platform/kernel/linux-starfive.git
6 years agoMerge branch 'More-complete-PHYLINK-support-for-mv88e6xxx'
David S. Miller [Thu, 9 Aug 2018 18:08:21 +0000 (11:08 -0700)]
Merge branch 'More-complete-PHYLINK-support-for-mv88e6xxx'

Andrew Lunn says:

====================
More complete PHYLINK support for mv88e6xxx

Previous patches added sufficient PHYLINK support to the mv88e6xxx
that it did not break existing use cases, basically fixed-link phys.

This patchset builds out the support so that SFP modules, up to
2.5Gbps can be supported, on mv88e6390X, on ports 9 and 10. It also
provides a framework which can be extended to support SFPs on ports
2-8 of mv88e6390X, 10Gbps PHYs, and SFP support on the 6352 family.

Russell King did much of the initial work, implementing the validate
and mac_link_state calls. However, there is an important TODO in the
commit message:

needs to call phylink_mac_change() when the port link comes up/goes down.

The remaining patches implement this, by adding more support for the
SERDES interfaces, in particular, interrupt support so we get notified
when the SERDES gains/looses sync.

This has been tested on the ZII devel C, using a Clearfog as peer
device.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Re-setup interrupts on CMODE change.
Andrew Lunn [Thu, 9 Aug 2018 13:38:49 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Re-setup interrupts on CMODE change.

When a port changes CMODE, the SERDES interface being used can change.
Disable interrupts for the old SERDES interface, and enable interrupts
on the new.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Add SERDES phydev_mac_change up for 6390
Andrew Lunn [Thu, 9 Aug 2018 13:38:48 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Add SERDES phydev_mac_change up for 6390

phylink wants to know when the MAC layers notices a change in the
link. For the 6390 family, this is a change in the SERDES state.

Add interrupt support for the SERDES interface used to implement
SGMII/1000Base-X/2500Base-X. This is currently limited to ports 9 and
10. Support for the 10G SERDES and other ports will be added later,
building on this basic framework.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: link mv88e6xxx_port to mv88e6xxx_chip
Andrew Lunn [Thu, 9 Aug 2018 13:38:47 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: link mv88e6xxx_port to mv88e6xxx_chip

An up coming change will register interrupts for individual switch
ports, using the mv88e6xxx_port as the interrupt context information.
Add members to the mv88e6xxx_port structure so we can link it back to
the mv88e6xxx_chip member the port belongs to and the port number of
the port.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Power on/off SERDES on cmode change
Andrew Lunn [Thu, 9 Aug 2018 13:38:46 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Power on/off SERDES on cmode change

The 6390 family has a number of SERDES interfaces per port. When the
cmode changes, eg 1000Base-X to XAUI, the SERDES interface in use will
also change. Power down the old SERDES interface and power up the new
SERDES interface.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Cache the port cmode
Andrew Lunn [Thu, 9 Aug 2018 13:38:45 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Cache the port cmode

The ports CMODE indicates the type of link between the MAC and the
PHY. It is used often in the SERDES code. Rather than read it each
time, cache its value.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: 2500Base-X uses the 1000Base-X SERDES
Andrew Lunn [Thu, 9 Aug 2018 13:38:44 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: 2500Base-X uses the 1000Base-X SERDES

The 6390 has three different SERDES interface types. 2500Base-X is
implemented by the SGMII/1000Base-X SERDES. So power on/off the
correct SERDES.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Add serdes register read/write helper
Andrew Lunn [Thu, 9 Aug 2018 13:38:43 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Add serdes register read/write helper

Add a helper for accessing SERDES registers of the 6390 family.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Rename sgmii/10g power functions
Andrew Lunn [Thu, 9 Aug 2018 13:38:42 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Rename sgmii/10g power functions

There is a need to add more functions manipulating the SERDES
interfaces. Cleanup the namespace.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: 6390 vs 6390X SERDES support
Andrew Lunn [Thu, 9 Aug 2018 13:38:41 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: 6390 vs 6390X SERDES support

The 6390 has two SERDES interfaces, used by ports 9 and 10.  The 6390X
has eight SERDES interfaces. These allow ports 9 and 10 to do 10G. Or
if lower speeds are used, some of the SERDES interfaces can be used by
ports 2-8 for 1000Base-X.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Refactor SERDES lane code
Andrew Lunn [Thu, 9 Aug 2018 13:38:40 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Refactor SERDES lane code

The 6390 family has 8 SERDES lanes. What ports use these lanes depends
on how ports 9 and 10 are configured. If 9 and 10 does not make use of
a line, one of the lower ports can use it.

Add a function to return the lane a port is using, if any, and simplify
the code to power up/down the lane.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: add phylink support
Russell King [Thu, 9 Aug 2018 13:38:39 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: add phylink support

Add rudimentary phylink support to mv88e6xxx.

TODO:
- needs to call phylink_mac_change() when the port link comes up/goes down.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agophylink: add helper for configuring 2500BaseX modes
Russell King [Thu, 9 Aug 2018 13:38:38 +0000 (15:38 +0200)]
phylink: add helper for configuring 2500BaseX modes

Add a helper for MAC drivers to use in their validate callback to deal
with 2500BaseX vs 1000BaseX modes, where the hardware supports both
but it is not possible to automatically select between them.

This helper defaults to 1000BaseX, as that is the 802.3 standard, and
will allow users to select 2500BaseX either by forcing the speed if
AN is disabled, or by changing the advertising mask if AN is enabled.
Disabling AN is not recommended as it is only the speed that we're
interested in controlling, not the duplex or pause mode parameters.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: mv88e6xxx: Add support to enabling pause
Andrew Lunn [Thu, 9 Aug 2018 13:38:37 +0000 (15:38 +0200)]
net: dsa: mv88e6xxx: Add support to enabling pause

The 6185 can enable/disable 802.3z pause be setting the MyPause bit in
the port status register. Add an op to support this.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'mlxsw-Various-updates'
David S. Miller [Thu, 9 Aug 2018 17:36:11 +0000 (10:36 -0700)]
Merge branch 'mlxsw-Various-updates'

Ido Schimmel says:

====================
mlxsw: Various updates

Patches 1-3 update the driver to use a new firmware version. Due to a
recently discovered issue, the version (and future ones) does not
support matching on VLAN ID at egress. This is enforced in the driver
and reported back to the user via extack.

Patch 4 adds a new selftest for the recently introduced algorithmic
TCAM.

Patch 5 converts the driver to use SPDX identifiers.

Patches 6-7 fix a bug in ethtool stats reporting and expose counters for
all 16 TCs, following recent MC-aware changes that utilize TCs 8-15.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: spectrum: Expose counter for all 16 TCs
Petr Machata [Thu, 9 Aug 2018 08:59:13 +0000 (11:59 +0300)]
mlxsw: spectrum: Expose counter for all 16 TCs

Before MC-aware mode was enabled in commit 7b8195306694 ("mlxsw:
spectrum: Configure MC-aware mode on mlxsw ports"), only 8 traffic
classes were used. Under MC-aware regime, however, besides using TCs
0-7 for UC traffic, it additionally uses TCs 8-15 for BUM traffic. It
is therefore desirable to show counters for these TCs as well.

Update ethtool stats pool length, mlxsw_sp_port_get_strings() and
mlxsw_sp_port_get_stats() to include artifacts for all 16 TCs. For
consistency and simplicity, expose tc_no_buffer_discard_uc_tc for BUM
TCs as well, even though it ought to stay at 0 all the time.

Signed-off-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: spectrum: Include RFC-2819 counters in stats length
Petr Machata [Thu, 9 Aug 2018 08:59:12 +0000 (11:59 +0300)]
mlxsw: spectrum: Include RFC-2819 counters in stats length

The function mlxsw_sp_port_get_sset_count() is supposed to return the
total number of ethtool strings that mlxsw supports. Specifically for
names of statistic counters (the only string type that mlxsw supports
as of now), that number is stored in MLXSW_SP_PORT_ETHTOOL_STATS_LEN.
However, when adding RFC-2891 counters, that define wasn't updated to
include the new counters. As a result, ethtool snips out the counters
towards the end of the list, which contains per-TC counters, and only
the first three traffic classes end up being reported.

Fix by adding MLXSW_SP_PORT_HW_RFC_2819_STATS_LEN as appropriate.

Fixes: 1222d15a01c7 ("mlxsw: spectrum: Expose counters for various packet sizes")
Signed-off-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: Replace license text with SPDX identifiers and adjust copyrights
Jiri Pirko [Thu, 9 Aug 2018 08:59:11 +0000 (11:59 +0300)]
mlxsw: Replace license text with SPDX identifiers and adjust copyrights

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoselftests: mlxsw: Add TC flower test for Spectrum-2
Ido Schimmel [Thu, 9 Aug 2018 08:59:10 +0000 (11:59 +0300)]
selftests: mlxsw: Add TC flower test for Spectrum-2

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: spectrum: Reset FW after flash
Jiri Pirko [Thu, 9 Aug 2018 08:59:09 +0000 (11:59 +0300)]
mlxsw: spectrum: Reset FW after flash

Recent FW fixes a bug and allows to load newly flashed FW image after
reset. So make sure the reset happens after flash. Indicate the need
down to PCI layer by -EAGAIN.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: spectrum: Update the supported firmware to version 13.1702.6
Nir Dotan [Thu, 9 Aug 2018 08:59:08 +0000 (11:59 +0300)]
mlxsw: spectrum: Update the supported firmware to version 13.1702.6

This new firmware contains:
        - Support for new types of cables
        - Support for flashing future firmware without reboot
        - Support for Router ARP BC and UC traps

Signed-off-by: Nir Dotan <nird@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agomlxsw: spectrum_flower: Disallow usage of vlan_id key on egress
Nir Dotan [Thu, 9 Aug 2018 08:59:07 +0000 (11:59 +0300)]
mlxsw: spectrum_flower: Disallow usage of vlan_id key on egress

As recent spectrum FW imposes a limitation on using vlan_id key for
egress ACL, disallow the usage of that key accordingly and return a
proper extack message.

Signed-off-by: Nir Dotan <nird@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'mlx5-next'
David S. Miller [Thu, 9 Aug 2018 02:34:55 +0000 (19:34 -0700)]
Merge branch 'mlx5-next'

Saeed Mahameed says:

====================
Mellanox, mlx5 next updates 2018-08-09

This series includes mlx5 core driver updates and mostly simple
cleanups.

From Denis: Use max #EQs reported by firmware to request MSIX vectors.

From Eli: Trivial cleanups, unused arguments/functions and reduce
command polling interval when command interface is in polling mode.

From Eran: Rename vport state enums, to better reflect their actual
usage.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5: Reduce command polling interval
Eli Cohen [Wed, 8 Aug 2018 23:23:53 +0000 (16:23 -0700)]
net/mlx5: Reduce command polling interval

Use cond_resched() instead of usleep_range() to decrease the time
between polling attempts thus reducing overall driver load time.

Below is a comparison before and after the change, of loading eight
virtual functions.

Before:
real    0m8.785s
user    0m0.093s
sys     0m0.090s

After:
real    0m5.730s
user    0m0.097s
sys     0m0.087s

Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5: Unexport functions that need not be exported
Eli Cohen [Wed, 8 Aug 2018 23:23:52 +0000 (16:23 -0700)]
net/mlx5: Unexport functions that need not be exported

mlx5_query_vport_state() and mlx5_modify_vport_admin_state() are used
only from within mlx5_core  - unexport them.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5: Remove unused mlx5_query_vport_admin_state
Eli Cohen [Wed, 8 Aug 2018 23:23:51 +0000 (16:23 -0700)]
net/mlx5: Remove unused mlx5_query_vport_admin_state

mlx5_query_vport_admin_state() is not used anywhere. Remove it.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5: E-Switch, Remove unused argument when creating legacy FDB
Eli Cohen [Wed, 8 Aug 2018 23:23:50 +0000 (16:23 -0700)]
net/mlx5: E-Switch, Remove unused argument when creating legacy FDB

Remove unused nvports argument.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5: Rename modify/query_vport state related enums
Eran Ben Elisha [Wed, 8 Aug 2018 23:23:49 +0000 (16:23 -0700)]
net/mlx5: Rename modify/query_vport state related enums

Modify and query vport state commands share the same admin_state and
op_mod values, rename the enums to fit them both.

In addition, remove the esw prefix from the admin state enum as this
also applied for vnic.

Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5: Use max_num_eqs for calculation of required MSIX vectors
Denis Drozdov [Wed, 8 Aug 2018 23:23:48 +0000 (16:23 -0700)]
net/mlx5: Use max_num_eqs for calculation of required MSIX vectors

New firmware has defined new HCA capability field called "max_num_eqs",
that is the number of available EQs after subtracting reserved FW EQs.

Before this capability the FW reported the EQ number in "log_max_eqs",
the reported value also contained FW reserved EQs, but the driver might
be failing to load on 320 cpus systems due to the fact that FW
reserved EQs were not available to the driver.

Now the driver has to obtain max_num_eqs value from new FW to get real
number of EQs available.

Signed-off-by: Denis Drozdov <denisd@mellanox.com>
Reviewed-by: Alex Vesker <valex@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoieee802154: hwsim: fix missing unlock on error in hwsim_add_one()
Wei Yongjun [Wed, 8 Aug 2018 03:10:39 +0000 (03:10 +0000)]
ieee802154: hwsim: fix missing unlock on error in hwsim_add_one()

Add the missing unlock before return from function hwsim_add_one()
in the error handling case.

Fixes: f25da51fdc38 ("ieee802154: hwsim: add replacement for fakelb")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoieee802154: hwsim: fix copy-paste error in hwsim_set_edge_lqi()
Wei Yongjun [Wed, 8 Aug 2018 02:43:46 +0000 (02:43 +0000)]
ieee802154: hwsim: fix copy-paste error in hwsim_set_edge_lqi()

The return value from kzalloc() is not checked correctly. The
test is done against a wrong variable. This patch fix it.

Fixes: f25da51fdc38 ("ieee802154: hwsim: add replacement for fakelb")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoieee802154: hwsim: fix rcu handling
Alexander Aring [Tue, 7 Aug 2018 23:32:49 +0000 (19:32 -0400)]
ieee802154: hwsim: fix rcu handling

This patch adds missing rcu_assign_pointer()/rcu_dereference() to used rcu
pointers. There was already a previous commit c5d99d2b35da ("ieee802154:
hwsim: fix rcu address annotation"), but there was more which was
pointed out on my side by using newest sparse version.

Cc: Stefan Schmidt <stefan@datenfreihafen.org>
Fixes: f25da51fdc38 ("ieee802154: hwsim: add replacement for fakelb")
Signed-off-by: Alexander Aring <aring@mojatatu.com>
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet-next: hinic: fix a problem in free_tx_poll()
Zhao Chen [Wed, 8 Aug 2018 06:37:30 +0000 (06:37 +0000)]
net-next: hinic: fix a problem in free_tx_poll()

This patch fixes the problem below. The problem can be reproduced by the
following steps:
1) Connecting all HiNIC interfaces
2) On server side
    # sudo ifconfig eth0 192.168.100.1 up #Using MLX CX4 card
    # iperf -s
3) On client side
    # sudo ifconfig eth0 192.168.100.2 up #Using our HiNIC card
    # iperf -c 192.168.101.1 -P 10 -t 100000

after hours of testing, we will see errors:

    hinic 0000:05:00.0: No MGMT msg handler, mod = 0
    hinic 0000:05:00.0: No MGMT msg handler, mod = 0
    hinic 0000:05:00.0: No MGMT msg handler, mod = 0
    hinic 0000:05:00.0: No MGMT msg handler, mod = 0

The errors are caused by the following problem.
1) The hinic_get_wqe() checks the "wq->delta" to allocate new WQEs:

if (atomic_sub_return(num_wqebbs, &wq->delta) <= 0) {
atomic_add(num_wqebbs, &wq->delta);
return ERR_PTR(-EBUSY);
}

If the WQE occupies multiple pages, the shadow WQE will be used. Then the
hinic_xmit_frame() fills the WQE.

2) While in parallel with 1), the free_tx_poll() checks the "wq->delta"
to free old WQEs:

if ((atomic_read(&wq->delta) + num_wqebbs) > wq->q_depth)
return ERR_PTR(-EBUSY);

There is a probability that the shadow WQE which hinic_xmit_frame() is
using will be damaged by copy_wqe_to_shadow():

if (curr_pg != end_pg) {
void *shadow_addr = &wq->shadow_wqe[curr_pg * wq->max_wqe_size];

copy_wqe_to_shadow(wq, shadow_addr, num_wqebbs, *cons_idx);
return shadow_addr;
}

This can cause WQE data error and you will see the above error messages.
This patch fixes the problem.

Signed-off-by: Zhao Chen <zhaochen6@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet:mod: remove unneeded variable 'ret' in init_p9
zhong jiang [Tue, 7 Aug 2018 11:20:09 +0000 (19:20 +0800)]
net:mod: remove unneeded variable 'ret' in init_p9

The ret is modified after initalization, so just remove it and
return 0.

Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet:af_iucv: get rid of the unneeded variable 'err' in afiucv_pm_freeze
zhong jiang [Tue, 7 Aug 2018 11:20:08 +0000 (19:20 +0800)]
net:af_iucv: get rid of the unneeded variable 'err' in afiucv_pm_freeze

We will not use the variable 'err' after initalization, So remove it and
return 0.

Signed-off-by: zhong jiang <zhongjiang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: nixge: Get rid of unused struct member 'last_link'
Moritz Fischer [Tue, 7 Aug 2018 23:35:20 +0000 (16:35 -0700)]
net: nixge: Get rid of unused struct member 'last_link'

Get rid of unused struct member 'last_link'

Signed-off-by: Moritz Fischer <mdf@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'net-ethernet-Mark-expected-switch-fall-throughs'
David S. Miller [Wed, 8 Aug 2018 00:54:21 +0000 (17:54 -0700)]
Merge branch 'net-ethernet-Mark-expected-switch-fall-throughs'

Gustavo A. R. Silva says:

====================
net: ethernet: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, this patchset aims
to add some annotations in order to mark switch cases where we are
expecting to fall through.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: ti: cpts: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:32:19 +0000 (18:32 -0500)]
net: ethernet: ti: cpts: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114813 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: tlan: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:31:46 +0000 (18:31 -0500)]
net: tlan: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 141440 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sfc: falcon: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:31:16 +0000 (18:31 -0500)]
net: sfc: falcon: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 1384500 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: ethernet: sxgbe: mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:30:22 +0000 (18:30 -0500)]
net: ethernet: sxgbe: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 1357414 ("Missing break in switch")
Addresses-Coverity-ID: 1357415 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqlge: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:29:12 +0000 (18:29 -0500)]
qlge: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114811 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqlcnic: Mark expected switch fall-througs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:28:20 +0000 (18:28 -0500)]
qlcnic: Mark expected switch fall-througs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 1410181 ("Missing break in switch")
Addresses-Coverity-ID: 1410184 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqede: qede_fp: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:27:36 +0000 (18:27 -0500)]
qede: qede_fp: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 1384501 ("Missing break in switch")
Addresses-Coverity-ID: 1398869 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonetxen_nic: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:26:49 +0000 (18:26 -0500)]
netxen_nic: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 1410182 ("Missing break in switch")
Addresses-Coverity-ID: 1410183 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqed: qed_dev: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:25:57 +0000 (18:25 -0500)]
qed: qed_dev: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Notice that in this particular case, I replaced the code comments with
a proper "fall through" annotation, which is what GCC is expecting
to find.

Addresses-Coverity-ID: 114809 ("Missing break in switch")
Addresses-Coverity-ID: 114810 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx5e: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:25:26 +0000 (18:25 -0500)]
net/mlx5e: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114808 ("Missing break in switch")
Addresses-Coverity-ID: 114802 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agovxge: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:24:38 +0000 (18:24 -0500)]
vxge: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114796 ("Missing break in switch")
Addresses-Coverity-ID: 114804 ("Missing break in switch")
Addresses-Coverity-ID: 114806 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigbvf: netdev: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:24:04 +0000 (18:24 -0500)]
igbvf: netdev: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114801 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: e1000_phy: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:23:31 +0000 (18:23 -0500)]
igb: e1000_phy: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114800 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb: e1000_82575: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:22:57 +0000 (18:22 -0500)]
igb: e1000_82575: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114799 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoigb_main: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:22:22 +0000 (18:22 -0500)]
igb_main: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 200521 ("Missing break in switch")
Addresses-Coverity-ID: 114797 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx4/en_rx: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:21:40 +0000 (18:21 -0500)]
net/mlx4/en_rx: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114794 ("Missing break in switch")
Addresses-Coverity-ID: 114795 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/mlx4/mcg: Mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:21:05 +0000 (18:21 -0500)]
net/mlx4/mcg: Mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114792 ("Missing break in switch")
Addresses-Coverity-ID: 114793 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoi40e_txrx: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:20:27 +0000 (18:20 -0500)]
i40e_txrx: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114791 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoi40e_main: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:19:42 +0000 (18:19 -0500)]
i40e_main: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114790 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns3: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:18:30 +0000 (18:18 -0500)]
net: hns3: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114789 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: hns: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:17:50 +0000 (18:17 -0500)]
net: hns: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114788 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agobe2net: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:17:08 +0000 (18:17 -0500)]
be2net: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114787 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: tulip: de4x5: mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:16:07 +0000 (18:16 -0500)]
net: tulip: de4x5: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114784 ("Missing break in switch")
Addresses-Coverity-ID: 114785 ("Missing break in switch")
Addresses-Coverity-ID: 114786 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: tulip_core: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:15:35 +0000 (18:15 -0500)]
net: tulip_core: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114782 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: thunderx: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:15:32 +0000 (18:15 -0500)]
net: thunderx: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114781 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocxgb3/l2t: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:14:18 +0000 (18:14 -0500)]
cxgb3/l2t: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114780 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocxgb4/t4_hw: mark expected switch fall-throughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:13:44 +0000 (18:13 -0500)]
cxgb4/t4_hw: mark expected switch fall-throughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114777 ("Missing break in switch")
Addresses-Coverity-ID: 114778 ("Missing break in switch")
Addresses-Coverity-ID: 114779 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agocxgb4/l2t: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:13:12 +0000 (18:13 -0500)]
cxgb4/l2t: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114910 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoliquidio: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:13:05 +0000 (18:13 -0500)]
liquidio: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 143135 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: macb: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:11:26 +0000 (18:11 -0500)]
net: macb: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agobnx2x: Mark expected switch fall-thoughs
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:11:14 +0000 (18:11 -0500)]
bnx2x: Mark expected switch fall-thoughs

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114878 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoalteon: acenic: mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:09:19 +0000 (18:09 -0500)]
alteon: acenic: mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114891 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years ago8390: axnet_cs: Mark expected switch fall-through
Gustavo A. R. Silva [Tue, 7 Aug 2018 23:09:09 +0000 (18:09 -0500)]
8390: axnet_cs: Mark expected switch fall-through

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114889 ("Missing break in switch")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoselftests: forwarding: gre_multipath: Update next-hop statistics match criteria
Nir Dotan [Tue, 7 Aug 2018 16:41:55 +0000 (19:41 +0300)]
selftests: forwarding: gre_multipath: Update next-hop statistics match criteria

gre_multipath test was using egress vlan_id matching on flows, for the
purpose of collecting next-hops statistics, later to be compared
against configured weights.
As matching on vlan_id on egress direction is not supported on all HW
devices, change the match criteria to use destination IP.

Signed-off-by: Nir Dotan <nird@mellanox.com>
Acked-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agotc-tests: initial version of nat action unit tests
Keara Leibovitz [Tue, 7 Aug 2018 19:18:43 +0000 (15:18 -0400)]
tc-tests: initial version of nat action unit tests

Initial set of nat action unit tests.

Signed-off-by: Keara Leibovitz <kleib@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'brcm-omega'
David S. Miller [Tue, 7 Aug 2018 22:48:38 +0000 (15:48 -0700)]
Merge branch 'brcm-omega'

Arun Parameswaran says:

====================
Add Broadcom Omega SoC internal switch and phy

The patchset is based on David Miller's "net-next" repo.

The patches add support for the Broadcom Omega SoC's internal ethernet
switch and the internal gphy.

The internal ethernet switch in the Omega is a b53 srab based switch.
The support for the switch is added to the b53 driver in the dsa
framework.

The gphy support is added to the bcm7xxx driver.
====================

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: phy: Add support for Broadcom Omega internal Combo GPHY
Arun Parameswaran [Tue, 7 Aug 2018 17:02:44 +0000 (10:02 -0700)]
net: phy: Add support for Broadcom Omega internal Combo GPHY

Add support for the Broadcom Omega SoC internal Combo Ethernet
GPHY to the bcm7xxx phy driver.

Signed-off-by: Arun Parameswaran <arun.parameswaran@broadcom.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: dsa: b53: Add support for Broadcom Omega SoC internal switch
Arun Parameswaran [Tue, 7 Aug 2018 17:02:43 +0000 (10:02 -0700)]
net: dsa: b53: Add support for Broadcom Omega SoC internal switch

Add support for the Broadcom Omega SoC internal ethernet switch
to the b53 srab driver in the DSA framework.

Signed-off-by: Arun Parameswaran <arun.parameswaran@broadcom.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agodt-bindings: net: dsa: Add compatibility strings for Broadcom Omega
Arun Parameswaran [Tue, 7 Aug 2018 17:02:42 +0000 (10:02 -0700)]
dt-bindings: net: dsa: Add compatibility strings for Broadcom Omega

Add compatibility strings for the internal switch in the Broadcom
Omega SoC family (BCM5831X/BCM1140X) to B53.

Signed-off-by: Arun Parameswaran <arun.parameswaran@broadcom.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Tue, 7 Aug 2018 22:43:12 +0000 (15:43 -0700)]
Merge branch '40GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
40GbE Intel Wired LAN Driver Updates 2018-08-07

This series contains updates to i40e and i40evf only.

Sergey cleans up a duplicate call to i40e_prep_for_reset() during
shutdown.

YueHaibing cleans up i40evf by removing code that was never being used
or called within the driver.

Jake updates the ethtool statistics to use a helper function since many
of the statistics use the same basic logic for copying strings into the
supplied buffer.  Cleaned up the use of a local variable that is no
longer needed or used.  Fixed additional stats issues, including the
failure to update the data pointer which was causing stats to be
reported incorrectly.

Mariusz fixes a bug where there was an oversight in configuring FEC when
link settings were forced which was causing 25G link to be configured
incorrectly.

Piotr adds a missing return code for when the firmware returns a busy
state.  Also added the process to command firmware to start
rearrangement when switching between old NVM structure to the new flat
NVM.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'qed-Add-Multi-TC-RoCE-support'
David S. Miller [Tue, 7 Aug 2018 20:22:11 +0000 (13:22 -0700)]
Merge branch 'qed-Add-Multi-TC-RoCE-support'

Denis Bolotin says:

====================
qed: Add Multi-TC RoCE support

This patch series adds support for multiple concurrent traffic classes for RoCE.
The first three patches enable the required parts of the driver to learn the TC
configuration, and the last one makes use of it to enable the feature.
Please consider applying this to net-next.

V1->V2:
-------
Avoid allocation in qed_dcbx_get_priority_tc().
Move qed_dcbx_get_priority_tc() out of CONFIG_DCB section since it doesn't call
qed_dcbx_query_params() anymore.

v2->V3:
-------
patch 1/3:
qed_dcbx_get_priority_tc() always returns a valid TC by value. In error cases,
it returns QED_DCBX_DEFAULT_TC (currently defined 0).
patch 3/3:
Cosmetic changes in qed_dev.c.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqed: Add Multi-TC RoCE support
Denis Bolotin [Tue, 7 Aug 2018 12:48:10 +0000 (15:48 +0300)]
qed: Add Multi-TC RoCE support

RoCE qps use a pair of physical queues (pq) received from the Queue Manager
(QM) - an offload queue (OFLD) and a low latency queue (LLT). The QM block
creates a pq for each TC, and allows RoCE qps to ask for a pq with a
specific TC. As a result, qps with different VLAN priorities can be mapped
to different TCs, and employ features such as PFC and ETS.

Signed-off-by: Michal Kalderon <michal.kalderon@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: Denis Bolotin <denis.bolotin@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqed: Add a flag which indicates if offload TC is set
Denis Bolotin [Tue, 7 Aug 2018 12:48:09 +0000 (15:48 +0300)]
qed: Add a flag which indicates if offload TC is set

Distinguish not set offload_tc from offload_tc 0 and add getters and
setters.

Signed-off-by: Michal Kalderon <michal.kalderon@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: Denis Bolotin <denis.bolotin@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoqed: Add DCBX API - qed_dcbx_get_priority_tc()
Denis Bolotin [Tue, 7 Aug 2018 12:48:08 +0000 (15:48 +0300)]
qed: Add DCBX API - qed_dcbx_get_priority_tc()

The API receives a priority and looks for the TC it is mapped to in the
operational DCBX configuration. The API returns QED_DCBX_DEFAULT_TC (0)
when DCBX is disabled.

Signed-off-by: Michal Kalderon <michal.kalderon@cavium.com>
Signed-off-by: Ariel Elior <ariel.elior@cavium.com>
Signed-off-by: Denis Bolotin <denis.bolotin@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoRDS: IB: fix 'passing zero to ERR_PTR()' warning
YueHaibing [Tue, 7 Aug 2018 11:34:16 +0000 (19:34 +0800)]
RDS: IB: fix 'passing zero to ERR_PTR()' warning

Fix a static code checker warning:
 net/rds/ib_frmr.c:82 rds_ib_alloc_frmr() warn: passing zero to 'ERR_PTR'

The error path for ib_alloc_mr failure should set err to PTR_ERR.

Fixes: 1659185fb4d0 ("RDS: IB: Support Fastreg MR (FRMR) memory registration mode")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'macb-add-pad-and-fcs-support'
David S. Miller [Tue, 7 Aug 2018 20:18:50 +0000 (13:18 -0700)]
Merge branch 'macb-add-pad-and-fcs-support'

Claudiu Beznea says:

====================
net: macb: add pad and fcs support

In [1] it was reported that UDP checksum is offloaded to hardware no mather
it was previously computed in software or not. The proposal on [1] was to
disable TX checksum offload.

This series (mostly patch 3/3) address the issue described at [1] by
setting NOCRC bit to TX buffer descriptor for SKBs that arrived from
networking stack with checksum computed. For these packets padding and FCS
need to be added (hardware doesn't compute them if NOCRC bit is set). The
minimum packet size that hardware expects is 64 bytes (including FCS).
This feature could not be used in case of GSO, so, it was used only for
no GSO SKBs.

For SKBs wich requires padding and FCS computation macb_pad_and_fcs()
checks if there is enough headroom and tailroom in SKB to avoid copying
SKB structure. Since macb_pad_and_fcs() may change SKB the
macb_pad_and_fcs() was places in macb_start_xmit() b/w macb_csum_clear()
and skb_headlen() calls.

This patch was tested with pktgen in kernel tool in a script like this:
(pktgen_sample01_simple.sh is at [2]):

minSize=1
maxSize=1500

for i in `seq $minSize $maxSize` ; do
copy="$(shuf -i 1-2000 -n 1)"
./pktgen_sample01_simple.sh -i eth0 \
-m <dst-mac-addr> -d <dst-ip-addr> -x -s $i -c $copy
done

minStep=1
maxStep=200
for i in `seq $minStep $maxStep` ; do
copy="$(shuf -i 1-2000 -n 1)"
size="$(shuf -i 1-1500 -n 1)"
./pktgen_sample01_simple.sh -i eth0 \
-m <dst-mac-addr> -d <dst-ip-addr> -x -s $size -c $copy
done

Changes since RFC:
- in patch 3/3 order local variables by their lenght (reverse christmas tree
  format)

[1] https://www.spinics.net/lists/netdev/msg505065.html
[2] https://github.com/netoptimizer/network-testing/blob/master/pktgen/pktgen_sample01_simple.sh
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: macb: add support for padding and fcs computation
Claudiu Beznea [Tue, 7 Aug 2018 09:25:14 +0000 (12:25 +0300)]
net: macb: add support for padding and fcs computation

For packets with computed IP/TCP/UDP checksum there is no need to tell
hardware to recompute it. For such kind of packets hardware expects the
packet to be at least 64 bytes and FCS to be computed.

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: macb: move checksum clearing outside of spinlock
Claudiu Beznea [Tue, 7 Aug 2018 09:25:13 +0000 (12:25 +0300)]
net: macb: move checksum clearing outside of spinlock

Move checksum clearing outside of spinlock. The SKB is protected by
networking lock (HARD_TX_LOCK()).

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: macb: use netdev_tx_t return type for ndo_start_xmit functions
Claudiu Beznea [Tue, 7 Aug 2018 09:25:12 +0000 (12:25 +0300)]
net: macb: use netdev_tx_t return type for ndo_start_xmit functions

Use netdev_tx_t return type for ndo_start_xmit function of macb driver.

Signed-off-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'ibmvnic-next'
David S. Miller [Tue, 7 Aug 2018 19:46:28 +0000 (12:46 -0700)]
Merge branch 'ibmvnic-next'

Thomas Falcon says:

====================
ibmvnic: Update firmware error reporting

This patch set cleans out a lot of dead code from the ibmvnic driver
and adds some more. The error ID field of the descriptor is not filled
in by firmware, so do not print it and do not use it to query for
more detailed information. Remove the unused code written for this.
Finally, update the message to print a string explainng the error
cause instead of just the error code.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoibmvnic: Update firmware error reporting with cause string
Thomas Falcon [Tue, 7 Aug 2018 02:39:59 +0000 (21:39 -0500)]
ibmvnic: Update firmware error reporting with cause string

Print a string instead of the error code. Since there is a
possibility that the driver can recover, classify it as a
warning instead of an error.

Signed-off-by: Thomas Falcon <tlfalcon@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoibmvnic: Remove code to request error information
Thomas Falcon [Tue, 7 Aug 2018 02:39:58 +0000 (21:39 -0500)]
ibmvnic: Remove code to request error information

When backing device firmware reports an error, it provides an
error ID, which is meant to be queried for more detailed error
information. Currently, however, an error ID is not provided by
the Virtual I/O server and there are not any plans to do so. For
now, it is always unfilled or zero, so request_error_information
will never be called.  Remove it.

Signed-off-by: Thomas Falcon <tlfalcon@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoliquidio: avoided acquiring post_lock for data only queues
Intiyaz Basha [Mon, 6 Aug 2018 20:09:40 +0000 (13:09 -0700)]
liquidio: avoided acquiring post_lock for data only queues

All control commands (soft commands) goes through only Queue 0
(control and data queue). So only queue-0 needs post_lock,
other queues are only data queues and does not need post_lock

Added a flag to indicate the queue can be used for soft commands.

If this flag is set, post_lock must be acquired before posting
a command to the queue.
If this flag is clear, post_lock is invalid for the queue.

Signed-off-by: Intiyaz Basha <intiyaz.basha@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoip6_tunnel: collect_md xmit: Use ip_tunnel_key's provided src address
Shmulik Ladkani [Mon, 6 Aug 2018 12:00:59 +0000 (15:00 +0300)]
ip6_tunnel: collect_md xmit: Use ip_tunnel_key's provided src address

When using an ip6tnl device in collect_md mode, the xmit methods ignore
the ipv6.src field present in skb_tunnel_info's key, both for route
calculation purposes (flowi6 construction) and for assigning the
packet's final ipv6h->saddr.

This makes it impossible specifying a desired ipv6 local address in the
encapsulating header (for example, when using tc action tunnel_key).

This is also not aligned with behavior of ipip (ipv4) in collect_md
mode, where the key->u.ipv4.src gets used.

Fix, by assigning fl6.saddr with given key->u.ipv6.src.
In case ipv6.src is not specified, ip6_tnl_xmit uses existing saddr
selection code.

Fixes: 8d79266bc48c ("ip6_tunnel: add collect_md mode to IPv6 tunnels")
Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Reviewed-by: Eyal Birger <eyal.birger@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet: sched: cls_flower: set correct offload data in fl_reoffload
Vlad Buslov [Mon, 6 Aug 2018 08:27:10 +0000 (11:27 +0300)]
net: sched: cls_flower: set correct offload data in fl_reoffload

fl_reoffload implementation sets following members of struct
tc_cls_flower_offload incorrectly:
 - masked key instead of mask
 - key instead of masked key

Fix fl_reoffload to provide correct data to offload callback.

Fixes: 31533cba4327 ("net: sched: cls_flower: implement offload tcf_proto_op")
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoMerge branch 'nfp-ttl-tos-geneve'
David S. Miller [Tue, 7 Aug 2018 19:22:15 +0000 (12:22 -0700)]
Merge branch 'nfp-ttl-tos-geneve'

Simon Horman says:

====================
nfp: flower: tunnel TTL & TOS, and Geneve options set & match support

this series contains updates for the TC Flower classifier
and the offload facility for it in the NFP driver.

* Patches 1 & 2: update the NFP driver to allow offload
  of matching and setting tunnel ToS/TTL of flows using the TC Flower
  classifier and tun_key action

* Patches 3 & 4: enhance the flow dissector and TC Flower classifier
  to allow match on Geneve options

* Patch 5 & 6: update the NFP driver to allow offload of
  matching and setting Geneve options of flows using the TC Flower
  classifier and tun_key action
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonfp: flower: add geneve option match offload
Pieter Jansen van Vuuren [Tue, 7 Aug 2018 15:36:03 +0000 (17:36 +0200)]
nfp: flower: add geneve option match offload

Introduce a new layer for matching on geneve options. This allows
offloading filters configured to match geneve with options.

Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonfp: flower: add geneve option push action offload
Pieter Jansen van Vuuren [Tue, 7 Aug 2018 15:36:02 +0000 (17:36 +0200)]
nfp: flower: add geneve option push action offload

Introduce new push geneve option action. This allows offloading
filters configured to entunnel geneve with options.

Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonet/sched: allow flower to match tunnel options
Pieter Jansen van Vuuren [Tue, 7 Aug 2018 15:36:01 +0000 (17:36 +0200)]
net/sched: allow flower to match tunnel options

Allow matching on options in Geneve tunnel headers.
This makes use of existing tunnel metadata support.

The options can be described in the form
CLASS:TYPE:DATA/CLASS_MASK:TYPE_MASK:DATA_MASK, where CLASS is
represented as a 16bit hexadecimal value, TYPE as an 8bit
hexadecimal value and DATA as a variable length hexadecimal value.

e.g.
 # ip link add name geneve0 type geneve dstport 0 external
 # tc qdisc add dev geneve0 ingress
 # tc filter add dev geneve0 protocol ip parent ffff: \
     flower \
       enc_src_ip 10.0.99.192 \
       enc_dst_ip 10.0.99.193 \
       enc_key_id 11 \
       geneve_opts 0102:80:1122334421314151/ffff:ff:ffffffffffffffff \
       ip_proto udp \
       action mirred egress redirect dev eth1

This patch adds support for matching Geneve options in the order
supplied by the user. This leads to an efficient implementation in
the software datapath (and in our opinion hardware datapaths that
offload this feature). It is also compatible with Geneve options
matching provided by the Open vSwitch kernel datapath which is
relevant here as the Flower classifier may be used as a mechanism
to program flows into hardware as a form of Open vSwitch datapath
offload (sometimes referred to as OVS-TC). The netlink
Kernel/Userspace API may be extended, for example by adding a flag,
if other matching options are desired, for example matching given
options in any order. This would require an implementation in the
TC software datapath. And be done in a way that drivers that
facilitate offload of the Flower classifier can reject or accept
such flows based on hardware datapath capabilities.

This approach was discussed and agreed on at Netconf 2017 in Seoul.

Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoflow_dissector: allow dissection of tunnel options from metadata
Simon Horman [Tue, 7 Aug 2018 15:36:00 +0000 (17:36 +0200)]
flow_dissector: allow dissection of tunnel options from metadata

Allow the existing 'dissection' of tunnel metadata to 'dissect'
options already present in tunnel metadata. This dissection is
controlled by a new dissector key, FLOW_DISSECTOR_KEY_ENC_OPTS.

This dissection only occurs when skb_flow_dissect_tunnel_info()
is called, currently only the Flower classifier makes that call.
So there should be no impact on other users of the flow dissector.

This is in preparation for allowing the flower classifier to
match on Geneve options.

Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonfp: flower: allow matching on ipv4 UDP tunnel tos and ttl
John Hurley [Tue, 7 Aug 2018 15:35:59 +0000 (17:35 +0200)]
nfp: flower: allow matching on ipv4 UDP tunnel tos and ttl

The addition of FLOW_DISSECTOR_KEY_ENC_IP to TC flower means that the ToS
and TTL of the tunnel header can now be matched on.

Extend the NFP tunnel match function to include these new fields.

Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agonfp: flower: set ip tunnel ttl from encap action
John Hurley [Tue, 7 Aug 2018 15:35:58 +0000 (17:35 +0200)]
nfp: flower: set ip tunnel ttl from encap action

The TTL for encapsulating headers in IPv4 UDP tunnels is taken from a
route lookup. Modify this to first check if a user has specified a TTL to
be used in the TC action.

Signed-off-by: John Hurley <john.hurley@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
6 years agoi40e: fix i40e_add_queue_stats data pointer update
Jacob Keller [Tue, 31 Jul 2018 10:41:48 +0000 (03:41 -0700)]
i40e: fix i40e_add_queue_stats data pointer update

This function accidentally failed to update the data pointer, which
caused the reported stats to be incorrect. Additionally, statistics
which follow queue stats in the output would potentially read non-zeroed
garbage data from the ethtool buffer.

This occurred because the data double pointer was not dereferenced
before incrementing the size.

Additionally, make sure this issue is more visible by adding a WARN_ONCE
to the i40e_get_ethtool_stats function. This warning will trigger
whenever the data pointer is not at the expected address, similar to the
check that we make in the i40e_get_stat_strings() function.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>