A timing hazard exists when the network interface is stopped that
allows a watchdog timeout to be processed by a separate core in
parallel. This creates the potential for the timeout handler to
wake the queues while the driver is shutting down, or access
registers after their clocks have been removed.

The more common case is that the watchdog timeout will produce a
warning message which doesn't lead to a crash. The chances of this
are greatly increased by the fact that bcmgenet_netif_stop stops
the transmit queues which can easily precipitate a watchdog time-
out because of stale trans_start data in the queues.

This commit corrects the behavior by ensuring that the watchdog
timeout is disabled before enterring bcmgenet_netif_stop. There
are currently only two users of the bcmgenet_netif_stop function:
close and suspend.

The close case already handles the issue by exiting the RUNNING
state before invoking the driver close service.

The suspend case now performs the netif_device_detach to exit the
PRESENT state before the call to bcmgenet_netif_stop rather than
after it.

These behaviors prevent any future scheduling of the driver timeout
service during the window. The netif_tx_stop_all_queues function
in bcmgenet_netif_stop is replaced with netif_tx_disable to ensure
synchronization with any transmit or timeout threads that may
already be executing on other cores.

For symmetry, the netif_device_attach call upon resume is moved to
after the call to bcmgenet_netif_start. Since it wakes the transmit
queues it is not necessary to invoke netif_tx_start_all_queues from
bcmgenet_netif_start so it is moved into the driver open service.

Fixes: 1c1008c7 ("net: bcmgenet: add main driver file")
Signed-off-by: NDoug Berger <opendmb@gmail.com>
Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If the network becomes (partially) unavailable, say by disabling IPv6, the
background ACK transmission routine can get itself into a tizzy by
proposing immediate ACK retransmission.  Since we're in the call event
processor, that happens immediately without returning to the workqueue
manager.

The condition should clear after a while when either the network comes back
or the call times out.

Fix this by:

 (1) When re-proposing an ACK on failed Tx, don't schedule it immediately.
     This will allow a certain amount of time to elapse before we try
     again.

 (2) Enforce a return to the workqueue manager after a certain number of
     iterations of the call processing loop.

 (3) Add a backoff delay that increases the delay on deferred ACKs by a
     jiffy per failed transmission to a limit of HZ.  The backoff delay is
     cleared on a successful return from kernel_sendmsg().

 (4) Cancel calls immediately if the opening sendmsg fails.  The layer
     above can arrange retransmission or rotate to another server.

Fixes: 248f219c ("rxrpc: Rewrite the data and ack handling code")
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Initialize mutex before use.  Avoid kernel complaint when
CONFIG_DEBUG_LOCK_ALLOC is enabled.

Fixes: b987e98e ("dsa: add DSA switch driver for Microchip KSZ9477")
Signed-off-by: NTristram Ha <Tristram.Ha@microchip.com>
Reviewed-by: NPavel Machek <pavel@ucw.cz>
Reviewed-by: NAndrew Lunn <andrew@lunn.ch>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

icmp6_send() function is expensive on systems with a large number of
interfaces. Every time it’s called, it has to verify that the source
address does not correspond to an existing anycast address by looping
through every device and every anycast address on the device.  This can
result in significant delays for a CPU when there are a large number of
neighbors and ND timers are frequently timing out and calling
neigh_invalidate().

Add anycast addresses to a global hashtable to allow quick searching for
matching anycast addresses.  This is based on inet6_addr_lst in addrconf.c.
Signed-off-by: NJeff Barnhill <0xeffeff@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We need to make sure, that the carrier check polling is disabled
while suspending. Otherwise we can end up with usbnet_read_cmd()
being issued when only usbnet_read_cmd_nopm() is allowed. If this
happens, read operations lock up.

Fixes: d69d1694 ("usbnet: smsc95xx: fix link detection for disabled autonegotiation")
Signed-off-by: NFrieder Schrempf <frieder.schrempf@kontron.de>
Reviewed-by: NRaghuram Chary J <RaghuramChary.Jallipalli@microchip.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove kernel-doc warning:

  net/core/skbuff.c:4953: warning: Function parameter or member 'skb' not described in 'skb_gso_size_check'
Signed-off-by: NMathieu Malaterre <malat@debian.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Most callers of pskb_trim_rcsum() simply drop the skb when
it fails, however, ip_check_defrag() still continues to pass
the skb up to stack. This is suspicious.

In ip_check_defrag(), after we learn the skb is an IP fragment,
passing the skb to callers makes no sense, because callers expect
fragments are defrag'ed on success. So, dropping the skb when we
can't defrag it is reasonable.

Note, prior to commit 88078d98, this is not a big problem as
checksum will be fixed up anyway. After it, the checksum is not
correct on failure.

Found this during code review.

Fixes: 88078d98 ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: NCong Wang <xiyou.wangcong@gmail.com>
Reviewed-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pull i2c fixes from Wolfram Sang:
 "I2C has a core bugfix & cleanup as well as an ID addition and
  MAINTAINERS update for you"

* 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  MAINTAINERS: add maintainer for IMX LPI2C driver
  dt-bindings: i2c: i2c-imx-lpi2c: add imx8qxp compatible string
  i2c: Clear client->irq in i2c_device_remove
  i2c: Remove unnecessary call to irq_find_mapping

Pull percpu fixes from Dennis Zhou:
 "Two small things for v4.20.

  The first fixes a clang uninitialized variable warning for arm64 in
  the default path calls BUILD_BUG(). The second removes an unnecessary
  unlikely() in a WARN_ON() use"

* 'for-4.20' of git://git.kernel.org/pub/scm/linux/kernel/git/dennis/percpu:
  arm64: percpu: Initialize ret in the default case
  mm: percpu: remove unnecessary unlikely()

Pull networking fixes from David Miller:

 1) BPF verifier fixes from Daniel Borkmann.

 2) HNS driver fixes from Huazhong Tan.

 3) FDB only works for ethernet devices, reject attempts to install FDB
    rules for others. From Ido Schimmel.

 4) Fix spectre V1 in vhost, from Jason Wang.

 5) Don't pass on-stack object to irq_set_affinity_hint() in mvpp2
    driver, from Marc Zyngier.

 6) Fix mlx5e checksum handling when RXFCS is enabled, from Eric
    Dumazet.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (49 commits)
  openvswitch: Fix push/pop ethernet validation
  net: stmmac: Fix stmmac_mdio_reset() when building stmmac as modules
  bpf: test make sure to run unpriv test cases in test_verifier
  bpf: add various test cases to test_verifier
  bpf: don't set id on after map lookup with ptr_to_map_val return
  bpf: fix partial copy of map_ptr when dst is scalar
  libbpf: Fix compile error in libbpf_attach_type_by_name
  kselftests/bpf: use ping6 as the default ipv6 ping binary if it exists
  selftests: mlxsw: qos_mc_aware: Add a test for UC awareness
  selftests: mlxsw: qos_mc_aware: Tweak for min shaper
  mlxsw: spectrum: Set minimum shaper on MC TCs
  mlxsw: reg: QEEC: Add minimum shaper fields
  net: hns3: bugfix for rtnl_lock's range in the hclgevf_reset()
  net: hns3: bugfix for rtnl_lock's range in the hclge_reset()
  net: hns3: bugfix for handling mailbox while the command queue reinitialized
  net: hns3: fix incorrect return value/type of some functions
  net: hns3: bugfix for hclge_mdio_write and hclge_mdio_read
  net: hns3: bugfix for is_valid_csq_clean_head()
  net: hns3: remove unnecessary queue reset in the hns3_uninit_all_ring()
  net: hns3: bugfix for the initialization of command queue's spin lock
  ...

Pull sparc fixes from David Miller:
 "Two small fixes"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  sparc64: Wire up compat getpeername and getsockname.
  sparc64: Remvoe set_fs() from perf_callchain_user().

Pull csky dtb fixups from Guo Ren:
 "These fix the csky dtb Kbuild to follow the new Devicetree dtb build
  rules"

* tag 'csky-for-linus-4.20-fixup-dtb' of https://github.com/c-sky/csky-linux:
  csky: use common dtb build rules
  csky: remove builtin-dtb Kbuild

Pull x86 platform driver updates from Darren Hart:

 - Move the Dell dcdbas and dell_rbu drivers into platform/drivers/x86
   as they are closely coupled with other drivers in this location.

 - Improve _init* usage for acerhdf and fix some usage issues with
   messages and module parameters.

 - Simplify asus-wmi by calling ACPI/WMI methods directly, eliminating
   workqueue overhead, eliminate double reporting of keyboard backlight.

 - Fix wake from USB failure on Bay Trail devices (intel_int0002_vgpio).

 - Notify intel_telemetry users when IPC1 device is not enabled.

 - Update various drivers with new laptop model IDs.

 - Update several intel drivers to use SPDX identifers and order headers
   alphabetically.

* tag 'platform-drivers-x86-v4.20-1' of git://git.infradead.org/linux-platform-drivers-x86: (64 commits)
  HID: asus: only support backlight when it's not driven by WMI
  platform/x86: asus-wmi: export function for evaluating WMI methods
  platform/x86: asus-wmi: Only notify kbd LED hw_change by fn-key pressed
  platform/x86: wmi: declare device_type structure as constant
  platform/x86: ideapad: Add Y530-15ICH to no_hw_rfkill
  platform/x86: Add Intel AtomISP2 dummy / power-management driver
  platform/x86: touchscreen_dmi: Add min-x and min-y settings for various models
  platform/x86: touchscreen_dmi: Add info for the Onda V80 Plus v3 tablet
  platform/x86: touchscreen_dmi: Add info for the Trekstor Primetab T13B tablet
  platform/x86: intel_telemetry: Get rid of custom macro
  platform/x86: intel_telemetry: report debugfs failure
  MAINTAINERS: intel_telemetry: Update maintainers info
  platform/x86: Add LG Gram laptop special features driver
  platform/x86: asus-wmi: Simplify the keyboard brightness updating process
  platform/x86: touchscreen_dmi: Add info for the Trekstor Primebook C11 convertible
  platform/x86: mlx-platform: Properly use mlxplat_mlxcpld_msn201x_items
  MAINTAINERS: intel_pmc_core: Update MAINTAINERS
  firmware: dcdbas: include linux/io.h
  platform/x86: intel-wmi-thunderbolt: Add dynamic debugging
  platform/x86: intel-wmi-thunderbolt: Convert to use SPDX identifier
  ...

Remove the Kbuild rules in arch/csky and use common dtb build rules.

This modification is based on:

commit 37c8a5fa ("kbuild: consolidate Devicetree dtb build rules")
Signed-off-by: NGuo Ren <ren_guo@c-sky.com>
Reviewed-by: NRob Herring <robh@kernel.org>
Acked-by: NArnd Bergmann <arnd@arndb.de>

Remove the builtin-dtb implementation in arch/csky.
Signed-off-by: NGuo Ren <ren_guo@c-sky.com>
Reviewed-by: NRob Herring <robh@kernel.org>
Reviewed-by: NArnd Bergmann <arnd@arndb.de>

When there are both pop and push ethernet header actions among the
actions to be applied to a packet, an unexpected EINVAL (Invalid
argument) error is obtained. This is due to mac_proto not being reset
correctly when those actions are validated.

Reported-at:
https://mail.openvswitch.org/pipermail/ovs-discuss/2018-October/047554.html
Fixes: 91820da6 ("openvswitch: add Ethernet push and pop actions")
Signed-off-by: NJaime Caamaño Ruiz <jcaamano@suse.com>
Tested-by: NGreg Rose <gvrose8192@gmail.com>
Reviewed-by: NGreg Rose <gvrose8192@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When building stmmac, it is only possible to select CONFIG_DWMAC_GENERIC,
or any of the glue drivers, when CONFIG_STMMAC_PLATFORM is set.
The only exception is CONFIG_STMMAC_PCI.

When calling of_mdiobus_register(), it will call our ->reset()
callback, which is set to stmmac_mdio_reset().

Most of the code in stmmac_mdio_reset() is protected by a
"#if defined(CONFIG_STMMAC_PLATFORM)", which will evaluate
to false when CONFIG_STMMAC_PLATFORM=m.

Because of this, the phy reset gpio will only be pulled when
stmmac is built as built-in, but not when built as modules.

Fix this by using "#if IS_ENABLED()" instead of "#if defined()".
Signed-off-by: NNiklas Cassel <niklas.cassel@linaro.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes: 8b30ca73 ("sparc: Add all necessary direct socket system calls.")
Reported-by: NJoseph Myers  <joseph@codesourcery.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2018-10-31

This series contains a various collection of fixes.

Miroslav Lichvar from Red Hat or should I say IBM now?  Updates the PHC
timecounter interval for igb so that it gets updated at least once
every 550 seconds.

Ngai-Mint provides a fix for fm10k to prevent a soft lockup or system
crash by adding a new condition to determine if the SM mailbox is in the
correct state before proceeding.

Jake provides several fm10k fixes, first one marks complier aborts as
non-fatal since on some platforms trigger machine check errors when the
compile aborts.  Added missing device ids to the in-kernel driver.  Due
to the recent fixes, bumped the driver version.

I (Jeff Kirsher) fixed a XFRM_ALGO dependency for both ixgbe and
ixgbevf.  This fix was based on the original work from Arnd Bergmann,
which only fixed ixgbe.

Mitch provides a fix for i40e/avf to update the status codes, which
resolves an issue between a mis-match between i40e and the iavf driver,
which also supports the ice LAN driver.

Radoslaw fixes the ixgbe where the driver is logging a message about
spoofed packets detected when the VF is re-started with a different MAC
address.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
pull-request: bpf 2018-11-01

The following pull-request contains BPF updates for your *net* tree.

The main changes are:

1) Fix tcp_bpf_recvmsg() to return -EAGAIN instead of 0 in non-blocking
   case when no data is available yet, from John.

2) Fix a compilation error in libbpf_attach_type_by_name() when compiled
   with clang 3.8, from Andrey.

3) Fix a partial copy of map pointer on scalar alu and remove id
   generation for RET_PTR_TO_MAP_VALUE return types, from Daniel.

4) Add unlimited memlock limit for kernel selftest's flow_dissector_load
   program, from Yonghong.

5) Fix ping for some BPF shell based kselftests where distro does not
   ship "ping -6" anymore, from Li.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
The series contains two fixes in BPF core and test cases. For details
please see individual patches. Thanks!
====================
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

Right now unprivileged tests are never executed as a BPF test run,
only loaded. Allow for running them as well so that we can check
the outcome and probe for regressions.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

Add some more map related test cases to test_verifier kselftest
to improve test coverage. Summary: 1012 PASSED, 0 SKIPPED, 0 FAILED
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

In the verifier there is no such semantics where registers with
PTR_TO_MAP_VALUE type have an id assigned to them. This is only
used in PTR_TO_MAP_VALUE_OR_NULL and later on nullified once the
test against NULL has been pattern matched and type transformed
into PTR_TO_MAP_VALUE.

Fixes: 3e6a4b3e ("bpf/verifier: introduce BPF_PTR_TO_MAP_VALUE")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Cc: Roman Gushchin <guro@fb.com>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

ALU operations on pointers such as scalar_reg += map_value_ptr are
handled in adjust_ptr_min_max_vals(). Problem is however that map_ptr
and range in the register state share a union, so transferring state
through dst_reg->range = ptr_reg->range is just buggy as any new
map_ptr in the dst_reg is then truncated (or null) for subsequent
checks. Fix this by adding a raw member and use it for copying state
over to dst_reg.

Fixes: f1174f77 ("bpf/verifier: rework value tracking")
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Cc: Edward Cree <ecree@solarflare.com>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

Merge tag 'tag-chrome-platform-for-v4.20' of git://git.kernel.org/pub/scm/linux/kernel/git/bleung/chrome-platform

Pull chrome-platform updates from Benson Leung:

 - Move mfd/cros_ec_lpc* includes to drivers/platform from mfd

 - Adding a new interrupt path for cros_ec_lpc

* tag 'tag-chrome-platform-for-v4.20' of git://git.kernel.org/pub/scm/linux/kernel/git/bleung/chrome-platform:
  platform/chrome: chromeos_tbmc - Remove unneeded const
  platform/chrome: Add a new interrupt path for cros_ec_lpc
  mfd: cros_ec: Fix and improve kerneldoc comments.
  platform/chrome: Move mfd/cros_ec_lpc* includes to drivers/platform.

The LPI2C is used in IMX7ULP/MX8 SoCs.
Signed-off-by: NDong Aisheng <aisheng.dong@nxp.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Add imx8qxp compatible string
Reviewed-by: NRob Herring <robh@kernel.org>
Signed-off-by: NDong Aisheng <aisheng.dong@nxp.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

The IRQ will be mapped in i2c_device_probe only if client->irq is zero and
i2c_device_remove does not clear this. When rebinding an I2C device,
whos IRQ provider has also been rebound this means that an IRQ mapping
will never be created, causing the I2C device to fail to acquire its
IRQ. Fix this issue by clearing client->irq in i2c_device_remove,
forcing i2c_device_probe to lookup the mapping again.
Signed-off-by: NCharles Keepax <ckeepax@opensource.cirrus.com>
Reviewed-by: NBenjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

irq_create_mapping calls irq_find_mapping internally and will use the
found mapping if one exists, so there is no need to manually call this
from i2c_smbus_host_notify_to_irq.
Signed-off-by: NCharles Keepax <ckeepax@opensource.cirrus.com>
Reviewed-by: NBenjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: NWolfram Sang <wsa@the-dreams.de>

Merge tag 'riscv-for-linus-4.20-mw2' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux

Pull more RISC-V updates from Palmer Dabbelt:
 "This contains the follow-on patches I'd like to target for the 4.20
  merge window. I'm being somewhat conservative here, as while there are
  a few patches on the mailing list that were posted early in the merge
  window I'd like to let those bake for another round -- this was a
  fairly big release as far as RISC-V is concerened, and we need to walk
  before we can run.

  As far as the patches that made it go:

   - A patch to ignore offline CPUs when calculating AT_HWCAP. This
     should fix GDB on the HiFive unleashed, which has an embedded core
     for hart 0 which is exposed to Linux as an offline CPU.

   - A move of EM_RISCV to elf-em.h, which is where it should have been
     to begin with.

   - I've also removed the 64-bit divide routines. I know I'm not really
     playing by my own rules here because I posted the patches this
     morning, but since they shouldn't be in the kernel I think it's
     better to err on the side of going too fast here.

  I don't anticipate any more patch sets for the merge window"

* tag 'riscv-for-linus-4.20-mw2' of git://git.kernel.org/pub/scm/linux/kernel/git/palmer/riscv-linux:
  Move EM_RISCV into elf-em.h
  RISC-V: properly determine hardware caps
  Revert "lib: Add umoddi3 and udivmoddi4 of GCC library routines"
  Revert "RISC-V: Select GENERIC_LIB_UMODDI3 on RV32"

Pull UML updates from Richard Weinberger:

 - removal of old and dead code

 - a bug fix for our tty driver

 - other minor cleanups across the code base

* 'for-linus-4.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/uml:
  um: Make line/tty semantics use true write IRQ
  um: trap: fix spelling mistake, EACCESS -> EACCES
  um: Don't hardcode path as it is architecture dependent
  um: NULL check before kfree is not needed
  um: remove unused AIO code
  um: Give start_idle_thread() a return code
  um: Remove update_debugregs()
  um: Drop own definition of PTRACE_SYSEMU/_SINGLESTEP

Pull c6x update from Mark Salter.

* tag 'for-linus' of git://linux-c6x.org/git/projects/linux-c6x-upstreaming:
  c6x: switch to NO_BOOTMEM

Arnaldo Carvalho de Melo reported build error in libbpf when clang
version 3.8.1-24 (tags/RELEASE_381/final) is used:

libbpf.c:2201:36: error: comparison of constant -22 with expression of
type 'const enum bpf_attach_type' is always false
[-Werror,-Wtautological-constant-out-of-range-compare]
                if (section_names[i].attach_type == -EINVAL)
                    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^  ~~~~~~~
1 error generated.

Fix the error by keeping "is_attachable" property of a program in a
separate struct field instead of trying to use attach_type itself.

Fixes: 956b620f ("libbpf: Introduce libbpf_attach_type_by_name")
Reported-by: NArnaldo Carvalho de Melo <acme@kernel.org>
Signed-off-by: NAndrey Ignatov <rdna@fb.com>
Tested-by: NArnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

ping binary on some distros doesn't support "ping -6" anymore.
Signed-off-by: NLi Zhijian <lizhijian@cn.fujitsu.com>
Acked-by: NSong Liu <songliubraving@fb.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Pull fuse updates from Miklos Szeredi:
 "As well as the usual bug fixes, this adds the following new features:

   - cached readdir and readlink

   - max I/O size increased from 128k to 1M

   - improved performance and scalability of request queues

   - copy_file_range support

  The only non-fuse bits are trivial cleanups of macros in
  <linux/bitops.h>"

* tag 'fuse-update-4.20' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse: (31 commits)
  fuse: enable caching of symlinks
  fuse: only invalidate atime in direct read
  fuse: don't need GETATTR after every READ
  fuse: allow fine grained attr cache invaldation
  bitops: protect variables in bit_clear_unless() macro
  bitops: protect variables in set_mask_bits() macro
  fuse: realloc page array
  fuse: add max_pages to init_out
  fuse: allocate page array more efficiently
  fuse: reduce size of struct fuse_inode
  fuse: use iversion for readdir cache verification
  fuse: use mtime for readdir cache verification
  fuse: add readdir cache version
  fuse: allow using readdir cache
  fuse: allow caching readdir
  fuse: extract fuse_emit() helper
  fuse: add FOPEN_CACHE_DIR
  fuse: split out readdir.c
  fuse: Use hash table to link processing request
  fuse: kill req->intr_unique
  ...

Pull ceph updates from Ilya Dryomov:
 "The highlights are:

   - a series that fixes some old memory allocation issues in libceph
     (myself). We no longer allocate memory in places where allocation
     failures cannot be handled and BUG when the allocation fails.

   - support for copy_file_range() syscall (Luis Henriques). If size and
     alignment conditions are met, it leverages RADOS copy-from
     operation. Otherwise, a local copy is performed.

   - a patch that reduces memory requirement of ceph_sync_read() from
     the size of the entire read to the size of one object (Zheng Yan).

   - fallocate() syscall is now restricted to FALLOC_FL_PUNCH_HOLE (Luis
     Henriques)"

* tag 'ceph-for-4.20-rc1' of git://github.com/ceph/ceph-client: (25 commits)
  ceph: new mount option to disable usage of copy-from op
  ceph: support copy_file_range file operation
  libceph: support the RADOS copy-from operation
  ceph: add non-blocking parameter to ceph_try_get_caps()
  libceph: check reply num_data_items in setup_request_data()
  libceph: preallocate message data items
  libceph, rbd, ceph: move ceph_osdc_alloc_messages() calls
  libceph: introduce alloc_watch_request()
  libceph: assign cookies in linger_submit()
  libceph: enable fallback to ceph_msg_new() in ceph_msgpool_get()
  ceph: num_ops is off by one in ceph_aio_retry_work()
  libceph: no need to call osd_req_opcode_valid() in osd_req_encode_op()
  ceph: set timeout conditionally in __cap_delay_requeue
  libceph: don't consume a ref on pagelist in ceph_msg_data_add_pagelist()
  libceph: introduce ceph_pagelist_alloc()
  libceph: osd_req_op_cls_init() doesn't need to take opcode
  libceph: bump CEPH_MSG_MAX_DATA_LEN
  ceph: only allow punch hole mode in fallocate
  ceph: refactor ceph_sync_read()
  ceph: check if LOOKUPNAME request was aborted when filling trace
  ...

Ido Schimmel says:

====================
mlxsw: Enable minimum shaper on MC TCs

Petr says:

An MC-aware mode was introduced in commit 7b819530 ("mlxsw:
spectrum: Configure MC-aware mode on mlxsw ports"). In MC-aware mode,
BUM traffic gets a special treatment by being assigned to a separate set
of traffic classes 8..15. Pairs of TCs 0 and 8, 1 and 9, etc., are then
configured to strictly prioritize the lower-numbered ones. The intention
is to prevent BUM traffic from flooding the switch and push out all UC
traffic, which would otherwise happen, and instead give UC traffic
precedence.

However strictly prioritizing UC traffic has the effect that UC overload
pushes out all BUM traffic, such as legitimate ARP queries. These
packets are kept in queues for a while, but under sustained UC overload,
their lifetime eventually expires and these packets are dropped. That is
detrimental to network performance as well.

In this patchset, MC TCs (8..15) are configured with minimum shaper of
200Mbps (a minimum permitted value) to allow a trickle of necessary
control traffic to get through.

First in patch #1, the QEEC register is extended with fields necessary
to configure the minimum shaper.

In patch #2, minimum shaper is enabled on TCs 8..15.

In patches #3 and #4, first the MC-awareness test is tweaked to support
the minimum shaper, and then a new test is introduced to test that MC
traffic behaves well under UC overload.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In a previous patch, mlxsw was updated to configure a minimum bandwidth
allowance on MC TCs. Test that this indeed fixes the problem of UC
traffic overload pushing out all MC traffic.

Fixes: b5638d46 ("selftests: mlxsw: Add a test for UC behavior under MC flood")
Signed-off-by: NPetr Machata <petrm@mellanox.com>
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since the minimum shaper is now being enabled for MC TCs, it's
unreasonable to expect no UC traffic loss. Minimal min shaper value is
200Mbps, which is 20% of the 1Gbps that this test configures on egress.
To cover for glitches, tolerate up to 25% UC degradation under MC
overload.

Fixes: b5638d46 ("selftests: mlxsw: Add a test for UC behavior under MC flood")
Signed-off-by: NPetr Machata <petrm@mellanox.com>
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>