- 10 Oct, 2021 21 commits
-
-
Christophe JAILLET authored
'destroy_workqueue()' already drains the queue before destroying it, so there is no need to flush it explicitly. Remove the redundant 'flush_workqueue()' calls. This was generated with coccinelle: @@ expression E; @@ - flush_workqueue(E); destroy_workqueue(E); Signed-off-by:
Christophe JAILLET <christophe.jaillet@wanadoo.fr> Reviewed-by: Leon Romanovsky <leonro@nvidia.com> #mlx* Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Li RongQing authored
networking benchmark shows that __rcu_read_lock and __rcu_read_unlock takes some cpu cycles, and we can avoid calling them partially in virtio rx path by check xdp_enabled of vi, and xdp is disabled most of time Signed-off-by:
Li RongQing <lirongqing@baidu.com> Signed-off-by:
-
Antoine Tenart authored
Cosmetic commit making dev_get_port_parent_id slightly more readable. There is no need to split the condition to return after calling devlink_compat_switch_id_get and after that 'recurse' is always true. Signed-off-by:
Antoine Tenart <atenart@kernel.org> Signed-off-by:
-
Ansuel Smith authored
Give a name to known debug regs from Documentation instead of using unknown hex values. Signed-off-by:
Ansuel Smith <ansuelsmth@gmail.com> Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Ansuel Smith authored
From original QCA source code the port was set to prefer master as port type in 1000BASE-T mode. Apply the same settings also here. Signed-off-by:
-
Ansuel Smith authored
QCA8327 internal phy require DAC amplitude adjustement set to +6% with 100m speed. Also add additional define to report a change of the same reg in QCA8337. (different scope it does set 1000m voltage) Add link_change_notify function to set the proper amplitude adjustement on PHY_RUNNING state and disable on any other state. Fixes: b4df02b5 ("net: phy: at803x: add support for qca 8327 A variant internal phy") Signed-off-by:
-
Ansuel Smith authored
From Documentation phy resume triggers phy reset and restart auto-negotiation. Add a dedicated function to wait reset to finish as it was notice a regression where port sometime are not reliable after a suspend/resume session. The reset wait logic is copied from phy_poll_reset. Add dedicated suspend function to use genphy_suspend only with QCA8337 phy and set only additional debug settings for QCA8327. With more test it was reported that QCA8327 doesn't proprely support this mode and using this cause the unreliability of the switch ports, especially the malfunction of the port0. Fixes: 15b9df4e ("net: phy: at803x: add resume/suspend function to qca83xx phy") Signed-off-by:
-
David S. Miller authored
Juhee Kang says: ==================== net-next: replace open code with helper functions Currently, there are many helper functions on netdevice.h. However, some code doesn't use the helper functions and remains open code. So this patchset replaces open code with an appropriate helper function. First patch modifies to use netif_is_rxfh_configured instead of dev->priv_flags & IFF_RXFH_CONFIGURED. Second patch replaces open code with netif_is_bond_master. Last patch substitutes netif_is_macsec() for dev->priv_flags & IFF_MACSEC. ==================== Signed-off-by: -
Juhee Kang authored
Open code which is dev->priv_flags & IFF_MACSEC has already defined as netif_is_macsec(). So use netif_is_macsec() instead of open code. This patch doesn't change logic. Signed-off-by:
Juhee Kang <claudiajkang@gmail.com> Signed-off-by:
-
Juhee Kang authored
Use netif_is_bond_master() function instead of open code, which is ((event_dev->priv_flags & IFF_BONDING) && (event_dev->flags & IFF_MASTER)). This patch doesn't change logic. Signed-off-by:
-
Juhee Kang authored
The open code which is dev->priv_flags & IFF_RXFH_CONFIGURED is defined as a helper function on netdevice.h. So use netif_is_rxfh_configured() function instead of open code. This patch doesn't change logic. Signed-off-by:
-
David S. Miller authored
Shannon Nelson says: ==================== ionic: add vlanid overflow management Add vlans to the existing rx_filter_sync mechanics currently used for managing mac filters. Older versions of our firmware had no enforced limits on the number of vlans that the driver could request, but requesting large numbers of vlans caused issues in FW memory management, so an arbitrary limit was added in the FW. The FW now returns -ENOSPC when it hits that limit, which the driver needs to handle. Unfortunately, the FW doesn't advertise the vlan id limit, as it does with mac filters, so the driver won't know the limit until it bumps into it. We'll grab the current vlan id count and use that as the limit from there on and thus prevent getting any more -ENOSPC errors. Just as is done for the mac filters, the device puts the device into promiscuous mode when -ENOSPC is seen for vlan ids, and the driver will track the vlans that aren't synced to the FW. When vlans are removed, the driver will retry the un-synced vlans. If all outstanding vlans are synced, the promiscuous mode will be disabled. The first 6 patches rework the existing filter management to make it flexible enough for additional filter types. Next we add the vlan ids into the management. The last 2 patches allow us to catch the max vlan -ENOSPC error without adding an unnecessary error message to the kernel log. ==================== Signed-off-by: -
Shannon Nelson authored
Override the automatic AdminQ error message in order to capture the potential No Space message when we hit the max vlan limit, and add additional messaging to detail what filter failed. Signed-off-by:
Shannon Nelson <snelson@pensando.io> Signed-off-by:
-
Shannon Nelson authored
The AdminQ handler has an error handler that automatically prints an error message when the request has failed. However, there are situations where the caller can expect that it might fail and has an alternative strategy, thus may not want the error message sent to the log, such as hitting -ENOSPC when adding a new vlan id. We add a new interface to the AdminQ API to allow for override of the default behavior, and an interface to the use standard error message formatting. Signed-off-by:
-
Shannon Nelson authored
Add vlans to the existing rx_filter_sync mechanics currently used for managing mac filters. Older versions of our firmware had no enforced limits on the number of vlans that the LIF could request, but requesting large numbers of vlans caused issues in FW memory management, so an arbitrary limit was added in the FW. The FW now returns -ENOSPC when it hits that limit, which the driver needs to handle. Unfortunately, the FW doesn't advertise the vlan id limit, as it does with mac filters, so the driver won't know the limit until it bumps into it. We'll grab the current vlan id count and use that as the limit from there on and thus prevent getting any more -ENOSPC errors. Just as is done for the mac filters, the device puts the device into promiscuous mode when -ENOSPC is seen for vlan ids, and the driver will track the vlans that aren't synced to the FW. When vlans are removed, the driver will retry the un-synced vlans. If all outstanding vlans are synced, the promiscuous mode will be disabled. Signed-off-by:
-
Shannon Nelson authored
Similar to the filter add, make a generic filter delete. Signed-off-by:
-
Shannon Nelson authored
In preparation for adding vlan overflow management, rework the ionic_lif_addr_add() function to something a little more generic that can be used for other filter types. Signed-off-by:
-
Shannon Nelson authored
In preparation for enhancing vlan filter management, add a filter search routine that can figure out for itself which type of filter search is needed. Signed-off-by:
-
Shannon Nelson authored
The overflow flags really aren't useful and we don't need lif struct elements to track them. Signed-off-by:
-
Shannon Nelson authored
The routines that add and delete mac addresses from the firmware really should be in the file with the rest of the filter management. This simply moves the functions with no logic changes. Signed-off-by:
-
Shannon Nelson authored
Dump the filter list to debugfs - includes the device-assigned filter id and the sync'd-to-hardware status. Signed-off-by:
-
- 09 Oct, 2021 9 commits
-
-
Vladimir Oltean authored
The Vitesse/Microsemi/Microchip family of switches supported by the Felix DSA driver is capable of selecting between the native tagging protocol and ocelot-8021q. This is necessary to enable flow control on the CPU port. Certain systems where these switches are integrated use the switch as a port multiplexer, so the termination throughput is paramount. Changing the tagging protocol at runtime is possible for these systems, but since it is known beforehand that one tagging protocol will provide strictly better performance than the other, just allow them to specify the preference in the device tree. Signed-off-by:
Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by:
-
Vladimir Oltean authored
There is a trivial typo when spelling "protocol", fix it. Signed-off-by:
-
Jakub Kicinski authored
Use dev_addr_set() instead of writing directly to netdev->dev_addr in various misc and old drivers. Signed-off-by:
Jakub Kicinski <kuba@kernel.org> Signed-off-by:
-
David S. Miller authored
Jakub Kicinski says: ==================== net: remove direct netdev->dev_addr writes Commit 406f42fa ("net-next: When a bond have a massive amount of VLANs...") introduced a rbtree for faster Ethernet address look up. To maintain netdev->dev_addr in this tree we need to make all the writes to it got through appropriate helpers. This series contains top 5 conversions in terms of LoC required to bring the driver into compliance. ==================== Signed-off-by:
-
Jakub Kicinski authored
8390 contains a lot of loops assigning netdev->dev_addr byte by byte. Convert what's possible directly to eth_hw_addr_set(), use local buf in other places. Signed-off-by:
-
Jakub Kicinski authored
Consify temporary variables pointing to netdev->dev_addr. A few places need local storage but pretty simple conversion over all. Note that macaddr[] is an array of ints, so we need to keep the loops. Signed-off-by:
-
Jakub Kicinski authored
Consify the casts of netdev->dev_addr. Convert pointless to eth_hw_addr_set() where possible. Use local buffers in a number of places. Signed-off-by:
-
Jakub Kicinski authored
tg3 does various forms of direct writes to netdev->dev_addr. Use a local buffer. Make sure local buffer is aligned since eth_platform_get_mac_address() may call ether_addr_copy(). tg3_get_device_address() returns whenever it finds a method that found a valid address. Instead of modifying all the exit points pass the buffer from the outside and commit the address in the caller. Constify the argument of the set addr helper. Signed-off-by:
Michael Chan <michael.chan@broadcom.com> Signed-off-by:
-
Jakub Kicinski authored
forcedeth writes to dev_addr byte by byte, make it use a local buffer instead. Commit the changes with eth_hw_addr_set() at the end. Signed-off-by:
Zhu Yanjun <zyjzyj2000@gmail.com> Signed-off-by:
-
- 08 Oct, 2021 10 commits
-
-
Ido Schimmel authored
mlxsw is using helpers to get / set fields in messages exchanged with the device. It is possible that some fields are only set or only get. This causes LLVM to emit warnings such as the following when building with W=1 [1]: drivers/net/ethernet/mellanox/mlxsw/core_acl_flex_actions.c:2022:1: warning: unused function 'mlxsw_afa_sampler_mirror_agent_get' The fact that some fields are only set or only get is very much intentional and not indicative of functions that need to be removed. Therefore, annotate the item helpers with '__maybe_unused' to suppress these warnings. [1] https://lkml.org/lkml/2021/9/29/685 Cc: Nathan Chancellor <nathan@kernel.org> Reported-by:
kernel test robot <lkp@intel.com> Signed-off-by:
Ido Schimmel <idosch@nvidia.com> Reviewed-by:
Jiri Pirko <jiri@nvidia.com> Link: https://lore.kernel.org/r/20211008132315.90211-1-idosch@idosch.orgSigned-off-by:
-
Tianjia Zhang authored
Add new cipher as a variant of standard tls selftests. Signed-off-by:
Tianjia Zhang <tianjia.zhang@linux.alibaba.com> Link: https://lore.kernel.org/r/20211008091745.42917-1-tianjia.zhang@linux.alibaba.comSigned-off-by:
-
Jean Sacren authored
Remove the redundant check of (tg3_asic_rev(tp) == ASIC_REV_5705) after it is checked to be true. Signed-off-by:
Jean Sacren <sakiwit@gmail.com> Reviewed-by:
-
Prabhakar Kushwaha authored
This patch fixes below compliation error in case CONFIG_QED_SRIOV not defined. drivers/net/ethernet/qlogic/qed/qed_dev.c: In function ‘qed_fw_err_handler’: drivers/net/ethernet/qlogic/qed/qed_dev.c:2390:3: error: implicit declaration of function ‘qed_sriov_vfpf_malicious’; did you mean ‘qed_iov_vf_task’? [-Werror=implicit-function-declaration] qed_sriov_vfpf_malicious(p_hwfn, &data->err_data); ^~~~~~~~~~~~~~~~~~~~~~~~ qed_iov_vf_task drivers/net/ethernet/qlogic/qed/qed_dev.c: In function ‘qed_common_eqe_event’: drivers/net/ethernet/qlogic/qed/qed_dev.c:2410:10: error: implicit declaration of function ‘qed_sriov_eqe_event’; did you mean ‘qed_common_eqe_event’? [-Werror=implicit-function-declaration] return qed_sriov_eqe_event(p_hwfn, opcode, echo, data, ^~~~~~~~~~~~~~~~~~~ qed_common_eqe_event Fixes: fe40a830 ("qed: Update qed_hsi.h for fw 8.59.1.0") Reported-by:Linux Kernel Functional Testing <lkft@linaro.org> Cc: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by:
Ariel Elior <aelior@marvell.com> Signed-off-by:
Shai Malin <smalin@marvell.com> Signed-off-by:
Omkar Kulkarni <okulkarni@marvell.com> Signed-off-by:
Prabhakar Kushwaha <pkushwaha@marvell.com> Signed-off-by:
-
Francesco Dolcini authored
Micrel KSZ9131 PHY LED behavior is not correct when configured in Individual Mode, LED1 (Activity LED) is in the ON state when there is no-link. Workaround this by setting bit 9 of register 0x1e after verifying that the LED configuration is Individual Mode. This issue is described in KSZ9131RNX Silicon Errata DS80000693B [*] and according to that it will not be corrected in a future silicon revision. [*] https://ww1.microchip.com/downloads/en/DeviceDoc/KSZ9131RNX-Silicon-Errata-and-Data-Sheet-Clarification-80000863B.pdfSigned-off-by:
Francesco Dolcini <francesco.dolcini@toradex.com> Reviewed-by:
Philippe Schenker <philippe.schenker@toradex.com> Reviewed-by:
-
David S. Miller authored
Antoine Tenart says: ==================== net: introduce a function to check if a netdev name is in use This was initially part of an RFC series[1] but has value on its own; hence the standalone report. (It will also help in not having a series too large). From patch 1: """ __dev_get_by_name is currently used to either retrieve a net device reference using its name or to check if a name is already used by a registered net device (per ns). In the later case there is no need to return a reference to a net device. Introduce a new helper, netdev_name_in_use, to check if a name is currently used by a registered net device without leaking a reference the corresponding net device. This helper uses netdev_name_node_lookup instead of __dev_get_by_name as we don't need the extra logic retrieving a reference to the corresponding net device. """ Two uses[2] of __dev_get_by_name weren't converted to this new function, as they are really looking for a net device, not only checking if a net device name is in use. While checking one or the other currently has the same result, that might change if the initial RFC series moves forward. I'll convert them later depending on the outcome of the initial series. Thanks, Antoine [1] https://lore.kernel.org/all/20210928125500.167943-1-atenart@kernel.org/ [2] drivers/net/Space.c:130 & drivers/nvme/host/tcp.c:2550 ==================== Signed-off-by:
-
Antoine Tenart authored
A new helper to detect if a net device name is in use was added. Use it here as the return reference from __dev_get_by_name was discarded. Signed-off-by:
-
Antoine Tenart authored
A new helper to detect if a net device name is in use was added. Use it here as the return reference from __dev_get_by_name was discarded. Signed-off-by:
-
Antoine Tenart authored
__dev_get_by_name is currently used to either retrieve a net device reference using its name or to check if a name is already used by a registered net device (per ns). In the later case there is no need to return a reference to a net device. Introduce a new helper, netdev_name_in_use, to check if a name is currently used by a registered net device without leaking a reference the corresponding net device. This helper uses netdev_name_node_lookup instead of __dev_get_by_name as we don't need the extra logic retrieving a reference to the corresponding net device. Signed-off-by:
-
David S. Miller authored
Ioana Ciornei says: ==================== net: enetc: add support for software TSO This series adds support for driver level TSO in the enetc driver. Ever since the ENETC MDIO erratum workaround is in place, the Tx path is incurring a penalty (enetc_lock_mdio/enetc_unlock_mdio) for each skb to be sent out. On top of this, ENETC does not support Tx checksum offloading. This means that software TSO would help performance just by the fact that one single mdio lock/unlock sequence would cover multiple packets sent. On the other hand, checksum needs to be computed in software since the controller cannot handle it. This is why, beside using the usual tso_build_hdr()/tso_build_data() this specific implementation also has to compute the checksum, both IP and L4, for each resulted segment. Even with that, the performance improvement of a TCP flow running on a single A72@1.3GHz of the LS1028A SoC (2.5Gbit/s port) is the following: before: 1.63 Gbits/sec after: 2.34 Gbits/sec Changes in v2: - declare NETIF_F_HW_CSUM instead of NETIF_F_IP_CSUM in 1/2 - add support for TSO over IPv6 (NETIF_F_TSO6 and csum compute) in 2/2 ==================== Signed-off-by:
-
