net/mlx5: E-Switch, Return eswitch max ports when eswitch is supported

mlx5_eswitch_get_total_vports() doesn't honor MLX5_ESWICH Kconfig flag.

When MLX5_ESWITCH is disabled, FS layer continues to initialize eswitch
specific ACL namespaces.
Instead, start honoring MLX5_ESWITCH flag and perform vport specific
initialization only when vport count is non zero.

Signed-off-by: Parav Pandit <parav@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Vu Pham <vuhuong@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>

phy: nxp-c45-tja11xx: add interrupt support

Added .config_intr and .handle_interrupt callbacks.

Link event interrupt will trigger an interrupt every time when the link
goes up or down.

Signed-off-by: Radu Pirea (NXP OSS) <radu-nicolae.pirea@oss.nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/atm: Fix spelling mistake "requed" -> "requeued"

There is a spelling mistake in a printk message. Fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests/net: bump timeout to 5 minutes

We found that with the latest mainline kernel (5.12.0-051200rc8) on
some KVM instances / bare-metal systems, the following tests will take
longer than the kselftest framework default timeout (45 seconds) to
run and thus got terminated with TIMEOUT error:
* xfrm_policy.sh - took about 2m20s
* pmtu.sh - took about 3m5s
* udpgso_bench.sh - took about 60s

Bump the timeout setting to 5 minutes to allow them have a chance to
finish.

https://bugs.launchpad.net/bugs/1856010
Signed-off-by: Po-Hsu Lin <po-hsu.lin@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'mptcp-msg-flags'

Mat Martineau says:

====================
mptcp: Compatibility with common msg flags

These patches from the MPTCP tree handle some of the msg flags that are
typically used with TCP, to make it easier to adapt userspace programs
for use with MPTCP.

Patches 1, 2, and 4 add support for MSG_ERRQUEUE (no-op for now),
MSG_TRUNC, and MSG_PEEK on the receive side.

Patch 3 ignores unsupported msg flags for send and receive.

Patch 5 adds a selftest for MSG_PEEK.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: mptcp: add a test case for MSG_PEEK

Extend mptcp_connect tool with MSG_PEEK support and add a test case in
mptcp_connect.sh that checks the data received from/after recv() with
MSG_PEEK.

Acked-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Yonglong Li <liyonglong@chinatelecom.cn>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mptcp: add MSG_PEEK support

This patch adds support for MSG_PEEK flag. Packets are not removed
from the receive_queue if MSG_PEEK set in recv() system call.

Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Yonglong Li <liyonglong@chinatelecom.cn>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mptcp: ignore unsupported msg flags

Currently mptcp_sendmsg() fails with EOPNOTSUPP if the
user-space provides some unsupported flag. That is unexpected
and may foul existing applications migrated to MPTCP, which
expect a different behavior.

Change the mentioned function to silently ignore the unsupported
flags except MSG_FASTOPEN. This is the only flags currently not
supported by MPTCP with user-space visible side-effects.

Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/162
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mptcp: implement MSG_TRUNC support

The mentioned flag is currently silenlty ignored. This
change implements the TCP-like behaviour, dropping the
pending data up to the specified length.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Sigend-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mptcp: implement dummy MSG_ERRQUEUE support

mptcp_recvmsg() currently silently ignores MSG_ERRQUEUE, returning
input data instead of error cmsg.

This change provides a dummy implementation for MSG_ERRQUEUE - always
returns no data. That is consistent with the current lack of a suitable
IP_RECVERR setsockopt() support.

Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
40GbE Intel Wired LAN Driver Updates 2021-04-23

This series contains updates to i40e and iavf drivers.

Aleksandr adds support for VIRTCHNL_VF_CAP_ADV_LINK_SPEED in i40e which
allows for reporting link speed to VF as a value instead of using an
enum; helper functions are created to remove repeated code.

Coiby Xu reduces memory use of i40e when using kdump by reducing Tx, Rx,
and admin queue to minimum values. Current use causes failure of kdump.

Stefan Assmann removes duplicated free calls in iavf.

Haiyue cleans up a loop to return directly when if the value is found
and changes some magic numbers to defines for better maintainability
in iavf.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'mlxsw-selftest-fixes'

Petr Machata says:

====================
selftests: mlxsw: Fixes

This patch set carries fixes to selftest issues that we have hit in our
nightly regression run. Almost all are in mlxsw selftests, though one is in
a generic forwarding selftest.

- In patch #1, in an ERSPAN test, install an FDB entry as static instead of
  (implicitly) as local.

- In the mlxsw resource-scale test, an if statement overrides the value of
  $?, which is supposed to contain the result of the test. As a result, the
  resource scale test can spuriously pass.

  In patches #2 and #3, remove the if statements to fix the issue in,
  respectively, port_scale test and tc_flower_scale tests.

- Again in the mlxsw resource-scale test, when more then one sub-test is
  run, a successful sub-test overrides any previous failures. This causes a
  spurious pass of the overall test. This is fixed in patch #4.

- In patch #5, increase a tolerance in a mlxsw-specific RED backlog test.
  This test is very noisy, due to rounding errors and the unpredictability
  of software traffic generation. By bumping the tolerance from 5 % to 10,
  get the failure rate to zero. This shouldn't impact the accuracy,
  mistakes in backlog configuration (e.g. due to wrong cell size) are
  likely to cause a much larger discrepancy.

- In patch #6, fix mausezahn invocation in the mlxsw ERSPAN scale
  test. The test failed because of the wrong invocation.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: mlxsw: Fix mausezahn invocation in ERSPAN scale test

The mirror_gre_scale test creates as many ERSPAN sessions as the underlying
chip supports, and tests that they all work. In order to determine that it
issues a stream of ICMP packets and checks if they are mirrored as
expected.

However, the mausezahn invocation missed the -6 flag to identify the use of
IPv6 protocol, and was sending ICMP messages over IPv6, as opposed to
ICMP6. It also didn't pass an explicit source IP address, which apparently
worked at some point in the past, but does not anymore.

To fix these issues, extend the function mirror_test() in mirror_lib by
detecting the IPv6 protocol addresses, and using a different ICMP scheme.
Fix __mirror_gre_test() in the selftest itself to pass a source IP address.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: mlxsw: Increase the tolerance of backlog buildup

The intention behind this test is to make sure that qdisc limit is
correctly projected to the HW. However, first, due to rounding in the
qdisc, and then in the driver, the number cannot actually be accurate. And
second, the approach to testing this is to oversubscribe the port with
traffic generated on the same switch. The actual backlog size therefore
fluctuates.

In practice, this test proved to be noisier than the rest, and spuriously
fails every now and then. Increase the tolerance to 10 % to avoid these
issues.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Acked-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: mlxsw: Return correct error code in resource scale tests

Currently, the resource scale test checks a few cases, when the error code
resets between the cases. So for example, if one case fails and the
consecutive case passes, the error code eventually will fit the last test
and will be 0.

Save a new return code that will hold the 'or' return codes of all the
cases, so the final return code will consider all the cases.

Signed-off-by: Danielle Ratson <danieller@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: mlxsw: Remove a redundant if statement in tc_flower_scale test

Currently, the error return code of the failure condition is lost after
using an if statement, so the test doesn't fail when it should.

Remove the if statement that separates the condition and the error code
check, so the test won't always pass.

Fixes: abfce9e062021 ("selftests: mlxsw: Reduce running time using offload indication")
Reported-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Danielle Ratson <danieller@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: mlxsw: Remove a redundant if statement in port_scale test

Currently, the error return code of the failure condition is lost after
using an if statement, so the test doesn't fail when it should.

Remove the if statement that separates the condition and the error code
check, so the test won't always pass.

Fixes: 5154b1b826d9b ("selftests: mlxsw: Add a scale test for physical ports")
Reported-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Danielle Ratson <danieller@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

selftests: net: mirror_gre_vlan_bridge_1q: Make an FDB entry static

The FDB roaming test installs a destination MAC address on the wrong
interface of an FDB database and tests whether the mirroring fails, because
packets are sent to the wrong port. The test by mistake installs the FDB
entry as local. This worked previously, because drivers were notified of
local FDB entries in the same way as of static entries. However that has
been fixed in the commit 6ab4c3117aec ("net: bridge: don't notify switchdev
for local FDB addresses"), and local entries are not notified anymore. As a
result, the HW is not reconfigured for the FDB roam, and mirroring keeps
working, failing the test.

To fix the issue, mark the FDB entry as static.

Fixes: 9c7c8a82442c ("selftests: forwarding: mirror_gre_vlan_bridge_1q: Add more tests")
Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'wireless-drivers-next-2021-04-23' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers-next

Kalle Valo says:

====================
wireless-drivers-next patches for v5.13

Third, and final, set of patches for v5.13. We got one more week
before the merge window and this includes from that extra week.
Smaller features to rtw88 and mt76, but mostly this contains fixes.

rtw88

* 8822c: Add gap-k calibration to improve long range performance

mt76

* parse rate power limits from DT

* debugfs file to test firmware crash

* debugfs to disable NAPI threaded mode
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'r8152-adjust-REALTEK_USB_DEVICE'

Hayes Wang says:

====================
r8152: adjust REALTEK_USB_DEVICE

Modify REALTEK_USB_DEVICE macro.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: redefine REALTEK_USB_DEVICE macro

Redefine REALTEK_USB_DEVICE macro with USB_DEVICE_INTERFACE_CLASS and
USB_DEVICE_AND_INTERFACE_INFO to simply the code.

Although checkpatch.pl shows the following error, it is more readable.

ERROR: Macros with complex values should be enclosed in parentheses

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: remove NCM mode from REALTEK_USB_DEVICE macro

The RTL8156 support CDC NCM mode. And users could set the configuration
of the USB device between vendor and NCM mode dynamically by themselves.
That is, the driver doesn't need to set vendor mode from NCM mode.

Fixes: 195aae321c82 ("r8152: support new chips")
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

enetc: fix locking for one-step timestamping packet transfer

The previous patch to support PTP Sync packet one-step timestamping
described one-step timestamping packet handling logic as below in
commit message:

- Trasmit packet immediately if no other one in transfer, or queue to
  skb queue if there is already one in transfer.
  The test_and_set_bit_lock() is used here to lock and check state.
- Start a work when complete transfer on hardware, to release the bit
  lock and to send one skb in skb queue if has.

There was not problem of the description, but there was a mistake in
implementation. The locking/test_and_set_bit_lock() should be put in
enetc_start_xmit() which may be called by worker, rather than in
enetc_xmit(). Otherwise, the worker calling enetc_start_xmit() after
bit lock released is not able to lock again for transfer.

Fixes: 7294380c5211 ("enetc: support PTP Sync packet one-step timestamping")
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Reviewed-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec-next

Steffen Klassert says:

====================
pull request (net-next): ipsec-next 2021-04-23

1) The SPI flow key in struct flowi has no consumers,
   so remove it. From Florian Westphal.

2) Remove stray synchronize_rcu from xfrm_init.
   From Florian Westphal.

3) Use the new exit_pre hook to reset the netlink socket
   on net namespace destruction. From Florian Westphal.

4) Remove an unnecessary get_cpu() in ipcomp, that
   code is always called with BHs off.
   From Sabrina Dubroca.

Please pull or let me know if there are problems.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'mk_eth_soc_fixes-perf-improvements'

Ilya Lipnitskiy says:

====================
mtk_eth_soc: fixes and performance improvements

Most of these changes come from OpenWrt where they have been present and
tested for months.

First three patches are bug fixes. The rest are performance
improvements. The last patch is a cleanup to use the iopoll.h macro for
busy-waiting instead of a custom loop.

v2:
 - Reverse christmas tree in "use iopoll.h macro for DMA init"
 - Use cond_resched() instead of iopoll.h macro in "reduce MDIO bus
   access latency"
 - Use napi_complete_done and rework NAPI callbacks in a new patch
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: use iopoll.h macro for DMA init

Replace a tight busy-wait loop without a pause with a standard
readx_poll_timeout_atomic routine with a 5 us poll period.

Tested by booting a MT7621 device to ensure the driver initializes
properly.

Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: set PPE flow hash as skb hash if present

This improves GRO performance

Signed-off-by: Felix Fietkau <nbd@nbd.name>
[Ilya: Use MTK_RXD4_FOE_ENTRY instead of GENMASK(13, 0)]
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: rework NAPI callbacks

Use napi_complete_done to communicate total TX and RX work done to NAPI.
Count total RX work up instead of remaining work down for clarity.
Remove unneeded local variables for clarity. Use do {} while instead of
goto for clarity.

Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: reduce unnecessary interrupts

Avoid rearming interrupt if napi_complete returns false

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: only read the full RX descriptor if DMA is done

Uncached memory access is expensive, and there is no need to access all
descriptor words if we can't process them anyway

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: cache HW pointer of last freed TX descriptor

The value is only updated by the CPU, so it is cheaper to access from the
ring data structure than from a hardware register.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: implement dynamic interrupt moderation

Reduces the number of interrupts under load

Signed-off-by: Felix Fietkau <nbd@nbd.name>
[Ilya: add documentation for new struct fields]
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: increase DMA ring sizes

256 descriptors is not enough for multi-gigabit traffic under load on
MT7622. Bump it to 512 to improve performance.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: use larger burst size for QDMA TX

Improves tx performance

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: remove unnecessary TX queue stops

When running short on descriptors, only stop the queue for the netdev that
tx was attempted for. By the time something tries to send on the other
netdev, the ring might have some more room already.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: reduce MDIO bus access latency

usleep_range often ends up sleeping much longer than the 10-20us provided
as a range here. This causes significant latency in mdio bus acceses,
which easily adds multiple seconds to the boot time on MT7621 when polling
DSA slave ports.
Use cond_resched instead of usleep_range, since the MDIO access does not
take much time

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: use napi_consume_skb

Should improve performance, since it can use bulk free

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: fix build_skb cleanup

In case build_skb fails, call skb_free_frag on the correct pointer. Also
update the DMA structures with the new mapping before exiting, because
the mapping was successful

Suggested-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: unmap RX data before calling build_skb

Since build_skb accesses the data area (for initializing shinfo), dma unmap
needs to happen before that call

Signed-off-by: Felix Fietkau <nbd@nbd.name>
[Ilya: split build_skb cleanup fix into a separate commit]
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: mtk_eth_soc: fix RX VLAN offload

The VLAN ID in the rx descriptor is only valid if the RX_DMA_VTAG bit is
set. Fixes frames wrongly marked with VLAN tags.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
[Ilya: fix commit message]
Signed-off-by: Ilya Lipnitskiy <ilya.lipnitskiy@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: mana: Use int to check the return value of mana_gd_poll_cq()

mana_gd_poll_cq() may return -1 if an overflow error is detected (this
should never happen unless there is a bug in the driver or the hardware).

Fix the type of the variable "comp_read" by using int rather than u32.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure Network Adapter (MANA)")
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

virtio-net: fix use-after-free in skb_gro_receive

When "headroom" > 0, the actual allocated memory space is the entire
page, so the address of the page should be used when passing it to
build_skb().

BUG: KASAN: use-after-free in skb_gro_receive (net/core/skbuff.c:4260)
Write of size 16 at addr ffff88811619fffc by task kworker/u9:0/534
CPU: 2 PID: 534 Comm: kworker/u9:0 Not tainted 5.12.0-rc7-custom-16372-gb150be05b806 #3382
Hardware name: QEMU MSN2700, BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
Workqueue: xprtiod xs_stream_data_receive_workfn [sunrpc]
Call Trace:
 <IRQ>
dump_stack (lib/dump_stack.c:122)
print_address_description.constprop.0 (mm/kasan/report.c:233)
kasan_report.cold (mm/kasan/report.c:400 mm/kasan/report.c:416)
skb_gro_receive (net/core/skbuff.c:4260)
tcp_gro_receive (net/ipv4/tcp_offload.c:266 (discriminator 1))
tcp4_gro_receive (net/ipv4/tcp_offload.c:316)
inet_gro_receive (net/ipv4/af_inet.c:1545 (discriminator 2))
dev_gro_receive (net/core/dev.c:6075)
napi_gro_receive (net/core/dev.c:6168 net/core/dev.c:6198)
receive_buf (drivers/net/virtio_net.c:1151) virtio_net
virtnet_poll (drivers/net/virtio_net.c:1415 drivers/net/virtio_net.c:1519) virtio_net
__napi_poll (net/core/dev.c:6964)
net_rx_action (net/core/dev.c:7033 net/core/dev.c:7118)
__do_softirq (./arch/x86/include/asm/jump_label.h:25 ./include/linux/jump_label.h:200 ./include/trace/events/irq.h:142 kernel/softirq.c:346)
irq_exit_rcu (kernel/softirq.c:221 kernel/softirq.c:422 kernel/softirq.c:434)
common_interrupt (arch/x86/kernel/irq.c:240 (discriminator 14))
</IRQ>

Fixes: fb32856b16ad ("virtio-net: page_to_skb() use build_skb when there's sufficient tailroom")
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Reported-by: Ido Schimmel <idosch@nvidia.com>
Tested-by: Ido Schimmel <idosch@nvidia.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sock: remove the unnecessary check in proto_register

tw_prot_cleanup will check the twsk_prot.

Fixes: 0f5907af3913 ("net: Fix potential memory leak in proto_register()")
Cc: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

iavf: redefine the magic number for FDIR GTP-U header fields

The flex-byte for GTP-U protocol header fields uses the magic number,
which is hard to maintain and understand, define the interested fields
with meaningful macro name, based on the GTP-U protocol stack:

GTP-U header
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     | 0x1 |1|0|1|0|0|     0xff      |           Length              |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                           TEID = 1654                         |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |    Sequence Number = 0        |N-PDU Number=0 |NextExtHdr=0x85|
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

GTP-U Extension Header (PDU Session Container)
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |  ExtHdrLen=2  |Type=0 | Spare |0|0|   QFI     | PPI |  Spare  |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     |                    Padding                    |NextExtHdr=0x0 |
     +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: enhance the duplicated FDIR list scan handling

When the FDIR entry is found, just return the result directly to break
the loop.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: change the flex-byte support number to macro definition

The maximum number (2) of flex-byte support is derived from ethtool
use-def data size (8 byte).

Change the magic number 2 to macro definition, and add the comment to
track the design thinking, so the code is clear and easily maintained.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: remove duplicate free resources calls

Both iavf_free_all_tx_resources() and iavf_free_all_rx_resources() have
already been called in the very same function.
Remove the duplicate calls.

Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

i40e: use minimal admin queue for kdump

The minimum size of admin send/receive queue is 1 and 2 respectively.
The admin send queue can't be set to 1 because in that case, the
firmware would fail to init.

Signed-off-by: Coiby Xu <coxu@redhat.com>
Tested-by: Dave Switzer <david.switzer@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

i40e: use minimal Rx and Tx ring buffers for kdump

Use the minimum of the number of descriptors thus we will allocate the
minimal ring buffers for kdump.

Signed-off-by: Coiby Xu <coxu@redhat.com>
Tested-by: Dave Switzer <david.switzer@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

i40e: use minimal Tx and Rx pairs for kdump

Set the number of the MSI-X vectors to 1. When MSI-X is enabled,
it's not allowed to use more TC queue pairs than MSI-X vectors
(pf->num_lan_msix) exist. Thus the number of Tx and Rx pairs
(vsi->num_queue_pairs) will be equal to the number of MSI-X vectors,
i.e., 1.

Signed-off-by: Coiby Xu <coxu@redhat.com>
Tested-by: Dave Switzer <david.switzer@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

i40e: refactor repeated link state reporting code

Refactor repeated link state reporting code into a separate helper
functions: i40e_set_vf_link_state() i40e_vc_link_speed2mbps().
Add support of VIRTCHNL_VF_CAP_ADV_LINK_SPEED;

Signed-off-by: Arkadiusz Kubalewski <arkadiusz.kubalewski@intel.com>
Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

Merge branch 'stmmac-swmac-desc-prefetch'

Mohammad Athari Bin Ismail says:

====================
Enable DWMAC HW descriptor prefetch

This patch series to add setting for HW descriptor prefetch for DWMAC
version 5.20 onwards. For Intel platform, enable the capability by
default.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: intel: Enable HW descriptor prefetch by default

Enable HW descriptor prefetch by default by setting plat->dma_cfg->dche =
true in intel_mgbe_common_data(). Need to be noted that this capability
only be supported in DWMAC core version 5.20 onwards. In stmmac, there is
a checking to check the core version. If the core version is below 5.20,
this capability wouldn`t be configured.

Below is the iperf result comparison between HW descriptor prefetch
disabled(DCHE=0b) and enabled(DCHE=1b). Tested on Intel Elkhartlake
platform with DWMAC Core 5.20. Observed line rate performance
improvement with HW descriptor prefetch enabled.

DCHE = 0b
[  5] local 169.254.1.162 port 42123 connected to 169.254.244.142 port 5201
[ ID] Interval           Transfer     Bitrate         Total Datagrams
[  5]   0.00-1.00   sec  96.7 MBytes   811 Mbits/sec  70050
[  5]   1.00-2.00   sec  96.5 MBytes   809 Mbits/sec  69850
[  5]   2.00-3.00   sec  96.3 MBytes   808 Mbits/sec  69720
[  5]   3.00-4.00   sec  95.9 MBytes   804 Mbits/sec  69450
[  5]   4.00-5.00   sec  96.0 MBytes   806 Mbits/sec  69530
[  5]   5.00-6.00   sec  96.8 MBytes   812 Mbits/sec  70080
[  5]   6.00-7.00   sec  96.9 MBytes   813 Mbits/sec  70140
[  5]   7.00-8.00   sec  96.8 MBytes   812 Mbits/sec  70080
[  5]   8.00-9.00   sec  97.0 MBytes   814 Mbits/sec  70230
[  5]   9.00-10.00  sec  96.9 MBytes   813 Mbits/sec  70170
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Jitter    Lost/Total Datagrams
[  5]   0.00-10.00  sec   966 MBytes   810 Mbits/sec  0.000 ms  0/699300 (0%)  sender
[  5]   0.00-10.00  sec   966 MBytes   810 Mbits/sec  0.011 ms  0/699265 (0%)  receiver

DCHE = 1b
[  5] local 169.254.1.162 port 49740 connected to 169.254.244.142 port 5201
[ ID] Interval           Transfer     Bitrate         Total Datagrams
[  5]   0.00-1.00   sec  97.9 MBytes   821 Mbits/sec  70880
[  5]   1.00-2.00   sec  98.1 MBytes   823 Mbits/sec  71060
[  5]   2.00-3.00   sec  98.2 MBytes   824 Mbits/sec  71140
[  5]   3.00-4.00   sec  98.2 MBytes   824 Mbits/sec  71090
[  5]   4.00-5.00   sec  98.1 MBytes   823 Mbits/sec  71050
[  5]   5.00-6.00   sec  98.1 MBytes   823 Mbits/sec  71040
[  5]   6.00-7.00   sec  98.1 MBytes   823 Mbits/sec  71050
[  5]   7.00-8.00   sec  98.2 MBytes   824 Mbits/sec  71140
[  5]   8.00-9.00   sec  98.2 MBytes   824 Mbits/sec  71120
[  5]   9.00-10.00  sec  98.3 MBytes   824 Mbits/sec  71150
- - - - - - - - - - - - - - - - - - - - - - - - -
[ ID] Interval           Transfer     Bitrate         Jitter    Lost/Total Datagrams
[  5]   0.00-10.00  sec   981 MBytes   823 Mbits/sec  0.000 ms  0/710720 (0%)  sender
[  5]   0.00-10.00  sec   981 MBytes   823 Mbits/sec  0.041 ms  0/710650 (0%) receiver

Signed-off-by: Mohammad Athari Bin Ismail <mohammad.athari.ismail@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: stmmac: Add HW descriptor prefetch setting for DWMAC Core 5.20 onwards

DWMAC Core 5.20 onwards supports HW descriptor prefetching.
Additionally, it also depends on platform specific RTL configuration.
This capability could be enabled by setting DMA_Mode bit-19 (DCHE).

So, to enable this cability, platform must set plat->dma_cfg->dche = true
and the DWMAC core version must be 5.20 onwards. Else, this capability
wouldn`t be configured

Signed-off-by: Mohammad Athari Bin Ismail <mohammad.athari.ismail@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx4: Treat VFs fair when handling comm_channel_events

Handling comm_channel_event in mlx4_master_comm_channel uses a double
loop to determine which slaves have requested work. The search is
always started at lowest slave. This leads to unfairness; lower VFs
tends to be prioritized over higher VFs.

The patch uses find_next_bit to determine which slaves to handle.
Fairness is implemented by always starting at the next to the last
start.

An MPI program has been used to measure improvements. It runs 500
ibv_reg_mr, synchronizes with all other instances and then runs 500
ibv_dereg_mr.

The results running 500 processes, time reported is for running 500
calls:

ibv_reg_mr:
             Mod.   Org.
mlx4_1    403.356ms 424.674ms
mlx4_2    403.355ms 424.674ms
mlx4_3    403.354ms 424.674ms
mlx4_4    403.355ms 424.674ms
mlx4_5    403.357ms 424.677ms
mlx4_6    403.354ms 424.676ms
mlx4_7    403.357ms 424.675ms
mlx4_8    403.355ms 424.675ms

ibv_dereg_mr:
             Mod.   Org.
mlx4_1    116.408ms 142.818ms
mlx4_2    116.434ms 142.793ms
mlx4_3    116.488ms 143.247ms
mlx4_4    116.679ms 143.230ms
mlx4_5    112.017ms 107.204ms
mlx4_6    112.032ms 107.516ms
mlx4_7    112.083ms 184.195ms
mlx4_8    115.089ms 190.618ms

Suggested-by: Håkon Bugge <haakon.bugge@oracle.com>
Signed-off-by: Hans Westgaard Ry <hans.westgaard.ry@oracle.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: replace return with break for ram code speedup mode timeout

When the timeout occurs, we still have to run the following process
for releasing patch request. Otherwise, the PHY would keep no link.
Therefore, use break to stop the loop of loading firmware and
release the patch request rather than return the function directly.

Fixes: 4a51b0e8a014 ("r8152: support PHY firmware for RTL8156 series")
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
100GbE Intel Wired LAN Driver Updates 2021-04-22

This series contains updates to virtchnl header file, ice, and iavf
drivers.

Vignesh adds support to warn about potentially malicious VFs; those that
are overflowing the mailbox for the ice driver.

Michal adds support for an allowlist/denylist of VF commands based on
supported capabilities for the ice driver.

Brett adds support for iavf UDP segmentation offload by adding the
capability bit to virtchnl, advertising support in the ice driver, and
enabling it in the iavf driver. He also adds a helper function for
getting the VF VSI for ice.

Colin Ian King removes an unneeded pointer assignment.

Qi enables support in the ice driver to support virtchnl requests from
the iavf to configure its own RSS input set. This includes adding new
capability bits, structures, and commands to virtchnl header file.

Haiyue enables configuring RSS flow hash via ethtool to support TCP, UDP
and SCTP protocols in iavf.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

vxge: avoid -Wemtpy-body warnings

There are a few warnings about empty debug macros in this driver:

drivers/net/ethernet/neterion/vxge/vxge-main.c: In function 'vxge_probe':
drivers/net/ethernet/neterion/vxge/vxge-main.c:4480:76: error: suggest braces around empty body in an 'if' statement [-Werror=empty-body]
 4480 |                                 "Failed in enabling SRIOV mode: %d\n", ret);

Change them to proper 'do { } while (0)' expressions to make the
code a little more robust and avoid the warnings.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: wwan: core: Return poll error in case of port removal

Ensure that the poll system call returns proper error flags when port
is removed (nullified port ops), allowing user side to properly fail,
without further read or write.

Fixes: 9a44c1cc6388 ("net: Add a WWAN subsystem")
Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netdevsim: Only use sampling truncation length when valid

When the sampling truncation length is invalid (zero), pass the length
of the packet. Without the fix, no payload is reported to user space
when the truncation length is zero.

Fixes: a8700c3dd0a4 ("netdevsim: Add dummy psample implementation")
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: enetc: fix link error again

A link time bug that I had fixed before has come back now that
another sub-module was added to the enetc driver:

ERROR: modpost: "enetc_ierb_register_pf" [drivers/net/ethernet/freescale/enetc/fsl-enetc.ko] undefined!

The problem is that the enetc Makefile is not actually used for
the ierb module if that is the only built-in driver in there
and everything else is a loadable module.

Fix it by always entering the directory this time, regardless
of which symbols are configured. This should reliably fix the
problem and prevent it from coming back another time.

Fixes: 112463ddbe82 ("net: dsa: felix: fix link error")
Fixes: e7d48e5fbf30 ("net: enetc: add a mini driver for the Integrated Endpoint Register Block")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: mana: fix PCI_HYPERV dependency

The MANA driver causes a build failure in some configurations when
it selects an unavailable symbol:

WARNING: unmet direct dependencies detected for PCI_HYPERV
  Depends on [n]: PCI [=y] && X86_64 [=y] && HYPERV [=n] && PCI_MSI [=y] && PCI_MSI_IRQ_DOMAIN [=y] && SYSFS [=y]
  Selected by [y]:
  - MICROSOFT_MANA [=y] && NETDEVICES [=y] && ETHERNET [=y] && NET_VENDOR_MICROSOFT [=y] && PCI_MSI [=y] && X86_64 [=y]
drivers/pci/controller/pci-hyperv.c: In function 'hv_irq_unmask':
drivers/pci/controller/pci-hyperv.c:1217:9: error: implicit declaration of function 'hv_set_msi_entry_from_desc' [-Werror=implicit-function-declaration]
 1217 |         hv_set_msi_entry_from_desc(&params->int_entry.msi_entry, msi_desc);
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~

A PCI driver should never depend on a particular host bridge
implementation in the first place, but if we have this dependency
it's better to express it as a 'depends on' rather than 'select'.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Dexuan Cui <decui@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

iavf: Support for modifying SCTP RSS flow hashing

Provide the ability to enable SCTP RSS hashing by ethtool.

It gives users option of generating RSS hash based on the SCTP source
and destination ports numbers, IPv4 or IPv6 source and destination
addresses.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: Support for modifying UDP RSS flow hashing

Provides the ability to enable UDP RSS hashing by ethtool.

It gives users option of generating RSS hash based on the UDP source
and destination ports numbers, IPv4 or IPv6 source and destination
addresses.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: Support for modifying TCP RSS flow hashing

Provides the ability to enable TCP RSS hashing by ethtool.

It gives users option of generating RSS hash based on the TCP source
and destination ports numbers, IPv4 or IPv6 source and destination
addresses.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: Add framework to enable ethtool RSS config

Add the virtchnl message interface to VF, so that VF can request RSS
input set(s) based on PF's capability.

This framework allows ethtool RSS config support on the VF driver.

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: Support RSS configure removal for AVF

Add the handler for virtchnl message VIRTCHNL_OP_DEL_RSS_CFG to remove
an existing RSS configuration with matching hashed fields.

Signed-off-by: Vignesh Sridhar <vignesh.sridhar@intel.com>
Co-developed-by: Jia Guo <jia.guo@intel.com>
Signed-off-by: Jia Guo <jia.guo@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Tested-by: Bo Chen <BoX.C.Chen@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: Enable RSS configure for AVF

Currently, RSS hash input is not available to AVF by ethtool, it is set
by the PF directly.

Add the RSS configure support for AVF through new virtchnl message, and
define the capability flag VIRTCHNL_VF_OFFLOAD_ADV_RSS_PF to query this
new RSS offload support.

Signed-off-by: Jia Guo <jia.guo@intel.com>
Signed-off-by: Qi Zhang <qi.z.zhang@intel.com>
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Tested-by: Bo Chen <BoX.C.Chen@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: Add helper function to get the VF's VSI

Currently, the driver gets the VF's VSI by using a long string of
dereferences (i.e. vf->pf->vsi[vf->lan_vsi_idx]). If the method to get
the VF's VSI were to change the driver would have to change it in every
location. Fix this by adding the helper ice_get_vf_vsi().

Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: remove redundant assignment to pointer vsi

Pointer vsi is being re-assigned a value that is never read,
the assignment is redundant and can be removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

iavf: add support for UDP Segmentation Offload

Add code to support UDP segmentation offload (USO) for
hardware that supports it.

Suggested-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: Advertise virtchnl UDP segmentation offload capability

As the hardware is capable of supporting UDP segmentation offload, add a
capability bit to virtchnl.h to communicate this and have the driver
advertise its support.

Suggested-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: Allow ignoring opcodes on specific VF

Declare bitmap of allowed commands on VF. Initialize default
opcodes list that should be always supported. Declare array of
supported opcodes for each caps used in virtchnl code.

Change allowed bitmap by setting or clearing corresponding
bit to allowlist (bit set) or denylist (bit clear).

Signed-off-by: Michal Swiatkowski <michal.swiatkowski@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

ice: warn about potentially malicious VFs

Attempt to detect malicious VFs and, if suspected, log the information but
keep going to allow the user to take any desired actions.

Potentially malicious VFs are identified by checking if the VFs are
transmitting too many messages via the PF-VF mailbox which could cause an
overflow of this channel resulting in denial of service. This is done by
creating a snapshot or static capture of the mailbox buffer which can be
traversed and in which the messages sent by VFs are tracked.

Co-developed-by: Yashaswini Raghuram Prathivadi Bhayankaram <yashaswini.raghuram.prathivadi.bhayankaram@intel.com>
Signed-off-by: Yashaswini Raghuram Prathivadi Bhayankaram <yashaswini.raghuram.prathivadi.bhayankaram@intel.com>
Co-developed-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Co-developed-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Vignesh Sridhar <vignesh.sridhar@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>

Merge tag 'mt76-for-kvalo-2021-04-21' of https://github.com/nbd168/wireless

mt76 patches for 5.13

* testmode improvements
* bugfixes
* device tree power limits support for 7615 and newer
* hardware recovery fixes
* mt7663 reset/init fixes
* mt7915 flash pre-calibration support
* mt7921/mt7663 runtime power management fixes

# gpg: Signature made Wed 21 Apr 2021 09:58:49 PM EEST using DSA key ID 02A76EF5
# gpg: Good signature from "Felix Fietkau <nbd@nbd.name>"
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 75D1 1A7D 91A7 710F 4900  42EF D77D 141D 02A7 6EF5

brcmfmac: Avoid GFP_ATOMIC when GFP_KERNEL is enough

A workqueue is not atomic, so constraints can be relaxed here.
GFP_KERNEL can be used instead of GFP_ATOMIC.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/b6e619415db4ee5de95389280d7195bb56e45f77.1618860716.git.christophe.jaillet@wanadoo.fr

qtnfmac: Fix possible buffer overflow in qtnf_event_handle_external_auth

Function qtnf_event_handle_external_auth calls memcpy without
checking the length.
A user could control that length and trigger a buffer overflow.
Fix by checking the length is within the maximum allowed size.

Signed-off-by: Lee Gibson <leegib@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210419145842.345787-1-leegib@gmail.com

wlcore: Fix buffer overrun by snprintf due to incorrect buffer size

The size of the buffer than can be written to is currently incorrect, it is
always the size of the entire buffer even though the snprintf is writing
as position pos into the buffer. Fix this by setting the buffer size to be
the number of bytes left in the buffer, namely sizeof(buf) - pos.

Addresses-Coverity: ("Out-of-bounds access")
Fixes: 7b0e2c4f6be3 ("wlcore: fix overlapping snprintf arguments in debugfs")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210419141405.180582-1-colin.king@canonical.com

wl3501_cs: Fix out-of-bounds warnings in wl3501_mgmt_join

Fix the following out-of-bounds warnings by adding a new structure
wl3501_req instead of duplicating the same members in structure
wl3501_join_req and wl3501_scan_confirm:

arch/x86/include/asm/string_32.h:182:25: warning: '__builtin_memcpy' offset [39, 108] from the object at 'sig' is out of the bounds of referenced subobject 'beacon_period' with type 'short unsigned int' at offset 36 [-Warray-bounds]
arch/x86/include/asm/string_32.h:182:25: warning: '__builtin_memcpy' offset [25, 95] from the object at 'sig' is out of the bounds of referenced subobject 'beacon_period' with type 'short unsigned int' at offset 22 [-Warray-bounds]

Refactor the code, accordingly:

$ pahole -C wl3501_req drivers/net/wireless/wl3501_cs.o
struct wl3501_req {
        u16                        beacon_period;        /*     0     2 */
        u16                        dtim_period;          /*     2     2 */
        u16                        cap_info;             /*     4     2 */
        u8                         bss_type;             /*     6     1 */
        u8                         bssid[6];             /*     7     6 */
        struct iw_mgmt_essid_pset  ssid;                 /*    13    34 */
        struct iw_mgmt_ds_pset     ds_pset;              /*    47     3 */
        struct iw_mgmt_cf_pset     cf_pset;              /*    50     8 */
        struct iw_mgmt_ibss_pset   ibss_pset;            /*    58     4 */
        struct iw_mgmt_data_rset   bss_basic_rset;       /*    62    10 */

        /* size: 72, cachelines: 2, members: 10 */
        /* last cacheline: 8 bytes */
};

$ pahole -C wl3501_join_req drivers/net/wireless/wl3501_cs.o
struct wl3501_join_req {
        u16                        next_blk;             /*     0     2 */
        u8                         sig_id;               /*     2     1 */
        u8                         reserved;             /*     3     1 */
        struct iw_mgmt_data_rset   operational_rset;     /*     4    10 */
        u16                        reserved2;            /*    14     2 */
        u16                        timeout;              /*    16     2 */
        u16                        probe_delay;          /*    18     2 */
        u8                         timestamp[8];         /*    20     8 */
        u8                         local_time[8];        /*    28     8 */
        struct wl3501_req          req;                  /*    36    72 */

        /* size: 108, cachelines: 2, members: 10 */
        /* last cacheline: 44 bytes */
};

$ pahole -C wl3501_scan_confirm drivers/net/wireless/wl3501_cs.o
struct wl3501_scan_confirm {
        u16                        next_blk;             /*     0     2 */
        u8                         sig_id;               /*     2     1 */
        u8                         reserved;             /*     3     1 */
        u16                        status;               /*     4     2 */
        char                       timestamp[8];         /*     6     8 */
        char                       localtime[8];         /*    14     8 */
        struct wl3501_req          req;                  /*    22    72 */
        /* --- cacheline 1 boundary (64 bytes) was 30 bytes ago --- */
        u8                         rssi;                 /*    94     1 */

        /* size: 96, cachelines: 2, members: 8 */
        /* padding: 1 */
        /* last cacheline: 32 bytes */
};

The problem is that the original code is trying to copy data into a
bunch of struct members adjacent to each other in a single call to
memcpy(). Now that a new struct wl3501_req enclosing all those adjacent
members is introduced, memcpy() doesn't overrun the length of
&sig.beacon_period and &this->bss_set[i].beacon_period, because the
address of the new struct object _req_ is used as the destination,
instead.

This helps with the ongoing efforts to globally enable -Warray-bounds
and get us closer to being able to tighten the FORTIFY_SOURCE routines
on memcpy().

Link: https://github.com/KSPP/linux/issues/109
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1fbaf516da763b50edac47d792a9145aa4482e29.1618442265.git.gustavoars@kernel.org

wl3501_cs: Fix out-of-bounds warnings in wl3501_send_pkt

Fix the following out-of-bounds warnings by enclosing structure members
daddr and saddr into new struct addr, in structures wl3501_md_req and
wl3501_md_ind:

arch/x86/include/asm/string_32.h:182:25: warning: '__builtin_memcpy' offset [18, 23] from the object at 'sig' is out of the bounds of referenced subobject 'daddr' with type 'u8[6]' {aka 'unsigned char[6]'} at offset 11 [-Warray-bounds]
arch/x86/include/asm/string_32.h:182:25: warning: '__builtin_memcpy' offset [18, 23] from the object at 'sig' is out of the bounds of referenced subobject 'daddr' with type 'u8[6]' {aka 'unsigned char[6]'} at offset 11 [-Warray-bounds]

Refactor the code, accordingly:

$ pahole -C wl3501_md_req drivers/net/wireless/wl3501_cs.o
struct wl3501_md_req {
u16                        next_blk;             /*     0     2 */
u8                         sig_id;               /*     2     1 */
u8                         routing;              /*     3     1 */
u16                        data;                 /*     4     2 */
u16                        size;                 /*     6     2 */
u8                         pri;                  /*     8     1 */
u8                         service_class;        /*     9     1 */
struct {
u8                 daddr[6];             /*    10     6 */
u8                 saddr[6];             /*    16     6 */
} addr;                                          /*    10    12 */

/* size: 22, cachelines: 1, members: 8 */
/* last cacheline: 22 bytes */
};

$ pahole -C wl3501_md_ind drivers/net/wireless/wl3501_cs.o
struct wl3501_md_ind {
u16                        next_blk;             /*     0     2 */
u8                         sig_id;               /*     2     1 */
u8                         routing;              /*     3     1 */
u16                        data;                 /*     4     2 */
u16                        size;                 /*     6     2 */
u8                         reception;            /*     8     1 */
u8                         pri;                  /*     9     1 */
u8                         service_class;        /*    10     1 */
struct {
u8                 daddr[6];             /*    11     6 */
u8                 saddr[6];             /*    17     6 */
} addr;                                          /*    11    12 */

/* size: 24, cachelines: 1, members: 9 */
/* padding: 1 */
/* last cacheline: 24 bytes */
};

The problem is that the original code is trying to copy data into a
couple of arrays adjacent to each other in a single call to memcpy().
Now that a new struct _addr_ enclosing those two adjacent arrays
is introduced, memcpy() doesn't overrun the length of &sig.daddr[0]
and &sig.daddr, because the address of the new struct object _addr_
is used, instead.

This helps with the ongoing efforts to globally enable -Warray-bounds
and get us closer to being able to tighten the FORTIFY_SOURCE routines
on memcpy().

Link: https://github.com/KSPP/linux/issues/109
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/d260fe56aed7112bff2be5b4d152d03ad7b78e78.1618442265.git.gustavoars@kernel.org

ath11k: fix warning in ath11k_mhi_config

Initialize static variable ath11k_mhi_config for all hw_rev,
return error for unknown hw_rev.
This patch fixes below Smatch warning:
    drivers/net/wireless/ath/ath11k/mhi.c:357 ath11k_mhi_register()
    error: uninitialized symbol 'ath11k_mhi_config'.

Tested-on: QCN9074 hw1.0 PCI WLAN.HK.2.4.0.1-01734-QCAHKSWPL_SILICONZ-1

Fixes: a233811ef600 ("ath11k: Add qcn9074 mhi controller config")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Anilkumar Kolli <akolli@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1617857830-19315-1-git-send-email-akolli@codeaurora.org

ath11k: qmi: Fix spelling mistake "requeqst" -> "request"

There is a spelling mistake in an ath11k_warn message. Fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210316091924.15627-1-colin.king@canonical.com

ath10k: Fix ath10k_wmi_tlv_op_pull_peer_stats_info() unlock without lock

ath10k_wmi_tlv_op_pull_peer_stats_info() could try to unlock RCU lock
winthout locking it first when peer reason doesn't match the valid
cases for this function.

Add a default case to return without unlocking.

Fixes: 09078368d516 ("ath10k: hold RCU lock when calling ieee80211_find_sta_by_ifaddr()")
Reported-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210406230228.31301-1-skhan@linuxfoundation.org

ath10k: Fix a use after free in ath10k_htc_send_bundle

In ath10k_htc_send_bundle, the bundle_skb could be freed by
dev_kfree_skb_any(bundle_skb). But the bundle_skb is used later
by bundle_skb->len.

As skb_len = bundle_skb->len, my patch replaces bundle_skb->len to
skb_len after the bundle_skb was freed.

Fixes: c8334512f3dd1 ("ath10k: add htt TX bundle for sdio")
Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210329120154.8963-1-lyl2019@mail.ustc.edu.cn

ath9k: Fix error check in ath9k_hw_read_revisions() for PCI devices

When the error check in ath9k_hw_read_revisions() was added, it checked for
-EIO which is what ath9k_regread() in the ath9k_htc driver uses. However,
for plain ath9k, the register read function uses ioread32(), which just
returns -1 on error. So if such a read fails, it still gets passed through
and ends up as a weird mac revision in the log output.

Fix this by changing ath9k_regread() to return -1 on error like ioread32()
does, and fix the error check to look for that instead of -EIO.

Fixes: 2f90c7e5d094 ("ath9k: Check for errors when reading SREV register")
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Reviewed-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210326180819.142480-1-toke@redhat.com

net: phy: marvell: don't use empty switch default case

This causes error reported by kernel test robot.

Signed-off-by: Marek Behún <kabel@kernel.org>
Fixes: 41d26bf4aba0 ("net: phy: marvell: refactor HWMON OOP style")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: bridge: fix error in br_multicast_add_port when CONFIG_NET_SWITCHDEV=n

When CONFIG_NET_SWITCHDEV is disabled, the shim for switchdev_port_attr_set
inside br_mc_disabled_update returns -EOPNOTSUPP. This is not caught,
and propagated to the caller of br_multicast_add_port, preventing ports
from joining the bridge.

Reported-by: Christian Borntraeger <borntraeger@de.ibm.com>
Fixes: ae1ea84b33da ("net: bridge: propagate error code and extack from br_mc_disabled_update")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mt76: mt7921: reinit wpdma during drv_own if necessary

Check dummy reg to reinitialized WPDMA during driver_own operation

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: move mt76_token_init in mt76_alloc_device

In order to remove duplicated code, move mt76_token_init in
mt76_alloc_device routine

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7921: fix a precision vs width bug in printk

Precision %.*s was intended instead of width %*s.  The original code
is potentially an information leak.

Fixes: c7cc5ec57303 ("mt76: mt7921: rework mt7921_mcu_debug_msg_event routine")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7915: fix a precision vs width bug in printk

Precision %.*s was intended instead of width %*s.  The original code
will still print unintended data from beyond the end of skb->data.

Fixes: 665b2c780d63 ("mt76: mt7915: limit firmware log message printk to buffer length")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7615: fix a precision vs width bug in printk

Precision "%.*s" was intended instead of width "%*s".  The original code
will print garbage from beyond the end of the skb->data.

Fixes: d76d6c3ba2b0 ("mt76: mt7615: limit firmware log message printk to buffer length")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: debugfs: introduce napi_threaded node

Introduce napi_threaded debugfs knob in order to enable/disable NAPI
threaded support

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7921: introduce mt7921_mcu_sta_add routine

mt7921_mcu_sta_add will be used to add and remove wtbl entries.
Create broadcast wtbl entry after AP association

Co-developed-by: Deren Wu <deren.wu@mediatek.com>
Signed-off-by: Deren Wu <deren.wu@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7921: mt7921_stop should put device in fw_own state

mt7921_stop should put device in fw_own state to reduce
power consumption.

Signed-off-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: connac: unschedule mac_work before going to sleep

In order to wake the device less frequently and so reduce power
consumpation, unschedule mac_work before going to sleep

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7663: add awake and doze time accounting

Similar to mt7921, introduce awake and doze time accounting
for runtime pm.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7921: improve doze opportunity

Increase mt7921 mac work timeout in oder to have move sleep
opportunities

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7921: get rid of mcu_reset function pointer

since mcu_reset it used only by mt7921, move the reset callback to
mt7921_mcu_parse_response routine and get rid of the function pointer.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7915: do not read rf value from efuse in flash mode

Do not read rf value from efuse when driver is configured to
use flash mode.

Tested-by: Bruce Chuang <bruce-ss.chuang@mediatek.com>
Signed-off-by: Shayne Chen <shayne.chen@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>