net: dsa: lan9303: remove unnecessary dev_set_drvdata()

Remove unnecessary dev_set_drvdata() in ->remove(), the driver_data will
be set to NULL in device_unbind_cleanup() after calling ->remove().

Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: dsa: hellcreek: remove unnecessary platform_set_drvdata()

Remove unnecessary platform_set_drvdata() in ->remove(), the driver_data
will be set to NULL in device_unbind_cleanup() after calling ->remove().

Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Acked-by: Kurt Kanzenbach <kurt@linutronix.de>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: dsa: loop: remove unnecessary dev_set_drvdata()

Remove unnecessary dev_set_drvdata() in ->remove(), the driver_data will
be set to NULL in device_unbind_cleanup() after calling ->remove().

Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: dsa: bcm_sf2: remove unnecessary platform_set_drvdata()

Remove unnecessary platform_set_drvdata() in ->remove(), the driver_data
will be set to NULL in device_unbind_cleanup() after calling ->remove().

Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: dsa: b53: remove unnecessary set_drvdata()

Remove unnecessary set_drvdata(NULL) function in ->remove(),
the driver_data will be set to NULL in device_unbind_cleanup()
after calling ->remove().

Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

ptp_ocp: use device_find_any_child() instead of custom approach

We have already a helper to get the first child device, use it and
drop custom approach.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Vadim Fedorenko <vadfed@fb.com>
Link: https://lore.kernel.org/r/20220921141005.2443-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

ethtool: tunnels: check the return value of nla_nest_start()

Check the return value of nla_nest_start(). When starting the entry
level nested attributes, if the tailroom of socket buffer is
insufficient to store the attribute header and payload, the return value
will be NULL.

There is, however, no real bug here since if the skb is full
nla_put_be16() will fail as well and we'll error out.

Signed-off-by: Li Zhong <floridsleeves@gmail.com>
Link: https://lore.kernel.org/r/20220921181716.1629541-1-floridsleeves@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'mlx5-macsec-extended-packet-number-and-replay-window-offload'

Saeed Mahameed says:

====================
mlx5 MACSec Extended packet number and replay window offload

This is a follow up series to the previously submitted mlx5 MACsec offload [1]
earlier this release cycle.

In this series we add the support for MACsec Extended packet number and
replay window offloads.

First patch is a simple modification (code movements) to the core macsec code
to allow exposing the EPN related user properties to the offloading
device driver.

The rest of the patches are mlx5 specific, we start off with fixing some
trivial issues with mlx5 MACsec code, and a simple refactoring to allow
additional functionality in mlx5 macsec to support EPN and window replay
offloads.
 A) Expose mkey creation functionality to MACsec
 B) Expose ASO object to MACsec, to allow advanced steering operations,
    ASO objects are used to modify MACsec steering objects in fastpath.

1) Support MACsec offload extended packet number (EPN)

    MACsec EPN splits the packet number (PN) into two 32-bits fields,
    epn_lsb (32 least significant bits (LSBs) of PN) and epn_msb (32
    most significant bits (MSBs) of PN).
    Epn_msb bits are managed by SW and for that HW is required to send
    an object change event of type EPN event notifying the SW to update
    the epn_msb in addition, once epn_msb is updated SW update HW with
    the new epn_msb value for HW to perform replay protection.
    To prevent HW from stopping while handling the event, SW manages
    another bit for HW called epn_overlap, HW uses the latter to get
    an indication regarding how to read the epn_msb value correctly
    while still receiving packets.
    Add epn event handling that updates the epn_overlap and epn_msb for
    every 2^31 packets according to the following logic:
    if epn_lsb crosses 2^31 (half sequence number wraparound) upon HW
    relevant event, SW updates the esn_overlap value to OLD (value = 1).
    When the epn_lsb crosses 2^32 (full sequence number wraparound)
    upon HW relevant event, SW updates the esn_overlap to NEW
    (value = 0) and increment the esn_msb.
    When using MACsec EPN a salt and short secure channel id (ssci)
    needs to be provided by the user, when offloading EPN need to pass
    this salt and ssci to the HW to be used in the initial vector (IV)
    calculations.

2) Support MACsec offload replay window

    Support setting replay window size for MACsec offload.
    Currently supported window size of 32, 64, 128 and 256
    bit. Other values will be returned as invalid parameter.

[1] https://lore.kernel.org/netdev/20220906052129.104507-1-saeed@kernel.org/
====================

Link: https://lore.kernel.org/r/20220921181054.40249-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Support MACsec offload replay window

Support setting replay window size for MACsec offload.
Currently supported window size of 32, 64, 128 and 256
bit. Other values will be returned as invalid parameter.

Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Support MACsec offload extended packet number (EPN)

MACsec EPN splits the packet number (PN) into two 32-bits fields,
epn_lsb (32 least significant bits (LSBs) of PN) and epn_msb (32
most significant bits (MSBs) of PN).
Epn_msb bits are managed by SW and for that HW is required to send
an object change event of type EPN event notifying the SW to update
the epn_msb in addition, once epn_msb is updated SW update HW with
the new epn_msb value for HW to perform replay protection.
To prevent HW from stopping while handling the event, SW manages
another bit for HW called epn_overlap, HW uses the latter to get
an indication regarding how to read the epn_msb value correctly
while still receiving packets.
Add epn event handling that updates the epn_overlap and epn_msb for
every 2^31 packets according to the following logic:
if epn_lsb crosses 2^31 (half sequence number wraparound) upon HW
relevant event, SW updates the esn_overlap value to OLD (value = 1).
When the epn_lsb crosses 2^32 (full sequence number wraparound)
upon HW relevant event, SW updates the esn_overlap to NEW
(value = 0) and increment the esn_msb.
When using MACsec EPN a salt and short secure channel id (ssci)
needs to be provided by the user, when offloading EPN need to pass
this salt and ssci to the HW to be used in the initial vector (IV)
calculations.

Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Move MACsec initialization from profile init stage to profile enable stage

Postpone MACsec initialization to the mlx5e profile enable stage to have
user access region (UAR) pages and other resources ready before MACsec
initialization to initialize advanced steering operation (ASO) hardware
resources.

Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Create advanced steering operation (ASO) object for MACsec

Add support for ASO work queue entry (WQE) data to allow reading
data upon querying the ASO work queue (WQ).
Register user mode memory registration (UMR) upon ASO WQ init,
de-register UMR upon ASO WQ cleanup.
MACsec uses UMR to determine the cause of the event triggered
by the HW since different scenarios could trigger the same event.
Setup MACsec ASO object to sync HW with SW about various macsec
flow stateful features like: replay window, lifetime limits e.t.c

Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Expose memory key creation (mkey) function

Expose mlx5e_create_mkey function, for future patches in the
macsec series to use.
The above function creates a memory key which describes a
region in memory that can be later used by both HW and SW.
The counterpart destroy functionality is already exposed.

Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5: Add ifc bits for MACsec extended packet number (EPN) and replay protection

Add ifc bits related to advanced steering operations (ASO) and general
object modify for macsec to use as part of offloading EPN and replay
protection features.

Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Fix MACsec initial packet number

Currently when creating MACsec object, next_pn which represents
the initial packet number (PN) is considered only in TX flow.
The above causes mismatch between TX and RX initial PN which
is reflected in packet drops.
Fix by considering next_pn in RX flow too.

Fixes: aae3454e4d4c ("net/mlx5e: Add MACsec offload Rx command support")
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Fix MACsec initialization error path

Currently MACsec initialization error path does not
destroy sci hash table in case of failure.
Fix by destroying hash table in case of failure.

Fixes: 9515978eee0b ("net/mlx5e: Implement MACsec Tx data path using MACsec skb_metadata_dst")
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5: Fix fields name prefix in MACsec

Fix ifc fields name to be consistent with the device spec document.

Fixes: 8385c51ff5bc ("net/mlx5: Introduce MACsec Connect-X offload hardware bits and structures")
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: macsec: Expose extended packet number (EPN) properties to macsec offload

Currently macsec invokes HW offload path before reading extended packet
number (EPN) related user properties i.e. salt and short secure channel
identifier (ssci), hence preventing macsec EPN HW offload.
Expose those by moving macsec EPN properties reading prior to HW offload
path.

Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'cleanup-in-huawei-hinic-driver'

Zhengchao Shao says:

====================
cleanup in Huawei hinic driver

Do code cleanup in Huawei hinic driver.
====================

Link: https://lore.kernel.org/r/20220921123358.63442-1-shaozhengchao@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: remove the unused input parameter prod_idx in sq_prepare_ctrl()

The input parameter prod_idx is not used in sq_prepare_ctrl(), remove it.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: replace magic numbers with macro

Replace magic numbers with macro.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: remove unused enumerated value

remove unused enumerated value.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: change hinic_deinit_vf_hw() to void

When hinic_pci_sriov_disable() calls hinic_deinit_vf_hw(), it doesn't
care about the return value of hinic_deinit_vf_hw(). Also
hinic_deinit_vf_hw() is return 0, so change it to void.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: simplify code logic

simplify code logic in hinic_ndo_set_vf_trust() and
hinic_ndo_set_vf_spoofchk().

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: remove duplicate macro definition

The macro HINIC_WAIT_SRIOV_CFG_TIMEOUT is defined twice, remove one.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: remove unused macro

remove unused macro.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: remove unused functions

hinic_hwdev_max_num_qpas() and hinic_msix_attr_get() are no longer called,
remove them. Also the macro HINIC_MSIX_ATTR_GET is also not called, remove
it.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: change type of function to be static

Functions are called only in one file, change their type to static.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: hinic: modify kernel doc comments

The type of cmdq_free_page() hinic_set_pf_action() and
link_status_event_handler() are void, modify the comments.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'refactor-duplicate-codes-in-the-qdisc-class-walk-function'

Zhengchao Shao says:

====================
refactor duplicate codes in the qdisc class walk function

The walk implementation of most qdisc class modules is basically the
same. That is, the values of count and skip are checked first. If count
is greater than or equal to skip, the registered fn function is
executed. Otherwise, increase the value of count. So the code can be
refactored.

The walk function is invoked during dump. Therefore, test cases related
 to the tdc filter need to be added.
====================

Link: https://lore.kernel.org/r/20220921024040.385296-1-shaozhengchao@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add show class case for red qdisc

Test 290a: Show RED class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add show class case for prio qdisc

Test 2410: Show prio class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add show class case for mq qdisc

Test 1023: Show mq class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add show class case for ingress qdisc

Test 0521: Show ingress class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for qfq qdisc

Test 0582: Create QFQ with default setting
Test c9a3: Create QFQ with class weight setting
Test 8452: Create QFQ with class maxpkt setting
Test d920: Create QFQ with multiple class setting
Test 0548: Delete QFQ with handle
Test 5901: Show QFQ class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for netem qdisc

Test cb28: Create NETEM with default setting
Test a089: Create NETEM with limit flag
Test 3449: Create NETEM with delay time
Test 3782: Create NETEM with distribution and corrupt flag
Test 2b82: Create NETEM with distribution and duplicate flag
Test a932: Create NETEM with distribution and loss flag
Test e01a: Create NETEM with distribution and loss state flag
Test ba29: Create NETEM with loss gemodel flag
Test 0492: Create NETEM with reorder flag
Test 7862: Create NETEM with rate limit
Test 7235: Create NETEM with multiple slot rate
Test 5439: Create NETEM with multiple slot setting
Test 5029: Change NETEM with loss state
Test 3785: Replace NETEM with delay time
Test 4502: Delete NETEM with handle
Test 0785: Show NETEM class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for multiq qdisc

Test 20ba: Add multiq Qdisc to multi-queue device (8 queues)
Test 4301: List multiq Class
Test 7832: Delete nonexistent multiq Qdisc
Test 2891: Delete multiq Qdisc twice
Test 1329: Add multiq Qdisc to single-queue device

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for mqprio qdisc

Test 9903: Add mqprio Qdisc to multi-queue device (8 queues)
Test 453a: Delete nonexistent mqprio Qdisc
Test 5292: Delete mqprio Qdisc twice
Test 45a9: Add mqprio Qdisc to single-queue device
Test 2ba9: Show mqprio class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for htb qdisc

Test 0904: Create HTB with default setting
Test 3906: Create HTB with default-N setting
Test 8492: Create HTB with r2q setting
Test 9502: Create HTB with direct_qlen setting
Test b924: Create HTB with class rate and burst setting
Test 4359: Create HTB with class mpu setting
Test 9048: Create HTB with class prio setting
Test 4994: Create HTB with class ceil setting
Test 9523: Create HTB with class cburst setting
Test 5353: Create HTB with class mtu setting
Test 346a: Create HTB with class quantum setting
Test 303a: Delete HTB with handle

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for hfsc qdisc

Test 3254: Create HFSC with default setting
Test 0289: Create HFSC with class sc and ul rate setting
Test 846a: Create HFSC with class sc umax and dmax setting
Test 5413: Create HFSC with class rt and ls rate setting
Test 9312: Create HFSC with class rt umax and dmax setting
Test 6931: Delete HFSC with handle
Test 8436: Show HFSC class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for fq_codel qdisc

Test 4957: Create FQ_CODEL with default setting
Test 7621: Create FQ_CODEL with limit setting
Test 6871: Create FQ_CODEL with memory_limit setting
Test 5636: Create FQ_CODEL with target setting
Test 630a: Create FQ_CODEL with interval setting
Test 4324: Create FQ_CODEL with quantum setting
Test b190: Create FQ_CODEL with noecn flag
Test 5381: Create FQ_CODEL with ce_threshold setting
Test c9d2: Create FQ_CODEL with drop_batch setting
Test 523b: Create FQ_CODEL with multiple setting
Test 9283: Replace FQ_CODEL with noecn setting
Test 3459: Change FQ_CODEL with limit setting
Test 0128: Delete FQ_CODEL with handle
Test 0435: Show FQ_CODEL class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for dsmark qdisc

Test 6345: Create DSMARK with default setting
Test 3462: Create DSMARK with default_index setting
Test ca95: Create DSMARK with set_tc_index flag
Test a950: Create DSMARK with multiple setting
Test 4092: Delete DSMARK with handle
Test 5930: Show DSMARK class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for drr qdisc

Test 0385: Create DRR with default setting
Test 2375: Delete DRR with handle
Test 3092: Show DRR class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for cbs qdisc

Test 1820: Create CBS with default setting
Test 1532: Create CBS with hicredit setting
Test 2078: Create CBS with locredit setting
Test 9271: Create CBS with sendslope setting
Test 0482: Create CBS with idleslope setting
Test e8f3: Create CBS with multiple setting
Test 23c9: Replace CBS with sendslope setting
Test a07a: Change CBS with idleslope setting
Test 43b3: Delete CBS with handle
Test 9472: Show CBS class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for cbq qdisc

Test 3460: Create CBQ with default setting
Test 0592: Create CBQ with mpu
Test 4684: Create CBQ with valid cell num
Test 4345: Create CBQ with invalid cell num
Test 4525: Create CBQ with valid ewma
Test 6784: Create CBQ with invalid ewma
Test 5468: Delete CBQ with handle
Test 492a: Show CBQ class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests/tc-testing: add selftests for cake qdisc

Test 1212: Create CAKE with default setting
Test 3281: Create CAKE with bandwidth limit
Test c940: Create CAKE with autorate-ingress flag
Test 2310: Create CAKE with rtt time
Test 2385: Create CAKE with besteffort flag
Test a032: Create CAKE with diffserv8 flag
Test 2349: Create CAKE with diffserv4 flag
Test 8472: Create CAKE with flowblind flag
Test 2341: Create CAKE with dsthost and nat flag
Test 5134: Create CAKE with wash flag
Test 2302: Create CAKE with flowblind and no-split-gso flag
Test 0768: Create CAKE with dual-srchost and ack-filter flag
Test 0238: Create CAKE with dual-dsthost and ack-filter-aggressive flag
Test 6572: Create CAKE with memlimit and ptm flag
Test 2436: Create CAKE with fwmark and atm flag
Test 3984: Create CAKE with overhead and mpu
Test 5421: Create CAKE with conservative and ingress flag
Test 6854: Delete CAKE with conservative and ingress flag
Test 2342: Replace CAKE with mpu
Test 2313: Change CAKE with mpu
Test 4365: Show CAKE class

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: use tc_qdisc_stats_dump() in qdisc

use tc_qdisc_stats_dump() in qdisc.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Reviewed-by: Victor Nogueira <victor@mojatatu.com>
Tested-by: Victor Nogueira <victor@mojatatu.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: sch_api: add helper for tc qdisc walker stats dump

The walk implementation of most qdisc class modules is basically the
same. That is, the values of count and skip are checked first. If
count is greater than or equal to skip, the registered fn function is
executed. Otherwise, increase the value of count. So we can reconstruct
them.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: taprio: remove unnecessary taprio_list_lock

The 3 functions that want access to the taprio_list:
taprio_dev_notifier(), taprio_destroy() and taprio_init() are all called
with the rtnl_mutex held, therefore implicitly serialized with respect
to each other. A spin lock serves no purpose.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Link: https://lore.kernel.org/r/20220921095632.1379251-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'support-256-bit-tls-keys-with-device-offload'

Gal Pressman says:

====================
Support 256 bit TLS keys with device offload

This series adds support for 256 bit TLS keys with device offload, and a
cleanup patch to remove repeating code:
- Patches #1-2 add cipher sizes descriptors which allow reducing the
  amount of code duplications.
- Patch #3 allows 256 bit keys to be TX offloaded in the tls module (RX
  already supported).
- Patch #4 adds 256 bit keys support to the mlx5 driver.
====================

Link: https://lore.kernel.org/r/20220920130150.3546-1-gal@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx5e: Support 256 bit keys with kTLS device offload

Add support for 256 bit TLS keys using device offload.

Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/tls: Support 256 bit keys with TX device offload

Add the missing clause for 256 bit keys in tls_set_device_offload(), and
the needed adjustments in tls_device_fallback.c.

Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/tls: Use cipher sizes structs

Use the newly introduced cipher sizes structs instead of the repeated
switch cases churn.

Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/tls: Describe ciphers sizes by const structs

Introduce cipher sizes descriptor. It helps reducing the amount of code
duplications and repeated switch/cases that assigns the proper sizes
according to the cipher type.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: replace reset with config timestamps

Any change to the hardware timestamps configuration triggers nic restart,
which breaks transmition and reception of network packets for a while.
But there is no need to fully restart the device because while configuring
hardware timestamps. The code for changing configuration runs after all
of the initialisation, when the NIC is actually up and running. This patch
changes the code that ioctl will only update configuration registers and
will not trigger carrier status change, but in case of timestamps for
all rx packetes it fallbacks to close()/open() sequnce because of
synchronization issues in the hardware. Tested on BCM57504.

Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Vadim Fedorenko <vfedorenko@novek.ru>
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20220922191038.29921-1-vfedorenko@novek.ru
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge git://git./linux/kernel/git/netdev/net

drivers/net/ethernet/freescale/fec.h
  7b15515fc1ca ("Revert "fec: Restart PPS after link state change"")
  40c79ce13b03 ("net: fec: add stop mode support for imx8 platform")
https://lore.kernel.org/all/20220921105337.62b41047@canb.auug.org.au/

drivers/pinctrl/pinctrl-ocelot.c
  c297561bc98a ("pinctrl: ocelot: Fix interrupt controller")
  181f604b33cd ("pinctrl: ocelot: add ability to be used in a non-mmio configuration")
https://lore.kernel.org/all/20220921110032.7cd28114@canb.auug.org.au/

tools/testing/selftests/drivers/net/bonding/Makefile
  bbb774d921e2 ("net: Add tests for bonding and team address list management")
  152e8ec77640 ("selftests/bonding: add a test for bonding lladdr target")
https://lore.kernel.org/all/20220921110437.5b7dbd82@canb.auug.org.au/

drivers/net/can/usb/gs_usb.c
  5440428b3da6 ("can: gs_usb: gs_can_open(): fix race dev->can.state condition")
  45dfa45f52e6 ("can: gs_usb: add RX and TX hardware timestamp support")
https://lore.kernel.org/all/84f45a7d-92b6-4dc5-d7a1-072152fab6ff@tessares.net/

Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: phy: micrel: Fix double spaces inside lan8814_config_intr

Inside the function lan8814_config_intr, there are double spaces when
assigning the return value of phy_write to err.

Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220921065444.637067-1-horatiu.vultur@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge tag 'net-6.0-rc7' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from wifi, netfilter and can.

  A handful of awaited fixes here - revert of the FEC changes, bluetooth
  fix, fixes for iwlwifi spew.

  We added a warning in PHY/MDIO code which is triggering on a couple of
  platforms in a false-positive-ish way. If we can't iron that out over
  the week we'll drop it and re-add for 6.1.

  I've added a new "follow up fixes" section for fixes to fixes in
  6.0-rcs but it may actually give the false impression that those are
  problematic or that more testing time would have caught them. So
  likely a one time thing.

  Follow up fixes:

   - nf_tables_addchain: fix nft_counters_enabled underflow

   - ebtables: fix memory leak when blob is malformed

   - nf_ct_ftp: fix deadlock when nat rewrite is needed

  Current release - regressions:

   - Revert "fec: Restart PPS after link state change" and the related
     "net: fec: Use a spinlock to guard `fep->ptp_clk_on`"

   - Bluetooth: fix HCIGETDEVINFO regression

   - wifi: mt76: fix 5 GHz connection regression on mt76x0/mt76x2

   - mptcp: fix fwd memory accounting on coalesce

   - rwlock removal fall out:
      - ipmr: always call ip{,6}_mr_forward() from RCU read-side
        critical section
      - ipv6: fix crash when IPv6 is administratively disabled

   - tcp: read multiple skbs in tcp_read_skb()

   - mdio_bus_phy_resume state warning fallout:
      - eth: ravb: fix PHY state warning splat during system resume
      - eth: sh_eth: fix PHY state warning splat during system resume

  Current release - new code bugs:

   - wifi: iwlwifi: don't spam logs with NSS>2 messages

   - eth: mtk_eth_soc: enable XDP support just for MT7986 SoC

  Previous releases - regressions:

   - bonding: fix NULL deref in bond_rr_gen_slave_id

   - wifi: iwlwifi: mark IWLMEI as broken

  Previous releases - always broken:

   - nf_conntrack helpers:
      - irc: tighten matching on DCC message
      - sip: fix ct_sip_walk_headers
      - osf: fix possible bogus match in nf_osf_find()

   - ipvlan: fix out-of-bound bugs caused by unset skb->mac_header

   - core: fix flow symmetric hash

   - bonding, team: unsync device addresses on ndo_stop

   - phy: micrel: fix shared interrupt on LAN8814"

* tag 'net-6.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (83 commits)
  selftests: forwarding: add shebang for sch_red.sh
  bnxt: prevent skb UAF after handing over to PTP worker
  net: marvell: Fix refcounting bugs in prestera_port_sfp_bind()
  net: sched: fix possible refcount leak in tc_new_tfilter()
  net: sunhme: Fix packet reception for len < RX_COPY_THRESHOLD
  udp: Use WARN_ON_ONCE() in udp_read_skb()
  selftests: bonding: cause oops in bond_rr_gen_slave_id
  bonding: fix NULL deref in bond_rr_gen_slave_id
  net: phy: micrel: fix shared interrupt on LAN8814
  net/smc: Stop the CLC flow if no link to map buffers on
  ice: Fix ice_xdp_xmit() when XDP TX queue number is not sufficient
  net: atlantic: fix potential memory leak in aq_ndev_close()
  can: gs_usb: gs_usb_set_phys_id(): return with error if identify is not supported
  can: gs_usb: gs_can_open(): fix race dev->can.state condition
  can: flexcan: flexcan_mailbox_read() fix return value for drop = true
  net: sh_eth: Fix PHY state warning splat during system resume
  net: ravb: Fix PHY state warning splat during system resume
  netfilter: nf_ct_ftp: fix deadlock when nat rewrite is needed
  netfilter: ebtables: fix memory leak when blob is malformed
  netfilter: nf_tables: fix percpu memory leak at nf_tables_addchain()
  ...

Merge tag 'efi-urgent-for-v6.0-2' of git://git./linux/kernel/git/efi/efi

Pull EFI fixes from Ard Biesheuvel:

 - Use the right variable to check for shim insecure mode

 - Wipe setup_data field when booting via EFI

 - Add missing error check to efibc driver

* tag 'efi-urgent-for-v6.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
  efi: libstub: check Shim mode using MokSBStateRT
  efi: x86: Wipe setup_data on pure EFI boot
  efi: efibc: Guard against allocation failure

Merge tag 'gpio-fixes-for-v6.0-rc7' of git://git./linux/kernel/git/brgl/linux

Pull gpio fixes from Bartosz Golaszewski:

 - fix a NULL-pointer dereference at driver unbind and a potential
   resource leak in error path in gpio-mockup

 - make the irqchip immutable in gpio-ftgpio010

 - fix dereferencing a potentially uninitialized variable in gpio-tqmx86

 - fix interrupt registering in gpiolib's character device code

* tag 'gpio-fixes-for-v6.0-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  gpiolib: cdev: Set lineevent_state::irq after IRQ register successfully
  gpio: tqmx86: fix uninitialized variable girq
  gpio: ftgpio010: Make irqchip immutable
  gpio: mockup: Fix potential resource leakage when register a chip
  gpio: mockup: fix NULL pointer dereference when removing debugfs

Merge tag 'perf-tools-fixes-for-v6.0-2022-09-21' of git://git./linux/kernel/git/acme/linux

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - Fix polling of system-wide events related to mixing per-cpu and
   per-thread events.

 - Do not check if /proc/modules is unchanged when copying /proc/kcore,
   that doesn't get in the way of post processing analysis.

 - Include program header in ELF files generated for JIT files, so that
   they can be opened by tools using elfutils libraries.

 - Enter namespaces when synthesizing build-ids.

 - Fix some bugs related to a recent cpu_map overhaul where we should be
   using an index and not the cpu number.

 - Fix BPF program ELF section name, using the naming expected by libbpf
   when using BPF counters in 'perf stat'.

 - Add a new test for perf stat cgroup BPF counter.

 - Adjust check on 'perf test wp' for older kernels, where the
   PERF_EVENT_IOC_MODIFY_ATTRIBUTES ioctl isn't supported.

 - Sync x86 cpufeatures with the kernel sources, no changes in tooling.

* tag 'perf-tools-fixes-for-v6.0-2022-09-21' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
  perf tools: Honor namespace when synthesizing build-ids
  tools headers cpufeatures: Sync with the kernel sources
  perf kcore_copy: Do not check /proc/modules is unchanged
  libperf evlist: Fix polling of system-wide events
  perf record: Fix cpu mask bit setting for mixed mmaps
  perf test: Skip wp modify test on old kernels
  perf jit: Include program header in ELF files
  perf test: Add a new test for perf stat cgroup BPF counter
  perf stat: Use evsel->core.cpus to iterate cpus in BPF cgroup counters
  perf stat: Fix cpu map index in bperf cgroup code
  perf stat: Fix BPF program section name

selftests: forwarding: add shebang for sch_red.sh

RHEL/Fedora RPM build checks are stricter, and complain when executable
files don't have a shebang line, e.g.

*** WARNING: ./kselftests/net/forwarding/sch_red.sh is executable but has no shebang, removing executable bit

Fix it by adding shebang line.

Fixes: 6cf0291f9517 ("selftests: forwarding: Add a RED test for SW datapath")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Link: https://lore.kernel.org/r/20220922024453.437757-1-liuhangbin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt: prevent skb UAF after handing over to PTP worker

When reading the timestamp is required bnxt_tx_int() hands
over the ownership of the completed skb to the PTP worker.
The skb should not be used afterwards, as the worker may
run before the rest of our code and free the skb, leading
to a use-after-free.

Since dev_kfree_skb_any() accepts NULL make the loss of
ownership more obvious and set skb to NULL.

Fixes: 83bb623c968e ("bnxt_en: Transmit and retrieve packet timestamps")
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20220921201005.335390-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: marvell: Fix refcounting bugs in prestera_port_sfp_bind()

In prestera_port_sfp_bind(), there are two refcounting bugs:
(1) we should call of_node_get() before of_find_node_by_name() as
it will automaitcally decrease the refcount of 'from' argument;
(2) we should call of_node_put() for the break of the iteration
for_each_child_of_node() as it will automatically increase and
decrease the 'child'.

Fixes: 52323ef75414 ("net: marvell: prestera: add phylink support")
Signed-off-by: Liang He <windhl@126.com>
Reviewed-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Link: https://lore.kernel.org/r/20220921133245.4111672-1-windhl@126.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ethernet: adi: Fix invalid parent name length

MII_BUS_ID_SIZE is larger than MAX_PHYS_ITEM_ID_LEN
so we use the former here to set the parent port id.

Fixes: bc93e19d088b ("net: ethernet: adi: Add ADIN1110 support")
Signed-off-by: Alexandru Tachici <alexandru.tachici@analog.com>
Link: https://lore.kernel.org/r/20220922063049.10388-1-alexandru.tachici@analog.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: sched: fix possible refcount leak in tc_new_tfilter()

tfilter_put need to be called to put the refount got by tp->ops->get to
avoid possible refcount leak when chain->tmplt_ops != NULL and
chain->tmplt_ops != tp->ops.

Fixes: 7d5509fa0d3d ("net: sched: extend proto ops with 'put' callback")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Link: https://lore.kernel.org/r/20220921092734.31700-1-hbh25y@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: sunhme: Fix packet reception for len < RX_COPY_THRESHOLD

There is a separate receive path for small packets (under 256 bytes).
Instead of allocating a new dma-capable skb to be used for the next packet,
this path allocates a skb and copies the data into it (reusing the existing
sbk for the next packet). There are two bytes of junk data at the beginning
of every packet. I believe these are inserted in order to allow aligned DMA
and IP headers. We skip over them using skb_reserve. Before copying over
the data, we must use a barrier to ensure we see the whole packet. The
current code only synchronizes len bytes, starting from the beginning of
the packet, including the junk bytes. However, this leaves off the final
two bytes in the packet. Synchronize the whole packet.

To reproduce this problem, ping a HME with a payload size between 17 and
214

$ ping -s 17 <hme_address>

which will complain rather loudly about the data mismatch. Small packets
(below 60 bytes on the wire) do not have this issue. I suspect this is
related to the padding added to increase the minimum packet size.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Sean Anderson <seanga2@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220920235018.1675956-1-seanga2@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

udp: Use WARN_ON_ONCE() in udp_read_skb()

Prevent udp_read_skb() from flooding the syslog.

Suggested-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Peilin Ye <peilin.ye@bytedance.com>
Link: https://lore.kernel.org/r/20220921005915.2697-1-yepeilin.cs@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'bonding-fix-null-deref-in-bond_rr_gen_slave_id'

Jonathan Toppins says:

====================
bonding: fix NULL deref in bond_rr_gen_slave_id

Fix a NULL dereference of the struct bonding.rr_tx_counter member because
if a bond is initially created with an initial mode != zero (Round Robin)
the memory required for the counter is never created and when the mode is
changed there is never any attempt to verify the memory is allocated upon
switching modes.
====================

Link: https://lore.kernel.org/r/cover.1663694476.git.jtoppins@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests: bonding: cause oops in bond_rr_gen_slave_id

This bonding selftest used to cause a kernel oops on aarch64
and should be architectures agnostic.

Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bonding: fix NULL deref in bond_rr_gen_slave_id

Fix a NULL dereference of the struct bonding.rr_tx_counter member because
if a bond is initially created with an initial mode != zero (Round Robin)
the memory required for the counter is never created and when the mode is
changed there is never any attempt to verify the memory is allocated upon
switching modes.

This causes the following Oops on an aarch64 machine:
    [  334.686773] Unable to handle kernel paging request at virtual address ffff2c91ac905000
    [  334.694703] Mem abort info:
    [  334.697486]   ESR = 0x0000000096000004
    [  334.701234]   EC = 0x25: DABT (current EL), IL = 32 bits
    [  334.706536]   SET = 0, FnV = 0
    [  334.709579]   EA = 0, S1PTW = 0
    [  334.712719]   FSC = 0x04: level 0 translation fault
    [  334.717586] Data abort info:
    [  334.720454]   ISV = 0, ISS = 0x00000004
    [  334.724288]   CM = 0, WnR = 0
    [  334.727244] swapper pgtable: 4k pages, 48-bit VAs, pgdp=000008044d662000
    [  334.733944] [ffff2c91ac905000] pgd=0000000000000000, p4d=0000000000000000
    [  334.740734] Internal error: Oops: 96000004 [#1] SMP
    [  334.745602] Modules linked in: bonding tls veth rfkill sunrpc arm_spe_pmu vfat fat acpi_ipmi ipmi_ssif ixgbe igb i40e mdio ipmi_devintf ipmi_msghandler arm_cmn arm_dsu_pmu cppc_cpufreq acpi_tad fuse zram crct10dif_ce ast ghash_ce sbsa_gwdt nvme drm_vram_helper drm_ttm_helper nvme_core ttm xgene_hwmon
    [  334.772217] CPU: 7 PID: 2214 Comm: ping Not tainted 6.0.0-rc4-00133-g64ae13ed4784 #4
    [  334.779950] Hardware name: GIGABYTE R272-P31-00/MP32-AR1-00, BIOS F18v (SCP: 1.08.20211002) 12/01/2021
    [  334.789244] pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
    [  334.796196] pc : bond_rr_gen_slave_id+0x40/0x124 [bonding]
    [  334.801691] lr : bond_xmit_roundrobin_slave_get+0x38/0xdc [bonding]
    [  334.807962] sp : ffff8000221733e0
    [  334.811265] x29: ffff8000221733e0 x28: ffffdbac8572d198 x27: ffff80002217357c
    [  334.818392] x26: 000000000000002a x25: ffffdbacb33ee000 x24: ffff07ff980fa000
    [  334.825519] x23: ffffdbacb2e398ba x22: ffff07ff98102000 x21: ffff07ff981029c0
    [  334.832646] x20: 0000000000000001 x19: ffff07ff981029c0 x18: 0000000000000014
    [  334.839773] x17: 0000000000000000 x16: ffffdbacb1004364 x15: 0000aaaabe2f5a62
    [  334.846899] x14: ffff07ff8e55d968 x13: ffff07ff8e55db30 x12: 0000000000000000
    [  334.854026] x11: ffffdbacb21532e8 x10: 0000000000000001 x9 : ffffdbac857178ec
    [  334.861153] x8 : ffff07ff9f6e5a28 x7 : 0000000000000000 x6 : 000000007c2b3742
    [  334.868279] x5 : ffff2c91ac905000 x4 : ffff2c91ac905000 x3 : ffff07ff9f554400
    [  334.875406] x2 : ffff2c91ac905000 x1 : 0000000000000001 x0 : ffff07ff981029c0
    [  334.882532] Call trace:
    [  334.884967]  bond_rr_gen_slave_id+0x40/0x124 [bonding]
    [  334.890109]  bond_xmit_roundrobin_slave_get+0x38/0xdc [bonding]
    [  334.896033]  __bond_start_xmit+0x128/0x3a0 [bonding]
    [  334.901001]  bond_start_xmit+0x54/0xb0 [bonding]
    [  334.905622]  dev_hard_start_xmit+0xb4/0x220
    [  334.909798]  __dev_queue_xmit+0x1a0/0x720
    [  334.913799]  arp_xmit+0x3c/0xbc
    [  334.916932]  arp_send_dst+0x98/0xd0
    [  334.920410]  arp_solicit+0xe8/0x230
    [  334.923888]  neigh_probe+0x60/0xb0
    [  334.927279]  __neigh_event_send+0x3b0/0x470
    [  334.931453]  neigh_resolve_output+0x70/0x90
    [  334.935626]  ip_finish_output2+0x158/0x514
    [  334.939714]  __ip_finish_output+0xac/0x1a4
    [  334.943800]  ip_finish_output+0x40/0xfc
    [  334.947626]  ip_output+0xf8/0x1a4
    [  334.950931]  ip_send_skb+0x5c/0x100
    [  334.954410]  ip_push_pending_frames+0x3c/0x60
    [  334.958758]  raw_sendmsg+0x458/0x6d0
    [  334.962325]  inet_sendmsg+0x50/0x80
    [  334.965805]  sock_sendmsg+0x60/0x6c
    [  334.969286]  __sys_sendto+0xc8/0x134
    [  334.972853]  __arm64_sys_sendto+0x34/0x4c
    [  334.976854]  invoke_syscall+0x78/0x100
    [  334.980594]  el0_svc_common.constprop.0+0x4c/0xf4
    [  334.985287]  do_el0_svc+0x38/0x4c
    [  334.988591]  el0_svc+0x34/0x10c
    [  334.991724]  el0t_64_sync_handler+0x11c/0x150
    [  334.996072]  el0t_64_sync+0x190/0x194
    [  334.999726] Code: b9001062 f9403c02 d53cd044 8b040042 (b8210040)
    [  335.005810] ---[ end trace 0000000000000000 ]---
    [  335.010416] Kernel panic - not syncing: Oops: Fatal exception in interrupt
    [  335.017279] SMP: stopping secondary CPUs
    [  335.021374] Kernel Offset: 0x5baca8eb0000 from 0xffff800008000000
    [  335.027456] PHYS_OFFSET: 0x80000000
    [  335.030932] CPU features: 0x0000,0085c029,19805c82
    [  335.035713] Memory Limit: none
    [  335.038756] Rebooting in 180 seconds..

The fix is to allocate the memory in bond_open() which is guaranteed
to be called before any packets are processed.

Fixes: 848ca9182a7d ("net: bonding: Use per-cpu rr_tx_counter")
CC: Jussi Maki <joamaki@gmail.com>
Signed-off-by: Jonathan Toppins <jtoppins@redhat.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: phy: micrel: fix shared interrupt on LAN8814

Since commit ece19502834d ("net: phy: micrel: 1588 support for LAN8814
phy") the handler always returns IRQ_HANDLED, except in an error case.
Before that commit, the interrupt status register was checked and if
it was empty, IRQ_NONE was returned. Restore that behavior to play nice
with the interrupt line being shared with others.

Fixes: ece19502834d ("net: phy: micrel: 1588 support for LAN8814 phy")
Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Reviewed-by: Divya Koppera <Divya.Koppera@microchip.com>
Link: https://lore.kernel.org/r/20220920141619.808117-1-michael@walle.cc
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'add-wed-support-for-mt7986-chipset'

Lorenzo Bianconi says:

====================
Add WED support for MT7986 chipset

Similar to MT7622, introduce Wireless Ethernet Dispatch (WED) support
for MT7986 chipset in order to offload to the hw packet engine traffic
received from LAN/WAN device to WLAN nic (MT7915E).
====================

Link: https://lore.kernel.org/r/cover.1663668203.git.lorenzo@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_soc: introduce flow offloading support for mt7986

Introduce hw flow offload support for mt7986 chipset. PPE is not enabled
yet in mt7986 since mt76 support is not available yet.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Co-developed-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_wed: add axi bus support

Other than pcie bus, introduce support for axi bus to mtk wed driver.
Axi bus is used to connect mt7986-wmac soc chip available on mt7986
device.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Co-developed-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_wed: add wed support for mt7986 chipset

Introduce Wireless Etherne Dispatcher support on transmission side
for mt7986 chipset

Tested-by: Daniel Golle <daniel@makrotopia.org>
Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Co-developed-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_wed: add mtk_wed_configure_irq and mtk_wed_dma_{enable/disable}

Introduce mtk_wed_configure_irq, mtk_wed_dma_enable and mtk_wed_dma_disable
utility routines.
This is a preliminary patch to introduce mt7986 wed support.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Co-developed-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_soc: add foe_entry_size to mtk_eth_soc

Introduce foe_entry_size to mtk_eth_soc data structure since mt7986
relies on a bigger mtk_foe_entry data structure.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_soc: move wdma_base definitions in mtk register map

This is a preliminary patch to introduce mt7986 wed support.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_soc: add the capability to run multiple ppe

mt7986 chipset support multiple packet engines for wlan <-> eth
packet forwarding.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Co-developed-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_soc: move ppe table hash offset to mtk_soc_data structure

This is a preliminary patch to introduce mt7986 hw packet engine.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Co-developed-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Sujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: mtk_eth_soc: move gdma_to_ppe and ppe_base definitions in mtk register map

This is a preliminary patch to introduce mt7986 hw packet engine.

Tested-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

dt-bindings: net: mediatek: add WED binding for MT7986 eth driver

Document the binding for the Wireless Ethernet Dispatch core on the
MT7986 ethernet driver

Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

arm64: dts: mediatek: mt7986: add support for Wireless Ethernet Dispatch

Introduce wed0 and wed1 nodes in order to enable offloading forwarding
between ethernet and wireless devices on the mt7986 chipset.

Co-developed-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Bo Jiao <Bo.Jiao@mediatek.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Merge branch 'separate-smc-parameter-settings-from-tcp-sysctls'

Wen Gu says:

====================
Separate SMC parameter settings from TCP sysctls

SMC shares some sysctls with TCP, but considering the difference
between these two protocols, it may not be very suitable for SMC
to reuse TCP parameter settings in some cases, such as keepalive
time or buffer size.

So this patch set aims to introduce some SMC specific sysctls to
independently and flexibly set the parameters that suit SMC.
====================

Link: https://lore.kernel.org/r/1663667542-119851-1-git-send-email-guwen@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net/smc: Unbind r/w buffer size from clcsock and make them tunable

Currently, SMC uses smc->sk.sk_{rcv|snd}buf to create buffers for
send buffer and RMB. And the values of buffer size are from tcp_{w|r}mem
in clcsock.

The buffer size from TCP socket doesn't fit SMC well. Generally, buffers
are usually larger than TCP for SMC-R/-D to get higher performance, for
they are different underlay devices and paths.

So this patch unbinds buffer size from TCP, and introduces two sysctl
knobs to tune them independently. Also, these knobs are per net
namespace and work for containers.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net/smc: Introduce a specific sysctl for TEST_LINK time

SMC-R tests the viability of link by sending out TEST_LINK LLC
messages over RoCE fabric when connections on link have been
idle for a time longer than keepalive interval (testlink time).

But using tcp_keepalive_time as testlink time maybe not quite
suitable because it is default no less than two hours[1], which
is too long for single link to find peer dead. The active host
will still use peer-dead link (QP) sending messages, and can't
find out until get IB_WC_RETRY_EXC_ERR error CQEs, which takes
more time than TEST_LINK timeout (SMC_LLC_WAIT_TIME) normally.

So this patch introduces a independent sysctl for SMC-R to set
link keepalive time, in order to detect link down in time. The
default value is 30 seconds.

[1] https://www.rfc-editor.org/rfc/rfc1122#page-101

Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net/smc: Stop the CLC flow if no link to map buffers on

There might be a potential race between SMC-R buffer map and
link group termination.

smc_smcr_terminate_all()     | smc_connect_rdma()
--------------------------------------------------------------
                             | smc_conn_create()
for links in smcibdev        |
        schedule links down  |
                             | smc_buf_create()
                             |  \- smcr_buf_map_usable_links()
                             |      \- no usable links found,
                             |         (rmb->mr = NULL)
                             |
                             | smc_clc_send_confirm()
                             |  \- access conn->rmb_desc->mr[]->rkey
                             |     (panic)

During reboot and IB device module remove, all links will be set
down and no usable links remain in link groups. In such situation
smcr_buf_map_usable_links() should return an error and stop the
CLC flow accessing to uninitialized mr.

Fixes: b9247544c1bc ("net/smc: convert static link ID instances to support multiple links")
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Link: https://lore.kernel.org/r/1663656189-32090-1-git-send-email-guwen@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: ethernet: altera: TSE: fix error return code in altera_tse_probe()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: fef2998203e1 ("net: altera: tse: convert to phylink")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Sun Ke <sunke32@huawei.com>
Link: https://lore.kernel.org/r/20220920020041.2685948-1-sunke32@huawei.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

efi: libstub: check Shim mode using MokSBStateRT

We currently check the MokSBState variable to decide whether we should
treat UEFI secure boot as being disabled, even if the firmware thinks
otherwise. This is used by shim to indicate that it is not checking
signatures on boot images. In the kernel, we use this to relax lockdown
policies.

However, in cases where shim is not even being used, we don't want this
variable to interfere with lockdown, given that the variable may be
non-volatile and therefore persist across a reboot. This means setting
it once will persistently disable lockdown checks on a given system.

So switch to the mirrored version of this variable, called MokSBStateRT,
which is supposed to be volatile, and this is something we can check.

Cc: <stable@vger.kernel.org> # v4.19+
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Peter Jones <pjones@redhat.com>

efi: x86: Wipe setup_data on pure EFI boot

When booting the x86 kernel via EFI using the LoadImage/StartImage boot
services [as opposed to the deprecated EFI handover protocol], the setup
header is taken from the image directly, and given that EFI's LoadImage
has no Linux/x86 specific knowledge regarding struct bootparams or
struct setup_header, any absolute addresses in the setup header must
originate from the file and not from a prior loading stage.

Since we cannot generally predict where LoadImage() decides to load an
image (*), such absolute addresses must be treated as suspect: even if a
prior boot stage intended to make them point somewhere inside the
[signed] image, there is no way to validate that, and if they point at
an arbitrary location in memory, the setup_data nodes will not be
covered by any signatures or TPM measurements either, and could be made
to contain an arbitrary sequence of SETUP_xxx nodes, which could
interfere quite badly with the early x86 boot sequence.

(*) Note that, while LoadImage() does take a buffer/size tuple in
addition to a device path, which can be used to provide the image
contents directly, it will re-allocate such images, as the memory
footprint of an image is generally larger than the PE/COFF file
representation.

Cc: <stable@vger.kernel.org> # v5.10+
Link: https://lore.kernel.org/all/20220904165321.1140894-1-Jason@zx2c4.com/
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Acked-by: Jason A. Donenfeld <Jason@zx2c4.com>

Merge branch 'master' of git://git./linux/kernel/git/netfilter/nf-next

Florian Westphal says:

====================
netfilter patches for net-next

Remove GPL license copypastry in uapi files, those have SPDX tags.
From Christophe Jaillet.

Remove unused variable in rpfilter, from Guillaume Nault.

Rework gc resched delay computation in conntrack, from Antoine Tenart.

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf-next:
  netfilter: rpfilter: Remove unused variable 'ret'.
  headers: Remove some left-over license text in include/uapi/linux/netfilter/
  netfilter: conntrack: revisit the gc initial rescheduling bias
  netfilter: conntrack: fix the gc rescheduling delay
====================

Link: https://lore.kernel.org/r/20220921095000.29569-1-fw@strlen.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch '100GbE' of git://git./linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2022-09-20 (ice)

Michal re-sets TC configuration when changing number of queues.

Mateusz moves the check and call for link-down-on-close to the specific
path for downing/closing the interface.

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  ice: Fix interface being down after reset with link-down-on-close flag on
  ice: config netdev tc before setting queues number
====================

Link: https://lore.kernel.org/r/20220920205344.1860934-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

ethernet: tundra: Drop forward declaration of static functions

Usually it's not necessary to declare static functions if the symbols are
in the right order. Moving the definition of tsi_eth_driver down in the
compilation unit allows to drop two such declarations.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220919131515.885361-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

flow_dissector: Do not count vlan tags inside tunnel payload

We've met the problem that when there is a vlan tag inside
GRE encapsulation, the match of num_of_vlans fails.
It is caused by the vlan tag inside GRE payload has been
counted into num_of_vlans, which is not expected.

One example packet is like this:
Ethernet II, Src: Broadcom_68:56:07 (00:10:18:68:56:07)
                   Dst: Broadcom_68:56:08 (00:10:18:68:56:08)
802.1Q Virtual LAN, PRI: 0, DEI: 0, ID: 100
Internet Protocol Version 4, Src: 192.168.1.4, Dst: 192.168.1.200
Generic Routing Encapsulation (Transparent Ethernet bridging)
Ethernet II, Src: Broadcom_68:58:07 (00:10:18:68:58:07)
                   Dst: Broadcom_68:58:08 (00:10:18:68:58:08)
802.1Q Virtual LAN, PRI: 0, DEI: 0, ID: 200
...
It should match the (num_of_vlans 1) rule, but it matches
the (num_of_vlans 2) rule.

The vlan tags inside the GRE or other tunnel encapsulated payload
should not be taken into num_of_vlans.
The fix is to stop counting the vlan number when the encapsulation
bit is set.

Fixes: 34951fcf26c5 ("flow_dissector: Add number of vlan tags dissector")
Signed-off-by: Qingqing Yang <qingqing.yang@broadcom.com>
Reviewed-by: Boris Sukholitko <boris.sukholitko@broadcom.com>
Link: https://lore.kernel.org/r/20220919074808.136640-1-qingqing.yang@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: sched: remove unused tcf_result extension

Added by:
commit e5cf1baf92cb ("act_mirred: use TC_ACT_REINSERT when possible")
but no longer useful.

Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Link: https://lore.kernel.org/r/20220919130627.3551233-1-jhs@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'clean-up-ocelot_reset-routine'

Colin Foster says:

====================
clean up ocelot_reset() routine

ocelot_reset() will soon be exported to a common library to be used by
the ocelot_ext system. This will make error values from regmap calls
possible, so they must be checked. Additionally, readx_poll_timeout()
can be substituted for the custom loop, as a simple cleanup.

I don't have hardware to verify this set directly, but there shouldn't
be any functional changes.
====================

Link: https://lore.kernel.org/r/20220917175127.161504-1-colin.foster@in-advantage.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: mscc: ocelot: check return values of writes during reset

The ocelot_reset() function utilizes regmap_field_write() but wasn't
checking return values. While this won't cause issues for the current MMIO
regmaps, it could be an issue for externally controlled interfaces.

Add checks for these return values.

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: mscc: ocelot: utilize readx_poll_timeout() for chip reset

Clean up the reset code by utilizing readx_poll_timeout instead of a custom
loop.

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-ll_temac-cleanup-for-clearing-static-warnings'

Haoyue Xu says:

====================
net: ll_temac: Cleanup for clearing static warnings

Most static warnings are detected by Checkpatch.pl, mainly about:
(1) #1: About the comments.
(2) #2: About function name in a string.
(3) #3: About the open parenthesis.
(4) #4: About the else branch.
(6) #6: About trailing statements.
(7) #5,#7: About blank lines and spaces.
====================

Link: https://lore.kernel.org/r/20220917103843.526877-1-xuhaoyue1@hisilicon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>