net: page_pool: factor out releasing DMA from releasing the page

Releasing the DMA mapping will be useful for other types
of pages, so factor it out. Make sure compiler inlines it,
to avoid any regressions.

Signed-off-by: Mina Almasry <almasrymina@google.com>
Reviewed-by: Shakeel Butt <shakeelb@google.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

page_pool: transition to reference count management after page draining

To support multiple users referencing the same fragment,
'pp_frag_count' is renamed to 'pp_ref_count', transitioning pp pages
from fragment management to reference count management after draining
based on the suggestion from [1].

The idea is that the concept of fragmenting exists before the page is
drained, and all related functions retain their current names.
However, once the page is drained, its management shifts to being
governed by 'pp_ref_count'. Therefore, all functions associated with
that lifecycle stage of a pp page are renamed.

[1]
http://lore.kernel.org/netdev/f71d9448-70c8-8793-dc9a-0eb48a570300@huawei.com

Signed-off-by: Liang Chen <liangchen.linux@gmail.com>
Reviewed-by: Yunsheng Lin <linyunsheng@huawei.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Mina Almasry <almasrymina@google.com>
Link: https://lore.kernel.org/r/20231212044614.42733-2-liangchen.linux@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

cxgb3: Avoid potential string truncation in desc

Builds with W=1 were warning about potential string truncations:

drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c: In function 'cxgb_up':
drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:394:38: warning: '%d' directive output may be truncated writing between 1 and 11 bytes into a region of size between 5 and 20 [-Wformat-truncation=]
  394 |                                  "%s-%d", d->name, pi->first_qset + i);
      |                                      ^~
In function 'name_msix_vecs',
    inlined from 'cxgb_up' at drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:1264:3: drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:394:34: note: directive argument in the range [-2147483641, 509]
  394 |                                  "%s-%d", d->name, pi->first_qset + i);
      |                                  ^~~~~~~
drivers/net/ethernet/chelsio/cxgb3/cxgb3_main.c:393:25: note: 'snprintf' output between 3 and 28 bytes into a destination of size 21
  393 |                         snprintf(adap->msix_info[msi_idx].desc, n,
      |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  394 |                                  "%s-%d", d->name, pi->first_qset + i);
      |                                  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Avoid open-coded %NUL-termination (this code was assuming snprintf
wasn't %NUL terminating when it does -- likely thinking of strncpy),
and grow the size of the string to handle a maximal value.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202312100937.ZPZCARhB-lkp@intel.com/
Cc: Raju Rangoju <rajur@chelsio.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20231212220954.work.219-kees@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

amd-xgbe: Avoid potential string truncation in name

Build with W=1 were warning about a potential string truncation:

drivers/net/ethernet/amd/xgbe/xgbe-drv.c: In function 'xgbe_alloc_channels':
drivers/net/ethernet/amd/xgbe/xgbe-drv.c:211:73: warning: '%u' directive output may be truncated writing between 1 and 10 bytes into a region of size 8 [-Wformat-truncation=]
  211 |                 snprintf(channel->name, sizeof(channel->name), "channel-%u", i);
      |                                                                         ^~
drivers/net/ethernet/amd/xgbe/xgbe-drv.c:211:64: note: directive argument in the range [0, 4294967294]
  211 |                 snprintf(channel->name, sizeof(channel->name), "channel-%u", i);
      |                                                                ^~~~~~~~~~~~
drivers/net/ethernet/amd/xgbe/xgbe-drv.c:211:17: note: 'snprintf' output between 10 and 19 bytes into a destination of size 16
  211 |                 snprintf(channel->name, sizeof(channel->name), "channel-%u", i);
      |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Increase the size of the "name" buffer to handle the full format range.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202312100937.ZPZCARhB-lkp@intel.com/
Cc: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20231212221312.work.830-kees@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

dpll: allocate pin ids in cycle

Pin ID is just a number. Nobody should rely on a certain value, instead,
user should use either pin-id-get op or RTNetlink to get it.

Unify the pin ID allocation behavior with what there is already
implemented for dpll devices.

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20231212150605.1141261-1-jiri@resnulli.us
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

sctp: support MSG_ERRQUEUE flag in recvmsg()

For some reason sctp_poll() generates EPOLLERR if sk->sk_error_queue
is not empty but recvmsg() can not drain the error queue yet.

This is needed to better support timestamping.

I had to export inet_recv_error(), since sctp
can be compiled as a module.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Cc: Willem de Bruijn <willemb@google.com>
Acked-by: Xin Long <lucien.xin@gmail.com>
Link: https://lore.kernel.org/r/20231212145550.3872051-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'idpf-add-get-set-for-ethtool-s-header-split-ringparam'

Alexander Lobakin says:

====================
idpf: add get/set for Ethtool's header split ringparam

Currently, the header split feature (putting headers in one smaller
buffer and then the data in a separate bigger one) is always enabled
in idpf when supported.
One may want to not have fragmented frames per each packet, for example,
to avoid XDP frags. To better optimize setups for particular workloads,
add ability to switch the header split state on and off via Ethtool's
ringparams, as well as to query the current status.
There's currently only GET in the Ethtool Netlink interface for now,
so add SET first. I suspect idpf is not the only one supporting this.
====================

Link: https://lore.kernel.org/r/20231212142752.935000-1-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

idpf: add get/set for Ethtool's header split ringparam

idpf supports the header split feature and that feature is always
enabled by default.
However, for flexibility reasons and to simplify some scenarios, it
would be useful to have the support for switching the header split
off (and on) from the userspace.

Address that need by adding the user config parameter, the functions
for disabling (or enabling) the header split feature, and calls to
them from the Ethtool ringparam callbacks.
It still is enabled by default if supported by the hardware.

Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Co-developed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Link: https://lore.kernel.org/r/20231212142752.935000-3-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

ethtool: add SET for TCP_DATA_SPLIT ringparam

Follow up commit 9690ae604290 ("ethtool: add header/data split
indication") and add the set part of Ethtool's header split, i.e.
ability to enable/disable header split via the Ethtool Netlink
interface. This might be helpful to optimize the setup for particular
workloads, for example, to avoid XDP frags, and so on.
A driver should advertise ``ETHTOOL_RING_USE_TCP_DATA_SPLIT`` in its
ops->supported_ring_params to allow doing that. "Unknown" passed from
the userspace when the header split is supported means the driver is
free to choose the preferred state.

Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Link: https://lore.kernel.org/r/20231212142752.935000-2-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

docs: networking: timestamping: mention MSG_EOR flag

TCP got MSG_EOR support in linux-4.7.

This is a canonical way of making sure no coalescing
will be performed on the skb, even if it could not be
immediately sent.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Martin KaFai Lau <kafai@fb.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Link: https://lore.kernel.org/r/20231212110608.3673677-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'add-support-for-dp83tg720s-phy'

Oleksij Rempel says:

====================
add support for DP83TG720S PHY
====================

Link: https://lore.kernel.org/r/20231212054144.87527-1-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: phy: Add support for the DP83TG720S Ethernet PHY

The DP83TG720S-Q1 device is an IEEE 802.3bp and Open Alliance compliant
automotive Ethernet physical layer transceiver.

This driver was tested with i.MX8MP EQOS (stmmac) on the MAC side and
same TI PHY on other side.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/20231212054144.87527-3-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: phy: c45: add genphy_c45_pma_read_ext_abilities() function

Move part of the genphy_c45_pma_read_abilities() code to a separate
function.

Some PHYs do not implement PMA/PMD status 2 register (Register 1.8) but
do implement PMA/PMD extended ability register (Register 1.11). To make
use of it, we need to be able to access this part of code separately.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/20231212054144.87527-2-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-sched-optimizations-around-action-binding-and-init'

Pedro Tammela says:

====================
net/sched: optimizations around action binding and init

Scaling optimizations for action binding in rtnl-less filters.
We saw a noticeable lock contention around idrinfo->lock when
testing in a 56 core system, which disappeared after the patches.
====================

Link: https://lore.kernel.org/r/20231211181807.96028-1-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: act_api: skip idr replace on bound actions

tcf_idr_insert_many will replace the allocated -EBUSY pointer in
tcf_idr_check_alloc with the real action pointer, exposing it
to all operations. This operation is only needed when the action pointer
is created (ACT_P_CREATED). For actions which are bound to (returned 0),
the pointer already resides in the idr making such operation a nop.

Even though it's a nop, it's still not a cheap operation as internally
the idr code walks the idr and then does a replace on the appropriate slot.
So if the action was bound, better skip the idr replace entirely.

Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Link: https://lore.kernel.org/r/20231211181807.96028-3-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: act_api: rely on rcu in tcf_idr_check_alloc

Instead of relying only on the idrinfo->lock mutex for
bind/alloc logic, rely on a combination of rcu + mutex + atomics
to better scale the case where multiple rtnl-less filters are
binding to the same action object.

Action binding happens when an action index is specified explicitly and
an action exists which such index exists. Example:
  tc actions add action drop index 1
  tc filter add ... matchall action drop index 1
  tc filter add ... matchall action drop index 1
  tc filter add ... matchall action drop index 1
  tc filter ls ...
     filter protocol all pref 49150 matchall chain 0 filter protocol all pref 49150 matchall chain 0 handle 0x1
     not_in_hw
           action order 1: gact action drop
            random type none pass val 0
            index 1 ref 4 bind 3

   filter protocol all pref 49151 matchall chain 0 filter protocol all pref 49151 matchall chain 0 handle 0x1
     not_in_hw
           action order 1: gact action drop
            random type none pass val 0
            index 1 ref 4 bind 3

   filter protocol all pref 49152 matchall chain 0 filter protocol all pref 49152 matchall chain 0 handle 0x1
     not_in_hw
           action order 1: gact action drop
            random type none pass val 0
            index 1 ref 4 bind 3

When no index is specified, as before, grab the mutex and allocate
in the idr the next available id. In this version, as opposed to before,
it's simplified to store the -EBUSY pointer instead of the previous
alloc + replace combination.

When an index is specified, rely on rcu to find if there's an object in
such index. If there's none, fallback to the above, serializing on the
mutex and reserving the specified id. If there's one, it can be an -EBUSY
pointer, in which case we just try again until it's an action, or an action.
Given the rcu guarantees, the action found could be dead and therefore
we need to bump the refcount if it's not 0, handling the case it's
in fact 0.

As bind and the action refcount are already atomics, these increments can
happen without the mutex protection while many tcf_idr_check_alloc race
to bind to the same action instance.

In case binding encounters a parallel delete or add, it will return
-EAGAIN in order to try again. Both filter and action apis already
have the retry machinery in-place. In case it's an unlocked filter it
retries under the rtnl lock.

Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Link: https://lore.kernel.org/r/20231211181807.96028-2-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'virtio-net-dynamic-coalescing-moderation'

Heng Qi says:

====================
virtio-net: support dynamic coalescing moderation

Now, virtio-net already supports per-queue moderation parameter
setting. Based on this, we use the linux dimlib to support
dynamic coalescing moderation for virtio-net.

Due to some scheduling issues, we only support and test the rx dim.

Some test results:

I. Sockperf UDP
=================================================
1. Env
rxq_0 with affinity to cpu_0.

2. Cmd
client: taskset -c 0 sockperf tp -p 8989 -i $IP -t 10 -m 16B
server: taskset -c 0 sockperf sr -p 8989

3. Result
dim off: 1143277.00 rxpps, throughput 17.844 MBps, cpu is 100%.
dim on:  1124161.00 rxpps, throughput 17.610 MBps, cpu is 83.5%.
=================================================

II. Redis
=================================================
1. Env
There are 8 rxqs, and rxq_i with affinity to cpu_i.

2. Result
When all cpus are 100%, ops/sec of memtier_benchmark client is
dim off:  978437.23
dim on:  1143638.28
=================================================

III. Nginx
=================================================
1. Env
There are 8 rxqs and rxq_i with affinity to cpu_i.

2. Result
When all cpus are 100%, requests/sec of wrk client is
dim off:  877931.67
dim on:  1019160.31
=================================================

IV. Latency of sockperf udp
=================================================
1. Rx cmd
taskset -c 0 sockperf sr -p 8989

2. Tx cmd
taskset -c 0 sockperf pp -i ${ip} -p 8989 -t 10

After running this cmd 5 times and averaging the results,

3. Result
dim off: 17.7735 usec
dim on:  18.0110 usec
=================================================

Changelog:
v7->v8:
- Add select DIMLIB.

v6->v7:
- Drop the patch titled "spin lock for ctrl cmd access"
- Use rtnl_trylock to avoid the deadlock.

v5->v6:
- Add patch(4/5): spin lock for ctrl cmd access
- Patch(5/5):
   - Use spin lock and cancel_work_sync to synchronize

v4->v5:
- Patch(4/4):
   - Fix possible synchronization issues with cancel_work_sync.
   - Reduce if/else nesting levels

v3->v4:
- Patch(5/5): drop.

v2->v3:
- Patch(4/5): some minor modifications.

v1->v2:
- Patch(2/5): a minor fix.
- Patch(4/5):
   - improve the judgment of dim switch conditions.
   - Cancel the work when vq reset.
- Patch(5/5): drop the tx dim implementation.
====================

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

virtio-net: support rx netdim

By comparing the traffic information in the complete napi processes,
let the virtio-net driver automatically adjust the coalescing
moderation parameters of each receive queue.

Signed-off-by: Heng Qi <hengqi@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

virtio-net: extract virtqueue coalescig cmd for reuse

Extract commands to set virtqueue coalescing parameters for reuse
by ethtool -Q, vq resize and netdim.

Signed-off-by: Heng Qi <hengqi@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

virtio-net: separate rx/tx coalescing moderation cmds

This patch separates the rx and tx global coalescing moderation
commands to support netdim switches in subsequent patches.

Signed-off-by: Heng Qi <hengqi@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

virtio-net: returns whether napi is complete

rx netdim needs to count the traffic during a complete napi process,
and start updating and comparing samples to make decisions after
the napi ends. Let virtqueue_napi_complete() return true if napi is done,
otherwise vice versa.

Signed-off-by: Heng Qi <hengqi@linux.alibaba.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'ionic-pci-errors'

Shannon Nelson says:

====================
ionic: updates to PCI error handling

These are improvements to our PCI error handling, including FLR and
AER events.
====================

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

ionic: fill out pci error handlers

Set up the pci_error_handlers error_detected and resume to be useful in
handling AER events.  If the error detected is pci_channel_io_frozen we
set up to do an FLR at the end of the AER handling - this tends to clear
things up well enough that traffic can continue.  Else, let the AER/PCI
machinery do what is needed for the less serious errors seen.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: lif debugfs refresh on reset

Remove and restore the lif's debugfs pointers on a reset,
and make sure to check for the dentry before removing it
in case an earlier reset failed to rebuild the lif.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: use timer_shutdown_sync

When stopping the watchdog timer at remove time we should
be using the new timer_shutdown_sync to assure the timer
doesn't ever get rearmed.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: no fw read when PCI reset failed

If there was a failed attempt to reset the PCI connection,
don't later try to read from PCI as the space is unmapped
and will cause a paging request crash.  When clearing the PCI
setup we can clear the dev_info register pointer, and check
it before using it in the fw_running test.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: prevent pci disable of already disabled device

If a reset fails, the PCI device is left in a disabled
state, so don't try to disable it again on driver remove.
This prevents a scary looking WARN trace in the kernel log.

    ionic 0000:2b:00.0: disabling already-disabled device

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: bypass firmware cmds when stuck in reset

If the driver or firmware is stuck in reset state, don't bother
trying to use adminq commands.  This speeds up shutdown and
prevents unnecessary timeouts and error messages.

This includes a bit of rework on ionic_adminq_post_wait()
and ionic_adminq_post_wait_nomsg() to both use
__ionic_adminq_post_wait() which can do the checks needed in
both cases.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: keep filters across FLR

Make sure we keep and replay the filters and RSS config across
an FLR by using our FW_RESET flag.  This gets checked on the
way down and on the way back up to help determine how much LIF
state to keep and restore across a reset action.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ionic: pass opcode to devcmd_wait

Don't rely on the PCI memory for the devcmd opcode because we
read a 0xff value if the PCI bus is broken, which can cause us
to report a bogus dev_cmd opcode later.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: stmmac: mmc: Support more counters for XGMAC Core

Complete all counters on XGMAC Core.
These can be useful for debugging.

Signed-off-by: Furong Xu <0x1207@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'net-at803x-cleanups'

Christian Marangi says:

====================
net: phy: at803x: cleanup

The intention of this big series is to try to cleanup the big
at803x PHY driver.

It currently have 3 different family of PHY in it. at803x, qca83xx
and qca808x.

The current codebase required lots of cleanup and reworking to
make the split possible as currently there is a greater use of
adding special function matching the phy_id.

This has been reworked to make the function actually generic
and make the change only in more specific one. The result
is the addition of micro additional function but that is for good
as it massively simplify splitting the driver later.

Consider that this is all in preparation for the addition of
qca807x PHY driver that will also uso some of the functions of
at803x.

Subsequent series will come with the actual PHY split and other
required cleanup. This is only to start the process with minor
changes.

Changes v4:
- Improve at8031_probe function
Changes v3:
- Add Reviewed-by tag from Andrew
- Split patch 10 (at8031 rename) to rename and move
Changes v2:
- Drop split part due to series too big
- Split changes even more
- Fix problem pointed out by Russell (flawed reworked function logic)
- Add Reviewed-by tag from Andrew
- Minor rework to prevent further code duplication for cdt
====================

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

net: phy: at803x: drop specific PHY ID check from cable test functions

Drop specific PHY ID check for cable test functions for at803x. This is
done to make functions more generic. While at it better describe what
the functions does by using more symbolic function names.

PHYs that requires to set additional reg are moved to specific function
calling the more generic one.

cdt_start and cdt_wait_for_completion are changed to take an additional
arg to pass specific values specific to the PHY.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move at8035 specific DT parse to dedicated probe

Move at8035 specific DT parse for clock out frequency to dedicated probe
to make at803x probe function more generic.

This is to tidy code and no behaviour change are intended.

Detection logic is changed, we check if the clk 25m mask is set and if
it's not zero, we assume the qca,clk-out-frequency property is set.

The property is checked in the generic at803x_parse_dt called by
at803x_probe.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move at8031 functions in dedicated section

Move at8031 functions in dedicated section with dedicated at8031
parse_dt and probe.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: make at8031 related DT functions name more specific

Rename at8031 related DT function name to a more specific name
referencing they are only related to at8031 and not to the generic
at803x PHY family.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move specific at8031 config_intr to dedicated function

Move specific at8031 config_intr bits to dedicated function to make
at803x_config_initr more generic.

This is needed in preparation for PHY driver split as qca8081 share the
same function to setup interrupts.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move specific at8031 WOL bits to dedicated function

Move specific at8031 WOL enable/disable to dedicated function to make
at803x_set_wol more generic.

This is needed in preparation for PHY driver split as qca8081 share the
same function to toggle WOL settings.

In this new implementation WOL module in at8031 is enabled after the
generic interrupt is setup. This should not cause any problem as the
WOL_INT has a separate implementation and only relay on MAC bits.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move specific at8031 config_init to dedicated function

Move specific at8031 config_init to dedicated function to make
at803x_config_init more generic and tidy things up.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move specific at8031 probe mode check to dedicated probe

Move specific at8031 probe mode check to dedicated probe to make
at803x_probe more generic and keep code tidy.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move specific DT option for at8031 to specific probe

Move specific DT options for at8031 to specific probe to tidy things up
and make at803x_parse_dt more generic.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move qca83xx specific check in dedicated functions

Rework qca83xx specific check to dedicated function to tidy things up
and drop useless phy_id check.

Also drop an useless link_change_notify for QCA8337 as it did nothing an
returned early.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: raname hw_stats functions to qca83xx specific name

The function and the struct related to hw_stats were specific to qca83xx
PHY but were called following the convention in the driver of calling
everything with at803x prefix.

To better organize the code, rename these function a more specific name
to better describe that they are specific to 83xx PHY family.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: move disable WOL to specific at8031 probe

Move the WOL disable call to specific at8031 probe to make at803x_probe
more generic and drop extra check for PHY ID.

Keep the same previous behaviour by first calling at803x_probe and then
disabling WOL.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: fix passing the wrong reference for config_intr

Fix passing the wrong reference for config_initr on passing the function
pointer, drop the wrong & from at803x_config_intr in the PHY struct.

Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

dpll: remove leftover mode_supported() op and use mode_get() instead

Mode supported is currently reported to the user exactly the same, as
the current mode. That's because mode changing is not implemented.
Remove the leftover mode_supported() op and use mode_get() to fill up
the supported mode exposed to user.

One, if even, mode changing is going to be introduced, this could be
very easily taken back. In the meantime, prevent drivers form
implementing this in wrong way (as for example recent netdevsim
implementation attempt intended to do).

Signed-off-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'bnxt_en-update-for-net-next'

Michael Chan says:

====================
bnxt_en: Update for net-next

The first 4 patches in the series fix issues in the net-next tree
introduced in the last 4 weeks.  The first 3 patches fix ring accounting
and indexing logic.  The 4th patch fix TX timeout when the TX ring is
very small.

The next 7 patches add new features on the P7 chips, including TX
coalesced completions, VXLAN GPE and UDP GSO stateless offload, a
new rx_filter_miss counters, and more QP backing store memory for
RoCE.

The last 2 patches are PTP improvements.
====================

Link: https://lore.kernel.org/r/20231212005122.2401-1-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Make PTP TX timestamp HWRM query silent

In a busy network, especially with flow control enabled, we may
experience timestamp query failures fairly regularly. After a while,
dmesg may be flooded with timestamp query failure error messages.

Silence the error message from the low level hwrm function that
sends the firmware message.  Change netdev_err() to netdev_WARN_ONCE()
if this FW call ever fails.

Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-14-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Skip nic close/open when configuring tstamp filters

We don't have to close and open the nic to make sure we have
valid rx timestamps. Once we have the timestamp filter applied to
the HW and the timestamp_fld_format bit is cleared in the rx
completion and the timestamp is non-zero, we can be sure that rx
timestamp is valid data.

Skip close/open when we set any timestamp filter.

Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-13-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Add support for UDP GSO on 5760X chips

The new 5760X chips supports UDP GSO.  Tested using udpgso_bench_tx.

Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-12-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: add rx_filter_miss extended stats

rx_filter_miss counter is newly added to the rx_port_stats_ext
stats structure for newer chips.  Newer firmware will return the
structure size that includes this counter.  Add this entry to
the bnxt_port_stats_ext_arr array and the ethtool -S code will
pick up this counter if it is supported.

Signed-off-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-11-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Configure UDP tunnel TPA

On the new P7 chips, TPA for tunnel packets can be independently
enabled for each VNIC.  The default TPA configuration should not
include UDP tunnels because the UDP ports for these tunnels are not
known yet.  The chip should not aggregate these UDP tunneled packets
using default UDP ports until the ports are known.

Add a new function bnxt_hwrm_vnic_update_tunl_tpa() to enable VXLAN
and Geneve TPA if the corresponding UDP ports are known.

Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-10-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Add support for VXLAN GPE

Add a new bnxt_udp_tunnels_p7 struct to support the new P7 chips that
can parse VXLAN GPE packets.  Add VXLAN GPE tunnel type handling to
the .set_port() and .unset_port() functions.  .ndo_features_check()
is also enhanced to support VXLAN GPE which may encapsulate inner
IP packets instead of ethernet packets.

Reviewed-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com>
Reviewed-by: Ajit Khaparde <ajit.khaparde@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-9-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Use proper TUNNEL_DST_PORT_ALLOC* commands

In bnxt_udp_tunnel_set_port(), use the proper ALLOC commands instead
of the FREE commands for correctness.  The ALLOC and FREE commands
happen to be identical so this is just a cosmetic fix for correctness.

Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-8-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Allocate extra QP backing store memory when RoCE FW reports it

The Fast QP modify destroy RoCE feature requires additional QP entries
in QP context backing store. FW reports the extra count to be
allocated during backing store query. Use this value and allocate extra
memory.  Note that this works for both the V1 and V1 backing store
FW APIs.

Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-7-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Support TX coalesced completion on 5760X chips

TX coalesced completions are supported on newer chips to provide
one TX completion record for multiple TX packets up to the
sq_cons_idx in the completion record.  This method saves PCIe
bandwidth by reducing the number of TX completions.

Only very minor changes are now required to support this mode
with the new framework that handles TX completions based on
the consumer indices.

Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-6-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Prevent TX timeout with a very small TX ring

If xmit_more condition is true, the driver may set the
TX_BD_FLAGS_NO_CMPL flag.  If after this packet, the TX ring can no
longer hold a packet with maximum fragments, we will stop the TX
queue.  When this happens, we must clear the TX_BD_FLAGS_NO_CMPL flag
on the last packet or there will be no completion and cause TX
timeout.

Fixes: c1056a59aee1 ("bnxt_en: Optimize xmit_more TX path")
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Reviewed-by: Hongguang Gao <hongguang.gao@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-5-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Fix TX ring indexing logic

Two spots were missed when modifying the TX ring indexing logic.
The use of unmasked TX index in bnxt_tx_int() will cause unnecessary
__bnxt_tx_int() calls.  The same issue in bnxt_tx_int_xdp() can
result in illegal array index.

Fixes: 6d1add95536b ("bnxt_en: Modify TX ring indexing logic.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-4-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Fix AGG ring check logic in bnxt_check_rings()

_bnxt_get_max_rings() that is invoked in bnxt_check_rings() already
accounts for the AGG ring(s) and gives a max value based on that.
Increasing for AGG rings before calling _bnxt_get_max_rings() will
result in checking for twice the number of rings than required and
it can fail.  Fix it by adjusting for AGG rings after calling
_bnxt_get_max_rings().

Fixes: f5b29c6afe36 ("bnxt_en: Add helper to get the number of CP rings required for TX rings")
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-3-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

bnxt_en: Fix trimming of P5 RX and TX rings

The recent commit to trim the RX and TX rings on P5 chips by assigning
each with max CP rings divided by 2 is not correct.  Max CP rings
divided by 2 may be bigger than the original RX or TX and would
lead to failure.  In other words, we may be checking for increased
RX/TX rings than required and it may fail.

Fix it by calling __bnxt_trim_rings() instead that would properly
trim RX and TX without the possibility of increasing their values.

Fixes: f5b29c6afe36 ("bnxt_en: Add helper to get the number of CP rings required for TX rings")
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Link: https://lore.kernel.org/r/20231212005122.2401-2-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: mdio-gpio: replace deprecated strncpy with strscpy

strncpy() is deprecated for use on NUL-terminated destination strings
[1] and as such we should prefer more robust and less ambiguous string
interfaces.

We expect new_bus->id to be NUL-terminated but not NUL-padded based on
its prior assignment through snprintf:
|       snprintf(new_bus->id, MII_BUS_ID_SIZE, "gpio-%x", bus_id);

Due to this, a suitable replacement is `strscpy` [2] due to the fact
that it guarantees NUL-termination on the destination buffer without
unnecessarily NUL-padding.

We can also use sizeof() instead of a length macro as this more closely
ties the maximum buffer size to the destination buffer. Do this for two
instances.

Link: https://www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings
Link: https://manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html
Link: https://github.com/KSPP/linux/issues/90
Signed-off-by: Justin Stitt <justinstitt@google.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20231211-strncpy-drivers-net-mdio-mdio-gpio-c-v3-1-76dea53a1a52@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge tag 'pef2256-framer' of git://git./linux/kernel/git/linusw/linux-pinctrl

Linus Walleij says:

====================
Immutable tag for the PEF2256 framer

* tag 'pef2256-framer' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  MAINTAINERS: Add the Lantiq PEF2256 driver entry
  pinctrl: Add support for the Lantic PEF2256 pinmux
  net: wan: framer: Add support for the Lantiq PEF2256 framer
  dt-bindings: net: Add the Lantiq PEF2256 E1/T1/J1 framer
  net: wan: Add framer framework support
====================

Link: https://lore.kernel.org/all/CACRpkdYT1J7noFUhObFgfA60XQAfL4rb=knEmWS__TKKtCMh7Q@mail.gmail.com/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

MAINTAINERS: Add the Lantiq PEF2256 driver entry

After contributing the driver, add myself as the maintainer for the
Lantiq PEF2256 driver.

Signed-off-by: Herve Codina <herve.codina@bootlin.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Link: https://lore.kernel.org/r/20231128132534.258459-6-herve.codina@bootlin.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

pinctrl: Add support for the Lantic PEF2256 pinmux

The Lantiq PEF2256 is a framer and line interface component designed to
fulfill all required interfacing between an analog E1/T1/J1 line and the
digital PCM system highway/H.100 bus.

This kind of component can be found in old telecommunication system.
It was used to digital transmission of many simultaneous telephone calls
by time-division multiplexing. Also using HDLC protocol, WAN networks
can be reached through the framer.

This pinmux support handles the pin muxing part (pins RP(A..D) and pins
XP(A..D)) of the PEF2256.

Signed-off-by: Herve Codina <herve.codina@bootlin.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20231128132534.258459-5-herve.codina@bootlin.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

net: wan: framer: Add support for the Lantiq PEF2256 framer

The Lantiq PEF2256 is a framer and line interface component designed to
fulfill all required interfacing between an analog E1/T1/J1 line and the
digital PCM system highway/H.100 bus.

Signed-off-by: Herve Codina <herve.codina@bootlin.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20231128132534.258459-4-herve.codina@bootlin.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

dt-bindings: net: Add the Lantiq PEF2256 E1/T1/J1 framer

The Lantiq PEF2256 is a framer and line interface component designed to
fulfill all required interfacing between an analog E1/T1/J1 line and the
digital PCM system highway/H.100 bus.

Signed-off-by: Herve Codina <herve.codina@bootlin.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20231128132534.258459-3-herve.codina@bootlin.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

net: wan: Add framer framework support

A framer is a component in charge of an E1/T1 line interface.
Connected usually to a TDM bus, it converts TDM frames to/from E1/T1
frames. It also provides information related to the E1/T1 line.

The framer framework provides a set of APIs for the framer drivers
(framer provider) to create/destroy a framer and APIs for the framer
users (framer consumer) to obtain a reference to the framer, and
use the framer.

This basic implementation provides a framer abstraction for:
 - power on/off the framer
 - get the framer status (line state)
 - be notified on framer status changes
 - get/set the framer configuration

Signed-off-by: Herve Codina <herve.codina@bootlin.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20231128132534.258459-2-herve.codina@bootlin.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

net: asix: fix fortify warning

When compiling with gcc version 14.0.0 20231129 (experimental) and
CONFIG_FORTIFY_SOURCE=y, I've noticed the following warning:

...
In function 'fortify_memcpy_chk',
    inlined from 'ax88796c_tx_fixup' at drivers/net/ethernet/asix/ax88796c_main.c:287:2:
./include/linux/fortify-string.h:588:25: warning: call to '__read_overflow2_field'
declared with attribute warning: detected read beyond size of field (2nd parameter);
maybe use struct_group()? [-Wattribute-warning]
  588 |                         __read_overflow2_field(q_size_field, size);
      |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
...

This call to 'memcpy()' is interpreted as an attempt to copy TX_OVERHEAD
(which is 8) bytes from 4-byte 'sop' field of 'struct tx_pkt_info' and
thus overread warning is issued. Since we actually want to copy both
'sop' and 'seg' fields at once, use the convenient 'struct_group()' here.

Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Acked-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20231211090535.9730-1-dmantipov@yandex.ru
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-dsa-realtek-two-rtl8366rb-fixes'

Linus Walleij says:

====================
net: dsa: realtek: Two RTL8366RB fixes

These minor fixes were found while digging into other
issues: a weirdly named variable and bogus MTU handling.
Fix it up.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
====================

Link: https://lore.kernel.org/r/20231209-rtl8366rb-mtu-fix-v1-0-df863e2b2b2a@linaro.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: dsa: realtek: Rewrite RTL8366RB MTU handling

The MTU callbacks are in layer 1 size, so for example 1500
bytes is a normal setting. Cache this size, and only add
the layer 2 framing right before choosing the setting. On
the CPU port this will however include the DSA tag since
this is transmitted from the parent ethernet interface!

Add the layer 2 overhead such as ethernet and VLAN framing
and FCS before selecting the size in the register.

This will make the code easier to understand.

The rtl8366rb_max_mtu() callback returns a bogus MTU
just subtracting the CPU tag, which is the only thing
we should NOT subtract. Return the correct layer 1
max MTU after removing headers and checksum.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: dsa: realtek: Rename bogus RTL8368S variable

Rename the register name to RTL8366RB instead of the bogus
RTL8368S (internal product name?)

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Reviewed-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: dns_resolver: the module is called dns_resolver, not dnsresolver

$ modinfo dnsresolver dns_resolver | grep name
modinfo: ERROR: Module dnsresolver not found.
filename: /lib/modules/6.1.0-9-amd64/kernel/net/dns_resolver/dns_resolver.ko
name:     dns_resolver

Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Link: https://lore.kernel.org/r/gh4sxphjxbo56n2spgmc66vtazyxgiehpmv5f2gkvgicy6f4rs@tarta.nabijaczleweli.xyz
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: dl2k: Use proper conversion of dev_addr before IO to device

The driver is using iowriteXX()/ioreadXX() APIs which are LE IO
accessors simplified as

  1. Convert given value _from_ CPU _to_ LE
  2. Write it to the device as is

The dev_addr is a byte stream, but because the driver uses 16-bit
IO accessors, it wants to perform double conversion on BE CPUs,
but it took it wrong, as it effectivelly does two times _from_ CPU
_to_ LE. What it has to do is to consider dev_addr as an array of
LE16 and hence do _from_ LE _to_ CPU conversion, followed by implied
_from_ CPU _to_ LE in the iowrite16().

To achieve that, use get_unaligned_le16(). This will make it correct
and allows to avoid sparse warning as reported by LKP.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202312030058.hfZPTXd7-lkp@intel.com/
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20231208153327.3306798-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

netlink: specs: devlink: add some(not all) missing attributes in devlink.yaml

Add some missing(not all) attributes in devlink.yaml.

Signed-off-by: Swarup Laxman Kotiaklapudi <swarupkotikalapudi@gmail.com>
Suggested-by: Jiri Pirko <jiri@resnulli.us>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20231208182515.1206616-1-swarupkotikalapudi@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-sched-conditional-notification-of-events-for-cls-and-act'

Pedro Tammela says:

====================
net/sched: conditional notification of events for cls and act

This is an optimization we have been leveraging on P4TC but we believe
it will benefit rtnl users in general.

It's common to allocate an skb, build a notification message and then
broadcast an event. In the absence of any user space listeners, these
resources (cpu and memory operations) are wasted. In cases where the subsystem
is lockless (such as in tc-flower) this waste is more prominent. For the
scenarios where the rtnl_lock is held it is not as prominent.

The idea is simple. Build and send the notification iif:
   - The user requests via NLM_F_ECHO or
   - Someone is listening to the rtnl group (tc mon)

On a simple test with tc-flower adding 1M entries, using just a single core,
there's already a noticeable difference in the cycles spent in tc_new_tfilter
with this patchset.

before:
   - 43.68% tc_new_tfilter
      + 31.73% fl_change
      + 6.35% tfilter_notify
      + 1.62% nlmsg_notify
        0.66% __tcf_qdisc_find.part.0
        0.64% __tcf_chain_get
        0.54% fl_get
      + 0.53% tcf_proto_lookup_ops

after:
   - 39.20% tc_new_tfilter
      + 34.58% fl_change
        0.69% __tcf_qdisc_find.part.0
        0.67% __tcf_chain_get
      + 0.61% tcf_proto_lookup_ops

Note, the above test is using iproute2:tc which execs a shell.
We expect people using netlink directly to observe even greater
reductions.

The qdisc side needs some refactoring of the notification routines to fit in
this new model, so they will be sent in a later patchset.
====================

Link: https://lore.kernel.org/r/20231208192847.714940-1-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: cls_api: conditional notification of events

As of today tc-filter/chain events are unconditionally built and sent to
RTNLGRP_TC. As with the introduction of rtnl_notify_needed we can check
before-hand if they are really needed. This will help to alleviate
system pressure when filters are concurrently added without the rtnl
lock as in tc-flower.

Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Link: https://lore.kernel.org/r/20231208192847.714940-8-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: cls_api: remove 'unicast' argument from delete notification

This argument is never called while set to true, so remove it as there's
no need for it.

Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20231208192847.714940-7-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: act_api: conditional notification of events

As of today tc-action events are unconditionally built and sent to
RTNLGRP_TC. As with the introduction of rtnl_notify_needed we can check
before-hand if they are really needed.

Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20231208192847.714940-6-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/sched: act_api: don't open code max()

Use max() in a couple of places that are open coding it with the
ternary operator.

Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://lore.kernel.org/r/20231208192847.714940-5-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

rtnl: add helper to send if skb is not null

This is a convenience helper for routines handling conditional rtnl
events, that is code that might send a notification depending on
rtnl_has_listeners/rtnl_notify_needed.

Instead of:
   if (skb)
      rtnetlink_send(...)

Use:
      rtnetlink_maybe_send(...)

Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Link: https://lore.kernel.org/r/20231208192847.714940-4-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

rtnl: add helper to check if a notification is needed

Building on the rtnl_has_listeners helper, add the rtnl_notify_needed
helper to check if we can bail out early in the notification routines.

Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Link: https://lore.kernel.org/r/20231208192847.714940-3-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

rtnl: add helper to check if rtnl group has listeners

As of today, rtnl code creates a new skb and unconditionally fills and
broadcasts it to the relevant group. For most operations this is okay
and doesn't waste resources in general.

When operations are done without the rtnl_lock, as in tc-flower, such
skb allocation, message fill and no-op broadcasting can happen in all
cores of the system, which contributes to system pressure and wastes
precious cpu cycles when no one will receive the built message.

Introduce this helper so rtnetlink operations can simply check if someone
is listening and then proceed if necessary.

Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Victor Nogueira <victor@mojatatu.com>
Signed-off-by: Pedro Tammela <pctammela@mojatatu.com>
Link: https://lore.kernel.org/r/20231208192847.714940-2-pctammela@mojatatu.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'ipv6-data-races'

Eric Dumazet says:

====================
ipv6: more data-race annotations

Small follow up series, taking care of races around
np->mcast_oif and np->ucast_oif.
====================

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

ipv6: annotate data-races around np->ucast_oif

np->ucast_oif is read locklessly in some contexts.

Make all accesses to this field lockless, adding appropriate
annotations.

This also makes setsockopt( IPV6_UNICAST_IF ) lockless.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: annotate data-races around np->mcast_oif

np->mcast_oif is read locklessly in some contexts.

Make all accesses to this field lockless, adding appropriate
annotations.

This also makes setsockopt( IPV6_MULTICAST_IF ) lockless.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Revert "net: rtnetlink: remove local list in __linkwatch_run_queue()"

This reverts commit b8dbbbc535a9 ("net: rtnetlink: remove local list
in __linkwatch_run_queue()"). It's evidently broken when there's a
non-urgent work that gets added back, and then the loop can never
finish.

While reverting, add a note about that.

Reported-by: Marek Szyprowski <m.szyprowski@samsung.com>
Fixes: b8dbbbc535a9 ("net: rtnetlink: remove local list in __linkwatch_run_queue()")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

nfp: support UDP segmentation offload

The device supports UDP hardware segmentation offload, which helps
improving the performance. Thus, this patch adds support for UDP
segmentation offload from the driver side.

Signed-off-by: Fei Qin <fei.qin@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'rswitch-jumbo-frames'

Yoshihiro Shimoda says:

====================
net: rswitch: Add jumbo frames support

This patch series is based on the latest net-next.git / main branch.

Changes from v3:
https://lore.kernel.org/all/20231204012058.3876078-1-yoshihiro.shimoda.uh@renesas.com/
 - Based on the latest net-next.git / main branch.
 - Modify for code consistancy in the patch 3/9.
 - Add a condition in the patch 3/9.
 - Fix usage of dma_addr in the patch 8/9.

Changes from v2:
https://lore.kernel.org/all/20231201054655.3731772-1-yoshihiro.shimoda.uh@renesas.com/
 - Based on the latest net-next.git / main branch.
 - Fix using a variable in the patch 8/9.
 - Add Reviewed-by tag in the patch 1/9.

Changes from v1:
https://lore.kernel.org/all/20231127115334.3670790-1-yoshihiro.shimoda.uh@renesas.com/
 - Based on the latest net-next.git / main branch.
 - Fix commit descriptions (s/near the future/the near future/).
====================

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

net: rswitch: Allow jumbo frames

Allow jumbo frames by changing maximum MTU size and number of RX queues.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Add jumbo frames handling for TX

If the driver would like to transmit a jumbo frame like 2KiB or more,
it should be split into multiple queues. In the near future, to support
this, add handling specific descriptor types F{START,MID,END}. However,
such jumbo frames will not happen yet because the maximum MTU size is
still default for now.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Add jumbo frames handling for RX

If this hardware receives a jumbo frame like 2KiB or more, it will be
split into multiple queues. In the near future, to support this,
add handling specific descriptor types F{START,MID,END}. However, such
jumbo frames will not happen yet because the maximum MTU size is still
default for now.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Set GWMDNC register

To support jumbo frames, set GWMDNC register with acceptable maximum
values for TX and RX.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Add a setting ext descriptor function

If the driver would like to transmit a jumbo frame like 2KiB or more,
it should be split into multiple queues. In the near future, to support
this, add a setting ext descriptor function to improve code readability.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Add unmap_addrs instead of dma address in each desc

If the driver would like to transmit a jumbo frame like 2KiB or more,
it should be split into multiple queues. In the near future, to support
this, add unmap_addrs array to unmap dma mapping address instead of dma
address in each TX descriptor because the descriptors may not have
the top dma address.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Use build_skb() for RX

If this hardware receives a jumbo frame like 2KiB or more, it will be
split into multiple queues. In the near future, to support this, use
build_skb() instead of netdev_alloc_skb_ip_align().

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Use unsigned int for desc related array index

Array index should not be negative, so use unsigned int for
descriptors related array index.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: rswitch: Drop unused argument/return value

Drop unused argument and return value of rswitch_tx_free() to
simplify the code.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: mdio_bus: replace deprecated strncpy with strscpy

strncpy() is deprecated for use on NUL-terminated destination strings
[1] and as such we should prefer more robust and less ambiguous string
interfaces.

We expect mdiodev->modalias to be NUL-terminated based on its usage with
strcmp():
|       return strcmp(mdiodev->modalias, drv->name) == 0;

Moreover, mdiodev->modalias is already zero-allocated:
|       mdiodev = kzalloc(sizeof(*mdiodev), GFP_KERNEL);
... which means the NUL-padding strncpy provides is not necessary.

Considering the above, a suitable replacement is `strscpy` [2] due to
the fact that it guarantees NUL-termination on the destination buffer
without unnecessarily NUL-padding.

Link: https://www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings
Link: https://manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html
Link: https://github.com/KSPP/linux/issues/90
Cc: linux-hardening@vger.kernel.org
Signed-off-by: Justin Stitt <justinstitt@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: replace deprecated strncpy with strscpy

strncpy() is deprecated for use on NUL-terminated destination strings
[1] and as such we should prefer more robust and less ambiguous string
interfaces.

We expect fw_info->fw_file_name to be NUL-terminated based on its use
within _request_firmware_prepare() wherein `name` refers to it:
|       if (firmware_request_builtin_buf(firmware, name, dbuf, size)) {
|               dev_dbg(device, "using built-in %s\n", name);
|               return 0; /* assigned */
|       }
... and with firmware_request_builtin() also via `name`:
|       if (strcmp(name, b_fw->name) == 0) {

There is no evidence that NUL-padding is required.

Additionally replace size macro (QLC_FW_FILE_NAME_LEN) with
sizeof(fw_info->fw_file_name) to more directly tie the maximum buffer
size to the destination buffer:

Considering the above, a suitable replacement is `strscpy` [2] due to
the fact that it guarantees NUL-termination on the destination buffer
without unnecessarily NUL-padding.

Link: https://www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings
Link: https://manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html
Link: https://github.com/KSPP/linux/issues/90
Cc: linux-hardening@vger.kernel.org
Signed-off-by: Justin Stitt <justinstitt@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ena: replace deprecated strncpy with strscpy

`strncpy` is deprecated for use on NUL-terminated destination strings
[1] and as such we should prefer more robust and less ambiguous string
interfaces.

A suitable replacement is `strscpy` [2] due to the fact that it
guarantees NUL-termination on the destination buffer without
unnecessarily NUL-padding.

host_info allocation is done in ena_com_allocate_host_info() via
dma_alloc_coherent() and is not zero initialized by alloc_etherdev_mq().

However zero initialization of the destination doesn't matter in this case,
because strscpy() guarantees a NULL termination.

Link: https://www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings
Link: https://manpages.debian.org/testing/linux-manual-4.8/strscpy.9.en.html
Link: https://github.com/KSPP/linux/issues/90
Cc: linux-hardening@vger.kernel.org
Signed-off-by: Justin Stitt <justinstitt@google.com>
Acked-by: Arthur Kiyanovski <akiyano@amazon.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>