while the I2C spec requires tLOW >= 1.3 us.

tLOW/tHIGH = 5/4 would meet both Standard-mode and Fast-mode:
  Standard-mode: tLOW = 5.56 us, tHIGH = 4.44 us
  Fast-mode:     tLOW = 1.39 us, tHIGH = 1.11 us

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoi2c: uniphier-f: fill TX-FIFO only in IRQ handler for repeated START
Masahiro Yamada [Thu, 6 Dec 2018 03:55:26 +0000 (12:55 +0900)]
i2c: uniphier-f: fill TX-FIFO only in IRQ handler for repeated START

- For a repeated START condition, this controller starts data transfer
   immediately after the slave address is written to the TX-FIFO.

 - Once the TX-FIFO empty interrupt is asserted, the controller makes
   a pause even if additional data are written to the TX-FIFO.

Given those circumstances, the data after a repeated START may not be
transferred if the interrupt is asserted while the TX-FIFO is being
filled up. A more reliable way is to append TX data only in the
interrupt handler.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoi2c: uniphier-f: fix timeout error after reading 8 bytes
Masahiro Yamada [Thu, 6 Dec 2018 03:55:25 +0000 (12:55 +0900)]
i2c: uniphier-f: fix timeout error after reading 8 bytes

I was totally screwed up in commit eaba68785c2d ("i2c: uniphier-f:
fix race condition when IRQ is cleared"). Since that commit, if the
number of read bytes is multiple of the FIFO size (8, 16, 24... bytes),
the STOP condition could be issued twice, depending on the timing.
If this happens, the controller will go wrong, resulting in the timeout
error.

It was more than 3 years ago when I wrote this driver, so my memory
about this hardware was vague. Please let me correct the description
in the commit log of eaba68785c2d.

Clearing the IRQ status on exiting the IRQ handler is absolutely
fine. This controller makes a pause while any IRQ status is asserted.
If the IRQ status is cleared first, the hardware may start the next
transaction before the IRQ handler finishes what it supposed to do.

This partially reverts the bad commit with clear comments so that I
will never repeat this mistake.

I also investigated what is happening at the last moment of the read
mode. The UNIPHIER_FI2C_INT_RF interrupt is asserted a bit earlier
(by half a period of the clock cycle) than UNIPHIER_FI2C_INT_RB.

I consulted a hardware engineer, and I got the following information:

UNIPHIER_FI2C_INT_RF
    asserted at the falling edge of SCL at the 8th bit.

UNIPHIER_FI2C_INT_RB
    asserted at the rising edge of SCL at the 9th (ACK) bit.

In order to avoid calling uniphier_fi2c_stop() twice, check the latter
interrupt. I also commented this because it is obscure hardware internal.

Fixes: eaba68785c2d ("i2c: uniphier-f: fix race condition when IRQ is cleared")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoi2c: scmi: Fix probe error on devices with an empty SMB0001 ACPI device node
Hans de Goede [Wed, 21 Nov 2018 09:19:55 +0000 (10:19 +0100)]
i2c: scmi: Fix probe error on devices with an empty SMB0001 ACPI device node

Some AMD based HP laptops have a SMB0001 ACPI device node which does not
define any methods.

This leads to the following error in dmesg:

[    5.222731] cmi: probe of SMB0001:00 failed with error -5

This commit makes acpi_smbus_cmi_add() return -ENODEV instead in this case
silencing the error. In case of a failure of the i2c_add_adapter() call
this commit now propagates the error from that call instead of -EIO.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoi2c: axxia: properly handle master timeout
Adamski, Krzysztof (Nokia - PL/Wroclaw) [Fri, 16 Nov 2018 13:24:41 +0000 (13:24 +0000)]
i2c: axxia: properly handle master timeout

According to Intel (R) Axxia TM Lionfish Communication Processor
Peripheral Subsystem Hardware Reference Manual, the AXXIA I2C module
have a programmable Master Wait Timer, which among others, checks the
time between commands send in manual mode. When a timeout (25ms) passes,
TSS bit is set in Master Interrupt Status register and a Stop command is
issued by the hardware.

The axxia_i2c_xfer(), does not properly handle this situation, however.
For each message a separate axxia_i2c_xfer_msg() is called and this
function incorrectly assumes that any interrupt might happen only when
waiting for completion. This is mostly correct but there is one
exception - a master timeout can trigger if enough time has passed
between individual transfers. It will, by definition, happen between
transfers when the interrupts are disabled by the code. If that happens,
the hardware issues Stop command.

The interrupt indicating timeout will not be triggered as soon as we
enable them since the Master Interrupt Status is cleared when master
mode is entered again (which happens before enabling irqs) meaning this
error is lost and the transfer is continued even though the Stop was
issued on the bus. The subsequent operations completes without error but
a bogus value (0xFF in case of read) is read as the client device is
confused because aborted transfer. No error is returned from
master_xfer() making caller believe that a valid value was read.

To fix the problem, the TSS bit (indicating timeout) in Master Interrupt
Status register is checked before each transfer. If it is set, there was
a timeout before this transfer and (as described above) the hardware
already issued Stop command so the transaction should be aborted thus
-ETIMEOUT is returned from the master_xfer() callback. In order to be
sure no timeout was issued we can't just read the status just before
starting new transaction as there will always be a small window of time
(few CPU cycles at best) where this might still happen. For this reason
we have to temporally disable the timer before checking for TSS bit.
Disabling it will, however, clear the TSS bit so in order to preserve
that information, we have to read it in ISR so we have to ensure that
the TSS interrupt is not masked between transfers of one transaction.
There is no need to call bus recovery or controller reinitialization if
that happens so it's skipped.

Signed-off-by: Krzysztof Adamski <krzysztof.adamski@nokia.com>
Reviewed-by: Alexander Sverdlin <alexander.sverdlin@nokia.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
5 years agoMerge branch 'mlxsw-Various-fixes'
David S. Miller [Thu, 6 Dec 2018 21:31:09 +0000 (13:31 -0800)]
Merge branch 'mlxsw-Various-fixes'

Ido Schimmel says:

====================
mlxsw: Various fixes

Patches #1 and #2 fix two VxLAN related issues. The first patch removes
warnings that can currently be triggered from user space. Second patch
avoids leaking a FID in an error path.

Patch #3 fixes a too strict check that causes certain host routes not to
be promoted to perform GRE decapsulation in hardware.

Last patch avoids a use-after-free when deleting a VLAN device via an
ioctl when it is enslaved to a bridge. I have a patchset for net-next
that reworks this code and makes the driver more robust.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
5 years agomlxsw: spectrum_switchdev: Fix VLAN device deletion via ioctl
Ido Schimmel [Thu, 6 Dec 2018 17:44:53 +0000 (17:44 +0000)]
mlxsw: spectrum_switchdev: Fix VLAN device deletion via ioctl

When deleting a VLAN device using an ioctl the netdev is unregistered
before the VLAN filter is updated via ndo_vlan_rx_kill_vid(). It can
lead to a use-after-free in mlxsw in case the VLAN device is deleted
while being enslaved to a bridge.

The reason for the above is that when mlxsw receives the CHANGEUPPER
event, it wrongly assumes that the VLAN device is no longer its upper
and thus destroys the internal representation of the bridge port despite
the reference count being non-zero.

Fix this by checking if the VLAN device is our upper using its real
device. In net-next I'm going to remove this trick and instead make
mlxsw completely agnostic to the order of the events.

Fixes: c57529e1d5d8 ("mlxsw: spectrum: Replace vPorts with Port-VLAN")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>