linux-dev/drivers/net/ethernet/sfc, branch master

sfc: Fix an error handling path in efx_pci_probe()

2022-11-01T16:09:47Z

If an error occurs after the first kzalloc() the corresponding memory allocation is never freed. Add the missing kfree() in the error handling path, as already done in the remove() function. Fixes: 7e773594dada ("sfc: Separate efx_nic memory from net_device memory") Signed-off-by: Christophe JAILLET Acked-by: Martin Habets Link: https://lore.kernel.org/r/dc114193121c52c8fa3779e49bdd99d4b41344a9.1667077009.git.christophe.jaillet@wanadoo.fr Signed-off-by: Jakub Kicinski

sfc: include vport_id in filter spec hash and equal()

2022-10-20T00:08:43Z

Filters on different vports are qualified by different implicit MACs and/or VLANs, so shouldn't be considered equal even if their other match fields are identical. Fixes: 7c460d9be610 ("sfc: Extend and abstract efx_filter_spec to cover Huntington/EF10") Co-developed-by: Edward Cree Signed-off-by: Edward Cree Signed-off-by: Pieter Jansen van Vuuren Reviewed-by: Martin Habets Link: https://lore.kernel.org/r/20221018092841.32206-1-pieter.jansen-van-vuuren@amd.com Signed-off-by: Jakub Kicinski

sfc: Change VF mac via PF as first preference if available.

2022-10-15T10:06:44Z

Changing a VF's mac address through the VF (rather than via the PF) fails with EPERM because the latter part of efx_ef10_set_mac_address attempts to change the vport mac address list as the VF. Even with this fixed it still fails with EBUSY because the vadaptor is still assigned on the VF - the vadaptor reassignment must be within a section where the VF has torn down its state. A major reason this has broken is because we have two functions that ostensibly do the same thing - have a PF and VF cooperate to change a VF mac address. Rather than do this, if we are changing the mac of a VF that has a link to the PF in the same VM then simply call sriov_set_vf_mac instead, which is a proven working function that does that. If there is no PF available, or that fails non-fatally, then attempt to change the VF's mac address as we would a PF, without updating the PF's data. Test case: Create a VF: echo 1 > /sys/class/net//device/sriov_numvfs Set the mac address of the VF directly: ip link set addr 00:11:22:33:44:55 Set the MAC address of the VF via the PF: ip link set vf 0 mac 00:11:22:33:44:66 Without this patch the last command will fail with ENOENT. Signed-off-by: Jonathan Cooper Reported-by: Íñigo Huguet Fixes: 910c8789a777 ("set the MAC address using MC_CMD_VADAPTOR_SET_MAC") Acked-by: Edward Cree Signed-off-by: David S. Miller

Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

2022-09-29T21:30:51Z

No conflicts. Signed-off-by: Jakub Kicinski

net: drop the weight argument from netif_napi_add

2022-09-29T01:57:14Z

We tell driver developers to always pass NAPI_POLL_WEIGHT as the weight to netif_napi_add(). This may be confusing to newcomers, drop the weight argument, those who really need to tweak the weight can use netif_napi_add_weight(). Acked-by: Marc Kleine-Budde # for CAN Link: https://lore.kernel.org/r/20220927132753.750069-1-kuba@kernel.org Signed-off-by: Jakub Kicinski

sfc: bare bones TC offload on EF100

2022-09-28T08:43:22Z

This is the absolute minimum viable TC implementation to get traffic to VFs and allow them to be tested; it supports no match fields besides ingress port, no actions besides mirred and drop, and no stats. Example usage: tc filter add dev $PF parent ffff: flower skip_sw \ action mirred egress mirror dev $VFREP tc filter add dev $VFREP parent ffff: flower skip_sw \ action mirred egress redirect dev $PF gives a VF unfiltered access to the network out the physical port ($PF acts here as a physical port representor). More matches, actions, and counters will be added in subsequent patches. Signed-off-by: Edward Cree Signed-off-by: David S. Miller

sfc: interrogate MAE capabilities at probe time

2022-09-28T08:43:22Z

Different versions of EF100 firmware and FPGA bitstreams support different matching capabilities in the Match-Action Engine. Probe for these at start of day; subsequent patches will validate TC offload requests against the reported capabilities. Signed-off-by: Edward Cree Signed-off-by: David S. Miller

sfc: add a hashtable for offloaded TC rules

2022-09-28T08:43:22Z

Nothing inserts into this table yet, but we have code to remove rules on FLOW_CLS_DESTROY or at driver teardown time, in both cases also attempting to remove the corresponding hardware rules. Signed-off-by: Edward Cree Signed-off-by: David S. Miller

sfc: optional logging of TC offload errors

2022-09-28T08:43:22Z

TC offload support will involve complex limitations on what matches and actions a rule can do, in some cases potentially depending on rules already offloaded. So add an ethtool private flag "log-tc-errors" which controls reporting the reasons for un-offloadable TC rules at NETIF_INFO. Signed-off-by: Edward Cree Signed-off-by: David S. Miller

sfc: bind indirect blocks for TC offload on EF100

2022-09-28T08:43:22Z

Bind indirect blocks for recognised tunnel netdevices. Currently these connect to a stub efx_tc_flower() that only returns -EOPNOTSUPP; subsequent patches will implement flower offloads to the Match-Action Engine. Signed-off-by: Edward Cree Signed-off-by: David S. Miller