aboutsummaryrefslogtreecommitdiff
path: root/drivers/net/ethernet/intel
AgeCommit message (Collapse)AuthorFilesLines
2023-11-14i40e: Use helpers to check running FW and AQ API versionsIvan Vecera4-58/+36
Use new helpers to check versions of running FW and provided AQ API to make the code more readable. Signed-off-by: Ivan Vecera <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Add other helpers to check version of running firmware and AQ APIIvan Vecera1-0/+54
Add another helper functions that will be used by subsequent patch to refactor existing open-coded checks whether the version of running firmware and AdminQ API is recent enough to provide certain capabilities. Signed-off-by: Ivan Vecera <[email protected]> Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Move i40e_is_aq_api_ver_ge helperIvan Vecera2-19/+18
Move i40e_is_aq_api_ver_ge helper function (used to check if AdminQ API version is recent enough) to header so it can be used from other .c files. Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Initialize hardware capabilities at single placeIvan Vecera4-72/+51
Some i40e_hw.caps bits are set in i40e_set_hw_caps(), some of them in i40e_init_adminq() and the rest of them in i40e_sw_init(). Consolidate the initialization to single proper place i40e_set_hw_caps(). Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Consolidate hardware capabilitiesIvan Vecera7-90/+85
Fields .caps in i40e_hw and .hw_features in i40e_pf both indicate capabilities provided by hardware. Move and merge i40e_pf.hw_features into i40e_hw.caps as this is more appropriate place for them and adjust their names to I40E_HW_CAP_... convention. Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Use DECLARE_BITMAP for flags field in i40e_hwIvan Vecera7-51/+55
Convert flags field in i40e_hw from u64 to bitmap and its usage to use bit access functions and rename the field to 'caps' as this field describes capabilities that are set once on init and read many times later. Changes: - Convert "hw_ptr->flags & FLAG" to "test_bit(FLAG, ...)" - Convert "hw_ptr->flags |= FLAG" to "set_bit(FLAG, ...)" - Convert "hw_ptr->flags &= ~FLAG" to "clear_bit(FLAG, ...)" - Rename i40e_hw.flags to i40e_hw.caps - Rename i40e_set_hw_flags() to i40e_set_hw_caps() - Adjust caps names so they are prefixed by I40E_HW_CAP_ and existing _CAPABLE suffixes are stripped Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Use DECLARE_BITMAP for flags and hw_features fields in i40e_pfIvan Vecera9-515/+544
Convert flags and hw_features fields from i40e_pf from u32 to bitmaps and their usage to use bit access functions. Changes: - Convert "pf_ptr->(flags|hw_features) & FL" to "test_bit(FL, ...)" - Convert "pf_ptr->(flags|hw_features) |= FL" to "set_bit(FL, ...)" - Convert "pf_ptr->(flags|hw_features) &= ~FL" to "clear_bit(FL, ...)" - Rename flag field to bitno in i40e_priv_flags and adjust ethtool callbacks to work with flags bitmap - Rename flag names where '_ENABLED'->'_ENA' and '_DISABLED'->'_DIS' like in ice driver Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Remove _t suffix from enum type namesIvan Vecera3-7/+7
Enum type names should not be suffixed by '_t'. Either to use 'typedef enum name name_t' to so plain 'name_t var' instead of 'enum name_t var'. Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Remove unused flagsIvan Vecera4-37/+31
The flag I40E_FLAG_RX_CSUM_ENABLED and I40E_HW_FLAG_DROP_MODE are set and never read. Remove them. Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: Change user notification of non-SFP module in i40e_get_module_info()Andrii Staikov1-2/+2
Make the driver not produce an error message on "ethtool -m ethX" command when a non-SFP module is encountered hence there is no possibility to read the EEPROM. Make the message to appear in the debug log rather than as an error string. Change the return code to -EOPNOTSUPP and the string to make it more verbose. Signed-off-by: Andrii Staikov <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> CC: Stefan Assmann <[email protected]> CC: Michal Schmidt <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: add an error code check in i40e_vsi_setupSu Hui1-0/+4
check the value of 'ret' after calling 'i40e_vsi_config_rss'. Signed-off-by: Su Hui <[email protected]> Reviewed-by: Dan Carpenter <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-14i40e: increase max descriptors for XL710Justin Bronder2-6/+20
In Tables 8-12 and 8-22 in the X710/XXV710/XL710 datasheet, the QLEN description states that the maximum size of the descriptor queue is 8k minus 32, or 8160. Signed-off-by: Justin Bronder <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-13igc: Add support for PTP .getcyclesx64()Vinicius Costa Gomes6-36/+101
Add support for using Timer 1 (i225/i226 have 4 timer registers) as a free-running clock (the "cycles" clock) in addition to Timer 0 (the default, "adjustable clock"). The objective is to allow taprio/etf offloading to coexist with PTP vclocks. Besides the implementation of .getcyclesx64() for i225/i226, to keep timestamping working when vclocks are in use, we also need to add support for TX and RX timestamping using the free running timer, when the requesting socket is bound to a vclock. On the RX side, i225/i226 can be configured to store the values of two timers in the received packet metadata area, so it's a matter of configuring the right registers and retrieving the right timestamp. The TX is a bit more involved because the hardware stores a single timestamp (with the selected timer in the TX descriptor) into one of the timestamp registers. Note some changes at how the timestamps are done for RX, the conversion and adjustment of timestamps are now done closer to the consumption of the timestamp instead of near the reception. Signed-off-by: Vinicius Costa Gomes <[email protected]> Tested-by: Naama Meir <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13igc: Simplify setting flags in the TX data descriptorVinicius Costa Gomes1-6/+4
We can re-use the IGC_SET_FLAG() macro to simplify setting some values in the TX data descriptor. With the macro it's easier to get the meaning of the operations. Signed-off-by: Vinicius Costa Gomes <[email protected]> Tested-by: Naama Meir <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: reserve number of CP queuesMichal Swiatkowski4-7/+58
Rebuilding CP VSI each time the PR is created drastically increase the time of maximum VFs creation. Add function to reserve number of CP queues to deal with this problem. Use the same function to decrease number of queues in case of removing VFs. Assume that caller of ice_eswitch_reserve_cp_queues() will also call ice_eswitch_attach/detach() correct number of times. Still one by one PR adding is handy for VF resetting routine. Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: adjust switchdev rebuild pathMichal Swiatkowski3-49/+28
There is no need to use specific functions for rebuilding path. Let's use current implementation by removing all representors and as the result remove switchdev environment. It will be added in devices rebuild path. For example during adding VFs, port representors for them also will be created. Rebuild control plane VSI before removing representors with INIT_VSI flag set to reinit VSI in hardware after reset. Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: add VF representors one by oneMichal Swiatkowski5-242/+228
Implement adding representors one by one. Always set switchdev environment when first representor is being added and clear environment when last one is being removed. Basic switchdev configuration remains the same. Code related to creating and configuring representor was changed. Instead of setting whole representors in one function handle only one representor in setup function. The same with removing representors. Stop representors when new one is being added or removed. Stop means, disabling napi, stopping traffic and removing slow path rule. It is needed because ::q_id will change after remapping, so each representor will need new rule. When representor are stopped rebuild control plane VSI with one more or one less queue. One more if new representor is being added, one less if representor is being removed. Bridge port is removed during unregister_netdev() call on PR, so there is no need to call it from driver side. After that do remap new queues to correct vector. At the end start all representors (napi enable, start queues, add slow path rule). Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: realloc VSI stats arraysMichal Swiatkowski1-19/+39
Previously only case when queues amount is lower was covered. Implement realloc for case when queues amount is higher than previous one. Use krealloc() function and zero new allocated elements. It has to be done before ice_vsi_def_cfg(), because stats element for ring is set there. Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: set Tx topology every time new repr is addedMichal Swiatkowski4-7/+59
It is needed to track correct Tx topology. Update it every time new representor is created or remove node in case of removing corresponding representor. Still clear all node when removing switchdev mode as part of Tx topology isn't related only to representors. Also clear ::rate_note value to prevent skipping this node next time Tx topology is created. Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: allow changing SWITCHDEV_CTRL VSI queuesMichal Swiatkowski1-3/+10
Implement mechanism to change number of queues for SWITCHDEV_CTRL VSI type. Value from ::req_txq/rxq will be written to ::alloc_txq/rxq after calling ice_vsi_rebuild(). Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: return pointer to representorMichal Swiatkowski1-7/+10
In follow up patches it will be easier to obtain created port representor pointer instead of the id. Without it the pattern from eswitch side will look like: - create PR - get PR based on the id Reviewed-by: Przemek Kitszel <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: make representor code genericMichal Swiatkowski8-92/+131
Representor code needs to be independent from specific device type, like in this case VF. Make generic add / remove representor function and specific add VF / rem VF function. New device types will follow this scheme. In bridge offload code there is a need to get representor pointer based on VSI. Implement helper function to achieve that. Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: remove VF pointer reference in eswitch codeMichal Swiatkowski4-22/+25
Make eswitch code generic by removing VF pointer reference in functions. It is needed to support eswitch mode for other type of devices. Previously queue id used for Rx was based on VF number. Use ::q_id saved in port representor instead. After adding or removing port representor ::q_id value can change. It isn't good idea to iterate over representors list using this value. Use xa_find starting from the first one instead to get next port representor to remap. The number of port representors has to be equal to ::num_rx/tx_q. Warn if it isn't true. Reviewed-by: Przemek Kitszel <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: track port representors in xarrayMichal Swiatkowski5-101/+94
Instead of assuming that each VF has pointer to port representor save it in xarray. It will allow adding port representor for other device types. Drop reference to VF where it is use only to get port representor. Get it from xarray instead. The functions will no longer by specific for VF, rename them. Track id assigned by xarray in port representor structure. The id can't be used as ::q_id, because it is fixed during port representor lifetime. ::q_id can change after adding / removing other port representors. Side effect of removing VF pointer is that we are losing VF MAC information used in unrolling. Store it in port representor as parent MAC. Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: use repr instead of vf->reprMichal Swiatkowski1-7/+7
Extract repr from vf->repr as it is often use in the ice_repr_rem(). Remove meaningless clearing of q_vector and netdev pointers as kfree is called on repr pointer. Reviewed-by: Przemek Kitszel <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: track q_id in representorMichal Swiatkowski3-1/+3
Previously queue index of control plane VSI used by port representor was always id of VF. If we want to allow adding port representors for different devices we have to track queue index in the port representor structure. Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: remove unused control VSI parameterMichal Swiatkowski1-4/+3
It isn't used in ice_eswitch_release_reprs(). Probably leftover. Remove it. Commit that has removed usage of ctrl_vsi: commit c1e5da5dd465 ("ice: improve switchdev's slow-path") Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: remove redundant max_vsi_num variableMichal Swiatkowski1-4/+0
It is a leftover from previous implementation. Accidentally it wasn't removed. Do it now. Commit that has removed it: commit c1e5da5dd465 ("ice: improve switchdev's slow-path") Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: rename switchdev to eswitchMichal Swiatkowski4-42/+43
Eswitch is used as a prefix for related functions. Main structure storing all data related to eswitch should also be named as eswitch instead of ice_switchdev_info. Rename it. Also rename switchdev to eswitch where the context is not about eswitch mode. ::uplink_netdev was changed to netdev for simplicity. There is no other netdev in function scope so it is obvious. Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Piotr Raczynski <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Michal Swiatkowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: fix DDP package download for packages without signature segmentDan Nowlin1-3/+100
Commit 3cbdb0343022 ("ice: Add support for E830 DDP package segment") incorrectly removed support for package download for packages without a signature segment. These packages include the signature buffer inline in the configurations buffers, and not in a signature segment. Fix package download by providing download support for both packages with (ice_download_pkg_with_sig_seg()) and without signature segment (ice_download_pkg_without_sig_seg()). Fixes: 3cbdb0343022 ("ice: Add support for E830 DDP package segment") Reported-by: Maciej Fijalkowski <[email protected]> Closes: https://lore.kernel.org/netdev/ZUT50a94kk2pMGKb@boxer/ Tested-by: Maciej Fijalkowski <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Dan Nowlin <[email protected]> Signed-off-by: Paul Greenwalt <[email protected]> Reviewed-by: Simon Horman <[email protected]> Tested-by: Arpana Arland <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: dpll: fix output pin capabilitiesArkadiusz Kubalewski3-4/+64
The dpll output pins which are used to feed clock signal of PHY and MAC circuits cannot be disconnected, those integrated circuits require clock signal for operation. By stopping assignment of DPLL_PIN_CAPABILITIES_STATE_CAN_CHANGE pin capability, prevent the user from invoking the state set callback on those pins, setting the state on those pins already returns error, as firmware doesn't allow the change of their state. Fixes: d7999f5ea64b ("ice: implement dpll interface to control cgu") Fixes: 8a3a565ff210 ("ice: add admin commands to access cgu configuration") Reviewed-by: Andrii Staikov <[email protected]> Signed-off-by: Arkadiusz Kubalewski <[email protected]> Tested-by: Sunitha Mekala <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: dpll: fix check for dpll input priority rangeArkadiusz Kubalewski2-7/+0
Supported priority value for input pins may differ with regard of NIC firmware version. E810T NICs with 3.20/4.00 FW versions would accept priority range 0-31, where firmware 4.10+ would support the range 0-9 and extra value of 255. Remove the in-range check as the driver has no information on supported values from the running firmware, let firmware decide if given value is correct and return extack error if the value is not supported. Fixes: d7999f5ea64b ("ice: implement dpll interface to control cgu") Reviewed-by: Przemek Kitszel <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Arkadiusz Kubalewski <[email protected]> Tested-by: Sunitha Mekala <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
2023-11-13ice: dpll: fix initial lock status of dpllArkadiusz Kubalewski1-2/+1
When dpll device is registered and dpll subsystem performs notify of a new device, the lock state value provided to dpll subsystem equals 0 which is invalid value for the `enum dpll_lock_status`. Provide correct value by obtaining it from firmware before registering the dpll device. Fixes: d7999f5ea64b ("ice: implement dpll interface to control cgu") Signed-off-by: Aleksandr Loktionov <[email protected]> Signed-off-by: Arkadiusz Kubalewski <[email protected]> Tested-by: Sunitha Mekala <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
2023-11-08Merge branch '100GbE' of ↵Jakub Kicinski2-41/+91
git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue Tony Nguyen says: ==================== Intel Wired LAN Driver Updates 2023-11-06 (ice) This series contains updates to ice driver only. Dave removes SR-IOV LAG attribute for only the interface being disabled to allow for proper unwinding of all interfaces. Michal Schmidt changes some LAG allocations from GFP_KERNEL to GFP_ATOMIC due to non-allowed sleeping. Aniruddha and Marcin fix redirection and drop rules for switchdev by properly setting and marking egress/ingress type. * '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue: ice: Fix VF-VF direction matching in drop rule in switchdev ice: Fix VF-VF filter rules in switchdev mode ice: lag: in RCU, use atomic allocation ice: Fix SRIOV LAG disable on non-compliant aggregate ==================== Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-08Merge branch '40GbE' of ↵Jakub Kicinski2-5/+6
git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue Tony Nguyen says: ==================== Intel Wired LAN Driver Updates 2023-11-06 (i40e) This series contains updates to i40e driver only. Ivan Vecera resolves a couple issues with devlink; removing a call to devlink_port_type_clear() and ensuring devlink port is unregistered after the net device. * '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue: i40e: Fix devlink port unregistering i40e: Do not call devlink_port_type_clear() ==================== Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-06idpf: fix potential use-after-free in idpf_tso()Eric Dumazet1-2/+4
skb_cow_head() can change skb->head (and thus skb_shinfo(skb)) We must not cache skb_shinfo(skb) before skb_cow_head(). Fixes: 6818c4d5b3c2 ("idpf: add splitq start_xmit") Signed-off-by: Eric Dumazet <[email protected]> Cc: Joshua Hay <[email protected]> Cc: Alan Brady <[email protected]> Cc: Madhu Chittim <[email protected]> Cc: Phani Burra <[email protected]> Cc: Sridhar Samudrala <[email protected]> Cc: Willem de Bruijn <[email protected]> Cc: Pavan Kumar Linga <[email protected]> Cc: Tony Nguyen <[email protected]> Cc: Bailey Forrest <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-11-06ice: Fix VF-VF direction matching in drop rule in switchdevMarcin Szycik1-1/+23
When adding a drop rule on a VF, rule direction is not being set, which results in it always being set to ingress (ICE_ESWITCH_FLTR_INGRESS equals 0). Because of this, drop rules added on port representors don't match any packets. To fix it, set rule direction in drop action to egress when netdev is a port representor, otherwise set it to ingress. Fixes: 0960a27bd479 ("ice: Add direction metadata") Reviewed-by: Michal Swiatkowski <[email protected]> Signed-off-by: Marcin Szycik <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Reviewed-by: Simon Horman <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-06ice: Fix VF-VF filter rules in switchdev modeAniruddha Paul1-28/+62
Any packet leaving VSI i.e VF's VSI is considered as egress traffic by HW, thus failing to match the added rule. Mark the direction for redirect rules as below: 1. VF-VF - Egress 2. Uplink-VF - Ingress 3. VF-Uplink - Egress 4. Link_Partner-Uplink - Ingress 5. Link_Partner-VF - Ingress Fixes: 0960a27bd479 ("ice: Add direction metadata") Reviewed-by: Przemek Kitszel <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Signed-off-by: Aniruddha Paul <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-06ice: lag: in RCU, use atomic allocationMichal Schmidt1-3/+3
Sleeping is not allowed in RCU read-side critical sections. Use atomic allocations under rcu_read_lock. Fixes: 1e0f9881ef79 ("ice: Flesh out implementation of support for SRIOV on bonded interface") Fixes: 41ccedf5ca8f ("ice: implement lag netdev event handler") Fixes: 3579aa86fb40 ("ice: update reset path for SRIOV LAG support") Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Reviewed-by: Simon Horman <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-06ice: Fix SRIOV LAG disable on non-compliant aggregateDave Ertman1-9/+3
If an attribute of an aggregate interface disqualifies it from supporting SRIOV, the driver will unwind the SRIOV support. Currently the driver is clearing the feature bit for all interfaces in the aggregate, but this is not allowing the other interfaces to unwind successfully on driver unload. Only clear the feature bit for the interface that is currently unwinding. Fixes: bf65da2eb279 ("ice: enforce interface eligibility and add messaging for SRIOV LAG") Signed-off-by: Dave Ertman <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Simon Horman <[email protected]> Tested-by: Sujai Buvaneswaran <[email protected]> Signed-off-by: Tony Nguyen <[email protected]>
2023-11-06i40e: Fix devlink port unregisteringIvan Vecera1-4/+6
Ensure that devlink port is unregistered after unregistering of net device. Reproducer: [root@host ~]# rmmod i40e [ 4742.939386] i40e 0000:02:00.1: i40e_ptp_stop: removed PHC on enp2s0f1np1 [ 4743.059269] ------------[ cut here ]------------ [ 4743.063900] WARNING: CPU: 21 PID: 10766 at net/devlink/port.c:1078 devl_port_unregister+0x69/0x80 ... Fixes: 9e479d64dc58 ("i40e: Add initial devlink support") Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jiri Pirko <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
2023-11-06i40e: Do not call devlink_port_type_clear()Ivan Vecera1-1/+0
Do not call devlink_port_type_clear() prior devlink port unregister and let devlink core to take care about it. Reproducer: [root@host ~]# rmmod i40e [ 4539.964699] i40e 0000:02:00.0: devlink port type for port 0 cleared without a software interface reference, device type not supported by the kernel? [ 4540.319811] i40e 0000:02:00.1: devlink port type for port 1 cleared without a software interface reference, device type not supported by the kernel? Fixes: 9e479d64dc58 ("i40e: Add initial devlink support") Signed-off-by: Ivan Vecera <[email protected]> Reviewed-by: Jiri Pirko <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
2023-10-27iavf: delete the iavf client interfaceMichal Schmidt6-871/+1
The iavf client interface was added in 2017 by commit ed0e894de7c1 ("i40evf: add client interface"), but there have never been any in-tree callers. It's not useful for future development either. The Intel out-of-tree iavf and irdma drivers instead use an auxiliary bus, which is a better solution. Remove the iavf client interface code. Also gone are the client_task work and the client_lock mutex. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: add a common function for undoing the interrupt schemeMichal Schmidt1-11/+15
Add a new function iavf_free_interrupt_scheme that does the inverse of iavf_init_interrupt_scheme. Symmetry is nice. And there will be three callers already. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: use unregister_netdevMichal Schmidt1-3/+1
Use unregister_netdev, which takes rtnl_lock for us. We don't have to check the reg_state under rtnl_lock. There's nothing to race with. We have just cancelled the finish_config work. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: rely on netdev's own registered stateMichal Schmidt2-7/+3
The information whether a netdev has been registered is already present in the netdev itself. There's no need for a driver flag with the same meaning. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Reviewed-by: Jacob Keller <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: fix the waiting time for initial resetMichal Schmidt1-1/+1
Every time I create VFs on ice, I receive at least one "Device is still in reset (-16), retrying" message per VF. It recovers fine, but typical usecases should not trigger scary-looking messages. The waiting for reset is too short. It makes no sense to check every 10 microseconds. Typical reset waiting times are at least tens of milliseconds and can be several seconds. I suspect the polling interval was meant to be 10 milliseconds all along. IAVF_RESET_WAIT_COMPLETE_COUNT is defined as 2000, so the total waiting time could be over 20 seconds. I have seen resets take 5 seconds (with 128 VFs on ice). The added benefit of not triggering the "Device is still in reset" path is that we avoid going through the __IAVF_INIT_FAILED state, which would take a full second before retrying. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: in iavf_down, don't queue watchdog_task if comms failedMichal Schmidt1-2/+4
The reason for queueing watchdog_task is to have it process the aq_required flags that are being set here. If comms failed, there's nothing to do, so return early. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: simplify mutex_trylock+sleep loopsMichal Schmidt1-4/+2
This pattern appears in two places in the iavf source code: while (!mutex_trylock(...)) usleep_range(...); That's just mutex_lock with extra steps. The pattern is a leftover from when iavf used bit flags instead of mutexes for locking. Commit 5ac49f3c2702 ("iavf: use mutexes for locking of critical sections") replaced test_and_set_bit with !mutex_trylock, preserving the pattern. Simplify it to mutex_lock. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
2023-10-27iavf: fix comments about old bit locksMichal Schmidt1-2/+2
Bit lock __IAVF_IN_CRITICAL_TASK does not exist anymore since commit 5ac49f3c2702 ("iavf: use mutexes for locking of critical sections"). Adjust the comments accordingly. Signed-off-by: Michal Schmidt <[email protected]> Reviewed-by: Wojciech Drewek <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Jacob Keller <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>
Powered by cgit, Gruvboxed by Blaster4385.