Linux supports 22 different interrupt coalescing parameters.
No driver implements them all. Some drivers just ignore the
ones they don't support, while others have to carry a long
list of checks to reject unsupported settings.

To simplify the drivers add the ability to specify inside
ethtool_ops which parameters are supported and let the core
reject attempts to set any other one.

This commit makes the mechanism an opt-in, only drivers which
set ethtool_opts->coalesce_types to a non-zero value will have
the checks enforced.

The same mask is used for global and per queue settings.

v3: - move the (temporary) check if driver defines types
      earlier (Michal)
    - rename used_types -> nonzero_params, and
      coalesce_types -> supported_coalesce_params (Alex)
    - use EOPNOTSUPP instead of EINVAL (Andrew, Michal)

Leaving the long series of ifs for now, it seems nice to
be able to grep for the field and flag names. This will
probably have to be revisited once netlink support lands.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In the commit e0a4b997 ("hsr: use upper/lower device infrastructure"),
dev_get() was removed but dev_put() in the error path wasn't removed.
So, if creating hsr interface command is failed, the reference counter leak
of lower interface would occur.

Test commands:
    ip link add dummy0 type dummy
    ip link add ipvlan0 link dummy0 type ipvlan mode l2
    ip link add ipvlan1 link dummy0 type ipvlan mode l2
    ip link add hsr0 type hsr slave1 ipvlan0 slave2 ipvlan1
    ip link del ipvlan0

Result:
[  633.271992][ T1280] unregister_netdevice: waiting for ipvlan0 to become free. Usage count = -1

Fixes: e0a4b997 ("hsr: use upper/lower device infrastructure")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Taehee Yoo says:

====================
net: rmnet: several code cleanup for rmnet module

This patchset is to cleanup rmnet module code.

1. The first patch is to add module alias
rmnet module can not be loaded automatically because there is no
alias name.

2. The second patch is to add extack error message code.
When rmnet netlink command fails, it doesn't print any error message.
So, users couldn't know the exact reason.
In order to tell the exact reason to the user, the extack error message
is used in this patch.

3. The third patch is to use GFP_KERNEL instead of GFP_ATOMIC.
In the sleepable context, GFP_KERNEL can be used.
So, in this patch, GFP_KERNEL is used instead of GFP_ATOMIC.

Change log:
 - v1->v2: change error message in the second patch.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

In the current code, rmnet_register_real_device() and rmnet_newlink()
are using GFP_ATOMIC.
But, these functions are allowed to sleep.
So, GFP_KERNEL can be used.
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When rmnet netlink command fails, it doesn't print any error message.
So, users couldn't know the exact reason.
In order to tell the exact reason to the user, the extack error message
is used in this patch.
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In the current rmnet code, there is no module alias.
So, RTNL couldn't load rmnet module automatically.

Test commands:
    ip link add dummy0 type dummy
    modprobe -rv rmnet
    ip link add rmnet0 link dummy0 type rmnet  mux_id 1
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Russell King says:

====================
marvell10g tunable and power saving support

This patch series adds support for:
- mdix configuration (auto, mdi, mdix)
- energy detect power down (edpd)
- placing in edpd mode at probe

for both the 88x3310 and 88x2110 PHYs.

Antione, could you test this for the 88x2110 PHY please?

v3: fix return code in get_tunable/set_tunable
v2: fix comments from Antione.
====================
Tested-by: Antoine Tenart <antoine.tenart@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Place the 88x3310 into powersaving mode when probing, which saves 600mW
per PHY. For both PHYs on the Macchiatobin double-shot, this saves
about 10% of the board idle power.
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add support for the energy detect power down tunable, which saves
around 600mW when the link is down. The 88x3310 supports off, rx-only
and NLP every second. Enable EDPD by default for 88x3310.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add support for controlling the MDI-X state of the PHY.
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'PCI-Add-and-use-constant-PCI_STATUS_ERROR_BITS-and-helper-pci_status_get_and_clear_errors'

Heiner Kallweit says:

====================
PCI: Add and use constant PCI_STATUS_ERROR_BITS and helper pci_status_get_and_clear_errors

Several drivers have own definitions for this constant, so move it
to the PCI core. In addition in multiple places the following code
sequence is used:
1. Read PCI_STATUS
2. Mask out non-error bits
3. Action based on set error bits
4. Write back set error bits to clear them

As this is a repeated pattern, add a helper to the PCI core.

Most affected drivers are network drivers. But as it's about core
PCI functionality, I suppose the series should go through the PCI
tree.

v2:
- fix formal issue with cover letter
v3:
- fix dumb typo in patch 7
v4:
- add patches 1-3
- move new constant PCI_STATUS_ERROR_BITS to include/linux/pci.h
- small improvements in commit messages
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Use new helper pci_status_get_and_clear_errors() to simplify the code.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use new constant PCI_STATUS_ERROR_BITS to simplify the code.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use new PCI core constant PCI_STATUS_ERROR_BITS to simplify the code.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use new helper pci_status_get_and_clear_errors() to simplify the code.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use new helper pci_status_get_and_clear_errors() to simplify the code.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Several drivers use the following code sequence:
1. Read PCI_STATUS
2. Mask out non-error bits
3. Action based on error bits set
4. Write back set error bits to clear them

As this is a repeated pattern, add a helper to the PCI core.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This collection of PCI error bits is used in more than one driver,
so move it to the PCI core.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In preparation of factoring out PCI_STATUS error bit handling let drivers
use the same collection of error bits. To facilitate bisecting we do this
in a separate patch per affected driver. For the r8169 driver we have to
add PCI_STATUS_PARITY to the error bits.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In preparation of factoring out PCI_STATUS error bit handling let drivers
use the same collection of error bits. To facilitate bisecting we do this
in a separate patch per affected driver. For the skfp driver we have to
add PCI_STATUS_REC_TARGET_ABORT to the error bits.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In preparation of factoring out PCI_STATUS error bit handling let drivers
use the same collection of error bits. To facilitate bisecting we do this
in a separate patch per affected driver. For the Marvell drivers we have
to add PCI_STATUS_SIG_TARGET_ABORT to the error bits.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Vladimir Oltean says:

====================
Allow unknown unicast traffic to CPU for Felix DSA

This is the continuation of the previous "[PATCH net-next] net: mscc:
ocelot: Workaround to allow traffic to CPU in standalone mode":

https://www.spinics.net/lists/netdev/msg631067.html

Following the feedback received from Allan Nielsen, the Ocelot and Felix
drivers were made to use the CPU port module in the same way (patch 1),
and Felix was made to additionally allow unknown unicast frames towards
the CPU port module (patch 2).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Compared to other DSA switches, in the Ocelot cores, the RX filtering is
a much more important concern.

Firstly, the primary use case for Ocelot is non-DSA, so there isn't any
secondary Ethernet MAC [the DSA master's one] to implicitly drop frames
having a DMAC we are not interested in.  So the switch driver itself
needs to install FDB entries towards the CPU port module (PGID_CPU) for
the MAC address of each switch port, in each VLAN installed on the port.
Every address that is not whitelisted is implicitly dropped. This is in
order to achieve a behavior similar to N standalone net devices.

Secondly, even in the secondary use case of DSA, such as illustrated by
Felix with the NPI port mode, that secondary Ethernet MAC is present,
but its RX filter is bypassed. This is because the DSA tags themselves
are placed before Ethernet, so the DMAC that the switch ports see is
not seen by the DSA master too (since it's shifter to the right).

So RX filtering is pretty important. A good RX filter won't bother the
CPU in case the switch port receives a frame that it's not interested
in, and there exists no other line of defense.

Ocelot is pretty strict when it comes to RX filtering: non-IP multicast
and broadcast traffic is allowed to go to the CPU port module, but
unknown unicast isn't. This means that traffic reception for any other
MAC addresses than the ones configured on each switch port net device
won't work. This includes use cases such as macvlan or bridging with a
non-Ocelot (so-called "foreign") interface. But this seems to be fine
for the scenarios that the Linux system embedded inside an Ocelot switch
is intended for - it is simply not interested in unknown unicast
traffic, as explained in Allan Nielsen's presentation [0].

On the other hand, the Felix DSA switch is integrated in more
general-purpose Linux systems, so it can't afford to drop that sort of
traffic in hardware, even if it will end up doing so later, in software.

Actually, unknown unicast means more for Felix than it does for Ocelot.
Felix doesn't attempt to perform the whitelisting of switch port MAC
addresses towards PGID_CPU at all, mainly because it is too complicated
to be feasible: while the MAC addresses are unique in Ocelot, by default
in DSA all ports are equal and inherited from the DSA master. This adds
into account the question of reference counting MAC addresses (delayed
ocelot_mact_forget), not to mention reference counting for the VLAN IDs
that those MAC addresses are installed in. This reference counting
should be done in the DSA core, and the fact that it wasn't needed so
far is due to the fact that the other DSA switches don't have the DSA
tag placed before Ethernet, so the DSA master is able to whitelist the
MAC addresses in hardware.

So this means that even regular traffic termination on a Felix switch
port happens through flooding (because neither Felix nor Ocelot learn
source MAC addresses from CPU-injected frames).

So far we've explained that whitelisting towards PGID_CPU:
- helps to reduce the likelihood of spamming the CPU with frames it
  won't process very far anyway
- is implemented in the ocelot driver
- is sufficient for the ocelot use cases
- is not feasible in DSA
- breaks use cases in DSA, in the current status (whitelisting enabled
  but no MAC address whitelisted)

So the proposed patch allows unknown unicast frames to be sent to the
CPU port module. This is done for the Felix DSA driver only, as Ocelot
seems to be happy without it.

[0]: https://www.youtube.com/watch?v=B1HhxEcU7JgSuggested-by: Allan W. Nielsen <allan.nielsen@microchip.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Allan W. Nielsen <allan.nielsen@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Ocelot has the concept of a CPU port. The CPU port is represented in the
forwarding and the queueing system, but it is not a physical device. The
CPU port can either be accessed via register-based injection/extraction
(which is the case of Ocelot), via Frame-DMA (similar to the first one),
or "connected" to a physical Ethernet port (called NPI in the datasheet)
which is the case of the Felix DSA switch.

In Ocelot the CPU port is at index 11.
In Felix the CPU port is at index 6.

The CPU bit is treated special in the forwarding, as it is never cleared
from the forwarding port mask (once added to it). Other than that, it is
treated the same as a normal front port.

Both Felix and Ocelot should use the CPU port in the same way. This
means that Felix should not use the NPI port directly when forwarding to
the CPU, but instead use the CPU port.

This patch is fixing this such that Felix will use port 6 as its CPU
port, and just use the NPI port to carry the traffic.

Therefore, eliminate the "ocelot->cpu" variable which was holding the
index of the NPI port for Felix, and the index of the CPU port module
for Ocelot, so the variable was actually configuring different things
for different drivers and causing at least part of the confusion.

Also remove the "ocelot->num_cpu_ports" variable, which is the result of
another confusion. The 2 CPU ports mentioned in the datasheet are
because there are two frame extraction channels (register based or DMA
based). This is of no relevance to the driver at the moment, and
invisible to the analyzer module.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Suggested-by: Allan W. Nielsen <allan.nielsen@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Leslie Monis says:

====================
pie: minor improvements

This patch series includes the following minor changes with
respect to the PIE/FQ-PIE qdiscs:

 - Patch 1 removes some ambiguity by using the term "backlog"
           instead of "qlen" when referring to the queue length
           in bytes.
 - Patch 2 removes redundant type casting on two expressions.
 - Patch 3 removes the pie_vars->accu_prob_overflows variable
           without affecting the precision in calculations and
           makes the size of the pie_vars structure exactly 64
           bytes.
 - Patch 4 realigns a comment affected by a change in patch 3.

Changes from v1 to v2:
 - Kept 8 as the argument to prandom_bytes() instead of changing it
   to 7 as suggested by David Miller.
====================
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Realign a comment after the change introduced by the
previous patch.
Signed-off-by: Leslie Monis <lesliemonis@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The variable pie_vars->accu_prob is used as an accumulator for
probability values. Since probabilty values are scaled using the
MAX_PROB macro denoting (2^64 - 1), pie_vars->accu_prob is
likely to overflow as it is of type u64.

The variable pie_vars->accu_prob_overflows counts the number of
times the variable pie_vars->accu_prob overflows.

The MAX_PROB macro needs to be equal to at least (2^39 - 1) in
order to do precise calculations without any underflow. Thus
MAX_PROB can be reduced to (2^56 - 1) without affecting the
precision in calculations drastically. Doing so will eliminate
the need for the variable pie_vars->accu_prob_overflows as the
variable pie_vars->accu_prob will never overflow.

Removing the variable pie_vars->accu_prob_overflows also reduces
the size of the structure pie_vars to exactly 64 bytes.
Signed-off-by: Mohit P. Tahiliani <tahiliani@nitk.edu.in>
Signed-off-by: Gautam Ramakrishnan <gautamramk@gmail.com>
Signed-off-by: Leslie Monis <lesliemonis@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In function pie_calculate_probability(), the variables alpha and
beta are of type u64. The variables qdelay, qdelay_old and
params->target are of type psched_time_t (which is also u64).
The explicit type casting done when calculating the value for
the variable delta is redundant and not required.
Signed-off-by: Mohit P. Tahiliani <tahiliani@nitk.edu.in>
Signed-off-by: Gautam Ramakrishnan <gautamramk@gmail.com>
Signed-off-by: Leslie Monis <lesliemonis@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove ambiguity by using the term backlog instead of qlen when
representing the queue length in bytes.
Signed-off-by: Mohit P. Tahiliani <tahiliani@nitk.edu.in>
Signed-off-by: Gautam Ramakrishnan <gautamramk@gmail.com>
Signed-off-by: Leslie Monis <lesliemonis@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Paul Blakey says:

====================
Fixes for tc act_ct software offload of established flows (diff v4->v6)

v4 of the original patchset was accidentally merged while we moved ahead
with v6 review. This two patches are the diff between v4 that was merged and
v6 that was the final revision, which was acked by the community.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

To make the filler functions more generic, use network
relative skb pulling.
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When checking the protocol number tcf_ct_flow_table_lookup() handles
the flow as if it's always ipv4, while it can be ipv6.

Instead, refactor the code to fetch the tcp header, if available,
in the relevant family (ipv4/ipv6) filler function, and do the
check on the returned tcp header.

Fixes: 46475bb2 ("net/sched: act_ct: Software offload of established flows")
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Core already zeroes out the struct ethtool_coalesce structure,
drivers don't have to set every field to 0 individually.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Vladimir Oltean says:

====================
Wire up Ocelot tc-flower to Felix DSA

This series is a proposal on how to wire up the tc-flower callbacks into
DSA. The example taken is the Microchip Felix switch, whose core
implementation is actually located in drivers/net/ethernet/mscc/.

The proposal is largely a compromise solution. The DSA middle layer
handles just enough to get to the interesting stuff (FLOW_CLS_REPLACE,
FLOW_CLS_DESTROY, FLOW_CLS_STATS), but also thin enough to let drivers
decide what filter keys and actions they support without worrying that
the DSA middle layer will grow exponentially. I am far from being an
expert, so I am asking reviewers to please voice your opinion if you
think it can be done differently, with better results.

The bulk of the work was actually refactoring the ocelot driver enough
to allow the VCAP (Versatile Content-Aware Processor) code for vsc7514
and the vsc9959 switch cores to live together.

Flow block offloads have not been tested yet, only filters attached to a
single port. It might be as simple as replacing ocelot_ace_rule_create
with something smarter, it might be more complicated, I haven't tried
yet.

I should point out that the tc-matchall filter offload is not
implemented in the same manner in current mainline. Florian has already
went all the way down into exposing actual per-action callbacks,
starting with port mirroring. Because currently only mirred is supported
by this DSA mid layer, everything else will return -EOPNOTSUPP. So even
though ocelot supports matchall (aka port-based) policers, we don't have
a call path to call into them.  Personally I think that this is not
going to scale for tc-matchall (there may be policers, traps, drops,
VLAN retagging, etc etc), and that we should consider whether further
matchall filter/action combinations should be just passed on to drivers
with no interpretation instead.
As for the existing mirroring callbacks in DSA, they can either be kept
as-is, or replaced with simple accessors to TC_CLSMATCHALL_REPLACE and
TC_CLSMATCHALL_DESTROY, just like for flower, and drivers which
currently implement the port mirroring callbacks will need to have some
extra "if" conditions now, in order for them to call their port
mirroring implementations.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Export the cls_flower methods from the ocelot driver and hook them up to
the DSA passthrough layer.

Tables for the VCAP IS2 parameters, as well as half key packing (field
offsets and lengths) need to be defined for the VSC9959 core, as they
are different from Ocelot, mainly due to the different port count.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Due to the immense variety of classification keys and actions available
for tc-flower, as well as due to potentially very different DSA switch
capabilities, it doesn't make a lot of sense for the DSA mid layer to
even attempt to interpret these. So just pass them on to the underlying
switch driver.

DSA implements just the standard boilerplate for binding and unbinding
flow blocks to ports, since nobody wants to deal with that.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove the definitions for the VCAP IS2 table from ocelot_ace.c, since
it is specific to VSC7514.

The VSC9959 VCAP IS2 table supports more rules (1024 instead of 64) and
has a different width for the action (89 bits instead of 99).
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The Felix driver is now using its own PHYLINK instance, not calling into
ocelot_adjust_link. So the port_pcs_init function pointer is an
unnecessary indirection. Remove it.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Reviewed-by: Allan W. Nielsen <allan.nielsen@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The IGR_PORT_MASK key width is different between the 11-port VSC7514 and
the 6-port VSC9959 switches. And since IGR_PORT_MASK is one of the first
fields of a VCAP key entry, it means that all further field
offset/length pairs are shifted between the 2.

The ocelot driver performs packing of VCAP half keys with the help of
some preprocessor macros:

- A set of macros for defining the HKO (Half Key Offset) and HKL (Half
  Key Length) of each possible key field. The offset of each field is
  defined as the sum between the offset and the sum of the previous
  field.

- A set of accessors on top of vcap_key_set for shorter (aka less
  typing) access to the HKO and HKL of each key field.

Since the field offsets and lengths are different between switches,
defining them through the preprocessor isn't going to fly. So introduce
a structure holding (offset, length) pairs and instantiate it in
ocelot_board.c for VSC7514. In a future patch, a similar structure will
be instantiated in felix_vsc9959.c for NXP LS1028A.

The accessors also need to go. They are based on macro name
concatenation, which is horrible to understand and follow.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Tested-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This is a cosmetic patch that makes the name of the driver private
variable be used uniformly in ocelot_ace.c as in the rest of the driver.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>