- 14 Dec, 2018 33 commits
-
-
Aviv Heller authored
With the introduction of SR-IOV LAG, checking whether LAG is active is no longer good enough, since RoCE and SR-IOV LAG each entails different behavior by both the core and infiniband drivers. This patch introduces facilities to discern LAG type, in addition to mlx5_lag_is_active(). These are implemented in such a way as to allow more complex mode combinations in the future. Signed-off-by: Aviv Heller <avivh@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Aviv Heller authored
The new name better represents the function's aim, and sets a precedent for a '__' prefix for internal, non-locked versions of LAG APIs. Signed-off-by: Aviv Heller <avivh@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Rabie Loulou authored
Enable setting uplink LAG if sriov is enabled on both ports in switchdev mode. Once the sriov mode is changed from switchdev for any of the ports, the LAG instance is disabled. Signed-off-by: Rabie Loulou <rabiel@mellanox.com> Signed-off-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Jianbo Liu <jianbol@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Rabie Loulou authored
Remove the lag forbid/allow functions, change the lag prereq check to run in the do-bond logic, so every change in the prereq state will cause LAG to be disabled/enabled accordingly after the next do-bond run. Add lag update function, so every component which changes the prereq state and want the LAG to re-calc the conditions can call the update function. Signed-off-by: Rabie Loulou <rabiel@mellanox.com> Signed-off-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Rabie Loulou authored
When HW lag is set/unset, roce must not be enabled on the port, as such we wrap such changes with roce enable/disable either directly or through re-creation of IB device. Currently, lag and sriov are mutually exclusive, so by definition this code doesn't run under sriov. Towards changing this exclusion, we need to make sure that roce will not be enabled on the eswitch manager port under sriov since this is requirement of the switchdev mode. We are going strict here and avoiding this all together under sriov. Signed-off-by: Rabie Loulou <rabiel@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Aviv Heller authored
Under uplink LAG, packets that match a flow might arrive on both uplink ports and transmitted through both as part of supporting aggregation and high-availability. When the netdevs representing the uplinks are set into LAG (bonding, teaming), duplicate the TC flow offloading into each of the per-uplink e-switches. Duplication is not required if the source is the bond device, since in this case it is assumed that the bond and the uplink netdevs share the same TC block, and thus duplication will occur naturally by the stack. Note that under encapsulation scheme, both flows will use the same neighbour and hence both will contribute to the last-used feedback towards the stack. Signed-off-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Rabie Loulou <rabiel@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Rabie Loulou authored
When parsing TC FDB actions, if the egress device is a bond/team net-device which enslaved the uplink representor of the e-switch, use the uplink representor as the destination in the HW rule. Signed-off-by: Rabie Loulou <rabiel@mellanox.com> Signed-off-by: Aviv Heller <avivh@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Rabie Loulou authored
When the uplink representors are put into lag, set all the representors (VFs and uplinks) of the same NIC to return the same switchdev id. Currently, the route lookup code on the encapsulation offload path assumes that same switchdev id for the source and dest devices means that the dest is also mlx5 HW netdev. This doesn't hold anymore when we align the switchdev Id of the uplinks to be same, which in turn causes the bond/team to return that id to the caller. As such, enhance the relevant check to take into account the uplink lag case. Signed-off-by: Rabie Loulou <rabiel@mellanox.com> Signed-off-by: Aviv Heller <avivh@mellanox.com> Reviewed-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Shahar Klein authored
Assign a counter dev attribute according to device capability and use it for management of counters related to offloaded eswitch TC flows. With upcoming support for uplink LAG, we have two HW rules per one logical SW (TC) rule. Although the HW supports attaching one counter to multiple rules, we are allocating counter per HW rule. We need this separation for two reasons: 1. "flow eswitch" counter affinity HW require the counter to be allocated on the device where the eswitch rule is set. 2. for some use-cases (multi-path routing) each HW flow relates to different neighbour, hence our neigh update logic must have a per-rule HW accountant in order to provide the proper feedback to the kernel. Signed-off-by: Shahar Klein <shahark@mellanox.com> Signed-off-by: Roi Dayan <roid@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Roi Dayan authored
Under uplink LAG or multipath schemes, traffic that matches one flow might arrive on both uplink ports and transmitted through both as part of supporting aggregation and high-availability. To cope with the fact that the SW model might use logical SW port (e.g uplink team or bond) but we have two HW ports with e-switch on each, there are cases where in order to offload a SW TC rule we need to duplicate it to two HW flows. Since each HW rule has its own counter we also aggregate the counter of both rules when a flow stats query is executed from user-space. Introduce the changes for the different elements (add/delete/stats), currently nothing is duplicated. Signed-off-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Shahar Klein <shahark@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Roi Dayan authored
In the sriov offloads mode, packets that are not matched by any other rule are sent towards the e-switch vport manager for further processing. Under upcoming patches (e.g for uplink LAG), packets sent from VF vports belonging to esw0 (e-switch related to PF0) might end up in esw1 (e-switch related to PF1) due to muxing logic applied by the FW. In such a case we still want the missed packet to be sent to the "base" esw manager vport in order to present the control plane a consistent view of the source (VF reresentor) port. Signed-off-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Shahar Klein <shahark@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Aviv Heller authored
This introduces devcom, a generic mechanism for performing operations on both physical functions of the same Connect-X card. The first user of this API is merged eswitch, which will be introduced in subsequent patches. Signed-off-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linuxSaeed Mahameed authored
mlx5-next shared branch with rdma subtree to avoid mlx5 rdma v.s. netdev conflicts. Highlights: 1) Lag refactroing and flow counter affinity bits. 2) mlx5 core cleanups By Roi Dayan (2) and others * 'mlx5-next' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux: net/mlx5: Fold the modify lag code into function net/mlx5: Add lag affinity info to log net/mlx5: Split the activate lag function into two routines net/mlx5: E-Switch, Introduce flow counter affinity IB/mlx5: Unify e-switch representors load approach between uplink and VFs net/mlx5: Use lowercase 'X' for hex values net/mlx5: Remove duplicated include from eswitch.c Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Shahar Klein authored
Handle the code of modifying the lag affinity within a separate function. Signed-off-by: Shahar Klein <shahark@mellanox.com> Signed-off-by: Roi Dayan <roid@mellanox.com> Reviewed-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Roi Dayan authored
Report the initial LAG port affinity upon LAG creation. Signed-off-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Shahar Klein <shahark@mellanox.com> Reviewed-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Roi Dayan authored
Split the activate lag function in order to be symmetric with the deactivate lag function. Signed-off-by: Roi Dayan <roid@mellanox.com> Signed-off-by: Shahar Klein <shahark@mellanox.com> Reviewed-by: Aviv Heller <avivh@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Shahar Klein authored
This dictates the device affinity for eswitch flow counters, set by the FW according to the HW device capabilities. Under "source eswitch" affinity, the counter should be allocated on the device related to the source vport in the match. This covers both non merged e-switch mode as well as old FW that does not advertise this cap. Under "flow eswitch" affinity, the counter should be allocated on the device where the eswitch rule is set. Signed-off-by: Shahar Klein <shahark@mellanox.com> Signed-off-by: Roi Dayan <roid@mellanox.com> Reviewed-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Mark Bloch authored
When in switchdev mode and the add function is called by the core level driver, make sure we only register the callbacks, but don't create the mlx5 IB device or initialize anything. With this change all the IB devices in switchdev mode are created only once the load callback is invoked by the e-switch core sub-module. This follows the design paradigm under which the all the Eth representors must be loaded before any of IB reprs is loaded. Signed-off-by: Mark Bloch <markb@mellanox.com> Acked-by: Or Gerlitz <ogerlitz@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
Saeed Mahameed authored
Apparently gcc is cool with upper case '0X' but it is not commonly used. Replace '0X' with lowercase '0x' in mlx5_ifc.h file. Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
David S. Miller authored
Petr Machata says: ==================== Introduce NETDEV_PRE_CHANGEADDR Spectrum devices have a limitation that all router interfaces need to have the same address prefix. In Spectrum-1, the requirement is for the initial 38 bits of all RIFs to be the same, in Spectrum-2 the limit is 36 bits. Currently violations of this requirement are not diagnosed. At the same time, if the condition is not upheld, the mismatched MAC address ends up overwriting the common prefix, and all RIF MAC addresses silently change to the new prefix. It is therefore desirable to be able at least to diagnose the issue, and better to reject attempts to change MAC addresses in ways that is incompatible with the device. Currently MAC address changes are notified through emission of NETDEV_CHANGEADDR, which is done after the change. Extending this message to allow vetoing is certainly possible, but several other notification types have instead adopted a simple two-stage approach: first a "pre" notification is sent to make sure all interested parties are OK with the change that's about to be done. Then the change is done, and afterwards a "post" notification is sent. This dual approach is easier to use: when the change is vetoed, nothing has changed yet, and it's therefore unnecessary to roll anything back. Therefore this patchset introduces it for NETDEV_CHANGEADDR as well. One prominent path to emitting NETDEV_CHANGEADDR is through dev_set_mac_address(). Therefore in patch #1, give this function an extack argument, so that a textual reason for rejection (or a warning) can be communicated back to the user. In patch #2, add the new notification type. In patch #3, have dev.c emit the notification for instances of dev_addr change, or addition of an address to dev_addrs list. In patches #4 and #5, extend the bridge driver to handle and emit the new notifier. In patch #6, change IPVLAN to emit the new notifier. Likewise for bonding driver in patches #7 and #8. Note that the team driver doesn't need this treatment, as it goes through dev_set_mac_address(). In patches #9, #10 and #11 adapt mlxsw to veto MAC addresses on router interfaces, if they violate the requirement that all RIF MAC addresses have the same prefix. Finally in patches #12 and #13, add a test for vetoing of a direct change of a port device MAC, and indirect change of a bridge MAC. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
When a FID RIF is created for a bridge with IP address, its MAC address must obey the same requirements as other RIFs. Test that attempts to change the address incompatibly by attaching a device are vetoed with extack. 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>
-
Petr Machata authored
Test that attempts to change address in a way that violates Spectrum requirements are vetoed with extack. 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>
-
Petr Machata authored
On NETDEV_PRE_CHANGEADDR, if the change is related to a RIF interface, verify that it satisfies the criterion that all RIF interfaces have the same MAC address prefix, as indicated by mlxsw_sp.mac_mask. Additionally, besides explicit address changes, check that the address of an interface for which a RIF is about to be added matches the required pattern as well. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
The Spectrum hardware demands that all router interfaces in the system have the same first 38 resp. 36 bits of MAC address: the former limit holds on Spectrum, the latter on Spectrum-2. Add a field that refers to the required prefix mask and initialize in mlxsw_sp1_init() and mlxsw_sp2_init(). Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
Prepare mlxsw_sp_netdevice_router_port_event() for handling of NETDEV_PRE_CHANGEADDR. Split out the part that deals with the actual changes and call it for the two events currently handled. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
Give interested parties an opportunity to veto an impending HW address change. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
Before NETDEV_CHANGEADDR, bond driver should emit NETDEV_PRE_CHANGEADDR, and allow consumers to veto the address change. To propagate further the return code from NETDEV_PRE_CHANGEADDR, give the function that implements address change a return value. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
A NETDEV_CHANGEADDR event implies a change of address of each of the IPVLANs of this IPVLAN device. Therefore propagate NETDEV_PRE_CHANGEADDR to all the IPVLANs. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
When a port device seeks approval of a potential new MAC address, make sure that should the bridge device end up using this address, all interested parties would agree with it. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
When a port is attached to a bridge, the address of the bridge in question may change as well. Even if it would not change at this point (because the current bridge address is lower), it might end up changing later as a result of detach of another port, which can't be vetoed. Therefore issue NETDEV_PRE_CHANGEADDR regardless of whether the address will be used at this point or not, and make sure all involved parties would agree with the change. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
When a device address is about to be changed, or an address added to the list of device HW addresses, it is necessary to ensure that all interested parties can support the address. Therefore, send the NETDEV_PRE_CHANGEADDR notification, and if anyone bails on it, do not change the address. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
The NETDEV_CHANGEADDR notification is emitted after a device address changes. Extending this message to allow vetoing is certainly possible, but several other notification types have instead adopted a simple two-stage approach: first a "pre" notification is sent to make sure all interested parties are OK with a change that's about to be done. Then the change is done, and afterwards a "post" notification is sent. This dual approach is easier to use: when the change is vetoed, nothing has changed yet, and it's therefore unnecessary to roll anything back. Therefore adopt it for NETDEV_CHANGEADDR as well. To that end, add NETDEV_PRE_CHANGEADDR and an info structure to go along with it. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
A follow-up patch will add a notifier type NETDEV_PRE_CHANGEADDR, which allows vetoing of MAC address changes. One prominent path to that notification is through dev_set_mac_address(). Therefore give this function an extack argument, so that it can be packed together with the notification. Thus a textual reason for rejection (or a warning) can be communicated back to the user. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
- 13 Dec, 2018 7 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linuxDavid S. Miller authored
Saeed Mahameed says: ==================== mlx5e-updates-2018-12-11 From Eli Britstein, Patches 1-10 adds remote mirroring support. Patches 1-4 refactor encap related code as pre-steps for using per destination encapsulation properties. Patches 5-7 use extended destination feature for single/multi destination scenarios that have a single encap destination. Patches 8-10 enable multiple encap destinations for a TC flow. From, Daniel Jurgens, Patch 11, Use CQE padding for Ethernet CQs, PPC showed up to a 24% improvement in small packet throughput From Eyal Davidovich, patches 12-14, FW monitor counter support FW monitor counters feature came to solve the delayed reporting of FW stats in the atomic get_stats64 ndo, since we can't access the FW at that stage, this feature will enable immediate FW stats updates in the driver via fw events on specific stats updates. Patch 12, cleanup to avoid querying a FW counter when it is not supported Patch 13, Monitor counters FW commands support Patch 14, Use monitor counters in ethernet netdevice to update FW stats reported in the atomic get_stats64 ndo. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
-
Biao Huang authored
Add MODULE_LICENSE info to fix this: WARNING: modpost: missing MODULE_LICENSE() in drivers/net/ethernet/stmicro/stmmac/dwmac-mediatek.o Signed-off-by: Biao Huang <biao.huang@mediatek.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
YueHaibing authored
Fixes gcc '-Wunused-but-set-variable' warning: drivers/net/ethernet/mellanox/mlx5/core/en_rep.c: In function 'mlx5e_vport_rep_load': drivers/net/ethernet/mellanox/mlx5/core/en_rep.c:1490:21: warning: variable 'upriv' set but not used [-Wunused-but-set-variable] drivers/net/ethernet/mellanox/mlx5/core/en_rep.c: In function 'mlx5e_vport_rep_unload': drivers/net/ethernet/mellanox/mlx5/core/en_rep.c:1557:21: warning: variable 'upriv' set but not used [-Wunused-but-set-variable] It not used any more since commit ef381359 ("net/mlx5e: Replace egdev with indirect block notifications"). Also remove unused variable 'uplink_rpriv' after this change. Signed-off-by: YueHaibing <yuehaibing@huawei.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Saeed Mahameed <saeedm@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
YueHaibing authored
Remove duplicated include. Signed-off-by: YueHaibing <yuehaibing@huawei.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
-
David S. Miller authored
Petr Machata says: ==================== Pass extack to SWITCHDEV_PORT_OBJ_ADD Drivers may need to do validation as a result of port object addition. An example is mlxsw, which needs to check the configuration of a VXLAN device attached to an offloaded bridge. Without a mapped VLAN, the invalidity of the device is not important, but as soon as a pvid, untagged VLAN is configured for the device, it has to be validated and offloaded. Should the validation fail, there's currently no way to communicate details of the failure to the user, beyond an error number. Because currently, extack is not available at all in that area of code, this patch starts down at the RTNL level and progresses up towards the driver(s). In patch #1, ndo_bridge_setlink is updated to include extack, and callbacks of all clients are updated as well (ignoring the argument). In patch #2, the bridge driver is updated to propagate the extack through to the switchdev border, br_switchdev_port_vlan_add(). Patches #3, #4 and #5 then gradually extend switchdev to pass the extack argument through to the switchdev blocking notifier chain. Patches #6 and #7 then update mlxsw to pass the extack argument from VXLAN events resp. port events on to mlxsw_sp_bridge_8021q_vxlan_join(). Finally in patches #8 and #9, the code paths from the previous two patches are verified to yield an error message. v2: - Patch #1: - In ndo_bridge_setlink(), keep the whole extack declaration on the same line. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
Test mapping a VLAN at a port device such that on the same VLAN, there already is an unoffloadable VXLAN device. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-
Petr Machata authored
Test mapping a VLAN at a VXLAN device that can't be offloaded. Signed-off-by: Petr Machata <petrm@mellanox.com> Acked-by: Jiri Pirko <jiri@mellanox.com> Reviewed-by: Ido Schimmel <idosch@mellanox.com> Signed-off-by: David S. Miller <davem@davemloft.net>
-