- 18 Nov, 2020 12 commits
-
-
Jakub Kicinski authored
Ido Schimmel says: ==================== mlxsw: Preparations for nexthop objects support - part 2/2 This patch set contains the second round of preparations towards nexthop objects support in mlxsw. Follow up patches can be found here [1]. The patches are mostly small and trivial and contain non-functional changes aimed at making it easier to integrate nexthop objects with mlxsw. Patch #1 is a fix for an issue introduced in previous submission. Found by Coverity. [1] https://github.com/idosch/linux/tree/submit/nexthop_objects ==================== Link: https://lore.kernel.org/r/20201117174704.291990-1-idosch@idosch.orgSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Ido Schimmel authored
The function is responsible for allocating the adjacency entries used by the nexthop group and populating them with the adjacency information such as egress RIF and MAC address. Allow the function to return an error when it encounters a problem and have the relevant call sites check it. Signed-off-by:
Ido Schimmel <idosch@nvidia.com> Reviewed-by:
Jiri Pirko <jiri@nvidia.com> Signed-off-by:
-
Ido Schimmel authored
Currently, a nexthop group is destroyed when the last FIB entry is detached from it. When nexthop objects are supported, this can no longer be the case, as the group is a separate object whose lifetime is managed by user space. Add an indication if a nexthop group can be destroyed and always set it to true for the existing IPv4 and IPv6 nexthop groups. Signed-off-by:
-
Ido Schimmel authored
When the IPv6 FIB info has a nexthop object, the nexthop offload indication is set on the nexthop object and not on the FIB info itself. Therefore, do not try to clear the offload indication from the FIB info when it has a nexthop object. Signed-off-by:
-
Ido Schimmel authored
Attach the FIB entry to the nexthop group after setting the offload flag on the IPv6 FIB info (i.e., 'struct fib6_info'). The second operation is not needed when the nexthop group is a nexthop object. This will allow us to have a common exit path from the function, regardless of the nexthop group's type. Signed-off-by:
-
Ido Schimmel authored
The previous patch associated a nexthop group with the FIB entry before the entry's type is determined. Make use of the nexthop group when determining the entry's type instead of relying on helpers that assume that the nexthop info is not a nexthop object (i.e., 'struct nexthop'). Signed-off-by:
-
Ido Schimmel authored
Each FIB entry has a type (e.g., remote, local) that determines how the entry is programmed to the device. In order to determine if the entry is local (directly connected) or remote (has a gateway) the relevant FIB info structures (e.g., 'struct fib_info') are checked. When entries that use nexthop objects are supported, these checks will need to be changed to take into account 'struct nexthop'. Instead, first associate the entry with a nexthop group so that the next patch could determine the entry's type based on the associated nexthop group's type. Signed-off-by:
-
Ido Schimmel authored
The sole caller of the function will soon only have the ifindex available, instead of the pointer itself. Therefore, change the function to take the ifindex as input and have it get the pointer. Signed-off-by:
-
Ido Schimmel authored
The ifindex of the nexthop device was never set for IPv4 nexthops, unlike IPv6 nexthops. This went unnoticed since only IPv6 nexthops use it. Set the ifindex for IPv4 nexthops in order to be consistent with IPv6 and also because it will be used by a later patch. Signed-off-by:
-
Ido Schimmel authored
The function allocates 'nhgi', not 'nh_grp', so it needs to free the former in its error path. Fixes: 7f7a417e ("mlxsw: spectrum_router: Split nexthop group configuration to a different struct") Addresses-Coverity: ("Memory - corruptions (USE_AFTER_FREE)") Reviewed-by:
-
Ahmad Fatoum authored
It's arguable most people interested in configuring a PPS signal want it as external output, not as kernel input. PTP_CLK_REQ_PPS is for input though. Add documentation to nudge readers into the correct direction. Signed-off-by:
Ahmad Fatoum <a.fatoum@pengutronix.de> Acked-by:
Richard Cochran <richardcochran@gmail.com> Link: https://lore.kernel.org/r/20201117213826.18235-1-a.fatoum@pengutronix.deSigned-off-by:
-
Gustavo A. R. Silva authored
Fix the following unreachable code issue: drivers/net/ethernet/netronome/nfp/crypto/tls.c: In function 'nfp_net_tls_add': include/linux/compiler_attributes.h:208:41: warning: statement will never be executed [-Wswitch-unreachable] 208 | # define fallthrough __attribute__((__fallthrough__)) | ^~~~~~~~~~~~~ drivers/net/ethernet/netronome/nfp/crypto/tls.c:299:3: note: in expansion of macro 'fallthrough' 299 | fallthrough; | ^~~~~~~~~~~ Reported-by:kernel test robot <lkp@intel.com> Signed-off-by:
Gustavo A. R. Silva <gustavoars@kernel.org> Reviewed-by:
Simon Horman <simon.horman@netronome.com> Link: https://lore.kernel.org/r/20201117171347.GA27231@embeddedorSigned-off-by:
-
- 17 Nov, 2020 28 commits
-
-
Jakub Kicinski authored
Mauro Carvalho Chehab says: ==================== Fix several bad kernel-doc markups Kernel-doc has always be limited to a probably bad documented rule: The kernel-doc markups should appear *imediatelly before* the function or data structure that it documents. On other words, if a C file would contain something like this: /** * foo - function foo * @args: foo args */ static inline void bar(int args); /** * bar - function bar * @args: foo args */ static inline void foo(void *args); The output (in ReST format) will be: .. c:function:: void bar (int args) function foo **Parameters** ``int args`` foo args .. c:function:: void foo (void *args) function bar **Parameters** ``void *args`` foo args Which is clearly a wrong result. Before this changeset, not even a warning is produced on such cases. As placing such markups just before the documented data is a common practice, on most cases this is fine. However, as patches touch things, identifiers may be renamed, and people may forget to update the kernel-doc markups to follow such changes. This has been happening for quite a while, as there are lots of files with kernel-doc problems. This series address those issues and add a file at the end that will enforce that the identifier will match the kernel-doc markup, avoiding this problem from keep happening as time goes by. ==================== Link: https://lore.kernel.org/r/cover.1605521731.git.mchehab+huawei@kernel.orgSigned-off-by:
-
Mauro Carvalho Chehab authored
Some identifiers have different names between their prototypes and the kernel-doc markup. In the specific case of netif_subqueue_stopped(), keep the current markup for __netif_subqueue_stopped(), adding a new one for netif_subqueue_stopped(). Signed-off-by:
Mauro Carvalho Chehab <mchehab+huawei@kernel.org> Signed-off-by:
-
Mauro Carvalho Chehab authored
Some identifiers have different names between their prototypes and the kernel-doc markup. Signed-off-by:
Kirill Tkhai <ktkhai@virtuozzo.com> Signed-off-by:
-
Mauro Carvalho Chehab authored
Some functions have different names between their prototypes and the kernel-doc markup. Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Jakub Kicinski authored
Naveen Mamindlapalli says: ==================== Add ethtool ntuple filters support This patch series adds support for ethtool ntuple filters, unicast address filtering, VLAN offload and SR-IOV ndo handlers. All of the above features are based on the Admin Function(AF) driver support to install and delete the low level MCAM entries. Each MCAM entry is programmed with the packet fields to match and what actions to take if the match succeeds. The PF driver requests AF driver to allocate set of MCAM entries to be used to install the flows by that PF. The entries will be freed when the PF driver is unloaded. * The patches 1 to 4 adds AF driver infrastructure to install and delete the low level MCAM flow entries. * Patch 5 adds ethtool ntuple filter support. * Patch 6 adds unicast MAC address filtering. * Patch 7 adds support for dumping the MCAM entries via debugfs. * Patches 8 to 10 adds support for VLAN offload. * Patch 10 to 11 adds support for SR-IOV ndo handlers. * Patch 12 adds support to read the MCAM entries. Misc: * Removed redundant mailbox NIX_RXVLAN_ALLOC. ==================== Link: https://lore.kernel.org/r/20201114195303.25967-1-naveenm@marvell.comSigned-off-by:
-
Subbaraya Sundeep authored
Since mailbox message for installing flows is in place, remove the RXVLAN_ALLOC mbox message which is redundant. Signed-off-by:
Subbaraya Sundeep <sbhatta@marvell.com> Signed-off-by:
Sunil Goutham <sgoutham@marvell.com> Signed-off-by:
Naveen Mamindlapalli <naveenm@marvell.com> Signed-off-by:
-
Naveen Mamindlapalli authored
This patch introduces new mailbox mesages to retrieve a given MCAM entry or base flow steering rule of a VF installed by its parent PF. This helps while updating the existing MCAM rules with out re-framing the whole mailbox request again. The INSTALL FLOW mailbox consumer can read-modify-write the existing entry. Similarly while installing new flow rules for a VF, the base flow steering rule match creteria is copied to the new flow rule and the deltas are appended to the new rule. Signed-off-by:
Vamsi Attunuru <vattunuru@marvell.com> Signed-off-by:
-
Hariprasad Kelam authored
This patch handles the VF mac address changes as given below. 1. mac addr configrued by VF will be retained until VF module unload. 2. mac addr configred by PF for VF will be retained until power cycle. 3. mac addr confgired by PF for its VF can't be overwritten by VF. Signed-off-by:Hariprasad Kelam <hkelam@marvell.com> Signed-off-by:
-
Naveen Mamindlapalli authored
This patch adds support for ndo_set_vf_mac, ndo_set_vf_vlan and ndo_get_vf_config handlers. The traffic redirection based on the VF mac address or vlan id is done by installing MCAM rules. Reserved RX_VTAG_TYPE7 in each NIXLF for VF VLAN which strips the VLAN tag from ingress VLAN traffic. The NIX PF allocates two MCAM entries for VF VLAN feature, one used for ingress VTAG strip and another entry for egress VTAG insertion. This patch also updates the MAC address in PF installed VF VLAN rule upon receiving nix_lf_start_rx mbox request for VF since Administrative Function driver will assign a valid MAC addr in nix_lf_start_rx function. Signed-off-by:
Tomasz Duszynski <tduszynski@marvell.com> Signed-off-by:
-
Hariprasad Kelam authored
This patch implements egress VLAN offload by appending NIX_SEND_EXT_S header to NIX_SEND_HDR_S. The VLAN TCI information is specified in the NIX_SEND_EXT_S. The VLAN offload in the ingress path is implemented by configuring the NIX_RX_VTAG_ACTION_S to strip and capture the outer vlan fields. The NIX PF allocates one MCAM entry for Rx VLAN offload. Signed-off-by:
-
Vamsi Attunuru authored
This patch modifies the existing nix_vtag_config mailbox message to allocate and free TX VTAG entries as requested by a NIX PF. The TX VTAG entries are global resource that shared by all PFs and each entry specifies the size of VTAG to insert and the VTAG header data to insert. The mailbox response contains the entry index which is used by mailbox requester in configuring the NPC_TX_VTAG_ACTION for any MCAM entry. Signed-off-by:
-
Subbaraya Sundeep authored
Add debugfs support to dump the MCAM rules installed using NPC_INSTALL_FLOW mbox message. Debugfs file can display mcam entry, counter if any, flow type and counter hits. Ethtool will dump the ntuple flows related to the PF only. The debugfs file gives systemwide view of the MCAM rules installed by all the PF's. Below is the example output when the debugfs file is read: ~ # mount -t debugfs none /sys/kernel/debug ~ # cat /sys/kernel/debug/octeontx2/npc/mcam_rules Installed by: PF1 direction: RX mcam entry: 227 udp source port 23 mask 0xffff Forward to: PF1 VF0 action: Direct to queue 0 enabled: yes counter: 1 hits: 0 Signed-off-by: -
Hariprasad Kelam authored
Add unicast MAC address filtering support using install flow message. Total of 8 MCAM entries are allocated for adding unicast mac filtering rules. If the MCAM allocation fails, the unicast filtering support will not be advertised. Signed-off-by:
-
Subbaraya Sundeep authored
This patch adds support for adding and deleting ethtool ntuple filters. The filters for ether, ipv4, ipv6, tcp, udp and sctp are supported. The mask is also supported. The supported actions are drop and direct to a queue. Additionally we support FLOW_EXT field vlan_tci and FLOW_MAC_EXT. The NIX PF will allocate total 32 MCAM entries for the use of ethtool ntuple filters. The Administrative Function(AF) will install/delete the MCAM rules when NIX PF sends mailbox message to install/delete the ntuple filters. Ethtool ntuple filters support is restricted to PFs as of now and PF can install ntuple filters to direct the traffic to its VFs. Hence added a separate callback for VFs to get/set RSS configuration. Signed-off-by:
-
Subbaraya Sundeep authored
Added new mailbox messages to install and delete MCAM rules. These mailbox messages will be used for adding/deleting ethtool n-tuple filters by NIX PF. The installed MCAM rules are stored in a list that will be traversed later to delete the MCAM entries when the interface is brought down or when PCIe FLR is received. The delete mailbox supports deleting a single MCAM entry or range of entries or all the MCAM entries owned by the pcifunc. Each MCAM entry can be associated with a HW match stat entry if the mailbox requester wants to check the hit count for debugging. Modified adding default unicast DMAC match rule using install flow API. The default unicast DMAC match entry installed by Administrative Function is saved and can be changed later by the mailbox user to fit additional fields, or the default MCAM entry rule action can be used for other flow rules installed later. Modified rvu_mbox_handler_nix_lf_free mailbox to add a flag to disable or delete the MCAM entries. The MCAM entries are disabled when the interface is brought down and deleted in FLR handler. The disabled MCAM entries will be re-enabled when the interface is brought up again. Signed-off-by:
-
Subbaraya Sundeep authored
Key Extraction(KEX) profile decides how the packet metadata such as layer information and selected packet data bytes at each layer are placed in MCAM search key. This patch reads the configured KEX profile parameters to find out the bit position and bit mask for each field. The information is used when programming the MCAM match data by SW to match a packet flow and take appropriate action on the flow. This patch also verifies the mandatory fields such as channel and DMAC are not overwritten by the KEX configuration of other fields. Signed-off-by:
-
Subbaraya Sundeep authored
This patch adds support to verify the channel number sent by mailbox requester before writing MCAM entry for Ingress packets. Similarly for Egress packets, verifying the PF_FUNC sent by the mailbox user. Signed-off-by:
Kiran Kumar K <kirankumark@marvell.com> Signed-off-by:
-
Stanislaw Kardach authored
The current default Key Extraction(KEX) profile can only use RX packet fields while generating the MCAM search key. The profile can't be used for matching TX packet fields. This patch modifies the default KEX profile to add support for extracting TX packet fields into MCAM search key. Enabled Tx KPU packet parsing by configuring TX PKIND in tx_parse_cfg. Modified the KEX profile to extract 2 bytes of VLAN TCI from an offset of 2 bytes from LB_PTR. The LB_PTR points to the byte offset where the VLAN header starts. The NPC KPU parser profile has been modified to point LB_PTR to the starting byte offset of VLAN header which points to the tpid field. Signed-off-by:
Stanislaw Kardach <skardach@marvell.com> Signed-off-by:
-
Xie He authored
The DLCI driver (dlci.c) implements the Frame Relay protocol. However, we already have another newer and better implementation of Frame Relay provided by the HDLC_FR driver (hdlc_fr.c). The DLCI driver's implementation of Frame Relay is used by only one hardware driver in the kernel - the SDLA driver (sdla.c). The SDLA driver provides Frame Relay support for the Sangoma S50x devices. However, the vendor provides their own driver (along with their own multi-WAN-protocol implementations including Frame Relay), called WANPIPE. I believe most users of the hardware would use the vendor-provided WANPIPE driver instead. (The WANPIPE driver was even once in the kernel, but was deleted in commit 8db60bcf ("[WAN]: Remove broken and unmaintained Sangoma drivers.") because the vendor no longer updated the in-kernel WANPIPE driver.) Cc: Mike McLagan <mike.mclagan@linux.org> Signed-off-by:
Xie He <xie.he.0141@gmail.com> Link: https://lore.kernel.org/r/20201114150921.685594-1-xie.he.0141@gmail.comSigned-off-by:
-
Jakub Kicinski authored
Huazhong Tan says: ==================== net: hns3: updates for -next There are several updates relating to the interrupt coalesce for the HNS3 ethernet driver. based on the frame quantity). a fixed value in code. based on the gap time). its new usage. change log: V4 - remove #5~#10 from this series, which needs more discussion. V3 - fix a typo error in #1 reported by Jakub Kicinski. rewrite #9 commit log. remove #11 from this series. V2 - reorder #2 & #3 to fix compiler error. fix some checkpatch warnings in #10 & #11. previous version: V3: https://patchwork.ozlabs.org/project/netdev/cover/1605151998-12633-1-git-send-email-tanhuazhong@huawei.com/ V2: https://patchwork.ozlabs.org/project/netdev/cover/1604892159-19990-1-git-send-email-tanhuazhong@huawei.com/ V1: https://patchwork.ozlabs.org/project/netdev/cover/1604730681-32559-1-git-send-email-tanhuazhong@huawei.com/ ==================== Link: https://lore.kernel.org/r/1605514854-11205-1-git-send-email-tanhuazhong@huawei.comSigned-off-by: -
Huazhong Tan authored
Besides GL(Gap Limiting), QL(Quantity Limiting) can be modified dynamically when DIM is supported. So rename gl_adapt_enable as adapt_enable in struct hns3_enet_coalesce. Signed-off-by:
Huazhong Tan <tanhuazhong@huawei.com> Signed-off-by:
-
Huazhong Tan authored
For device whose version is above V3(include V3), the GL configuration can set as 1us unit, so adds support for configuring this field. Signed-off-by:
-
Huazhong Tan authored
For maintainability and compatibility, add support for querying the maximum value of GL. Signed-off-by:
-
Huazhong Tan authored
QL(quantity limiting) means that hardware supports the interrupt coalesce based on the frame quantity. QL can be configured when int_ql_max in device's specification is non-zero, so add support to configure it. Also, rename two coalesce init function to fit their purpose. Signed-off-by:
-
Jakub Kicinski authored
Ioana Ciornei says: ==================== net: phy: add support for shared interrupts (part 2) This patch set aims to actually add support for shared interrupts in phylib and not only for multi-PHY devices. While we are at it, streamline the interrupt handling in phylib. For a bit of context, at the moment, there are multiple phy_driver ops that deal with this subject: - .config_intr() - Enable/disable the interrupt line. - .ack_interrupt() - Should quiesce any interrupts that may have been fired. It's also used by phylib in conjunction with .config_intr() to clear any pending interrupts after the line was disabled, and before it is going to be enabled. - .did_interrupt() - Intended for multi-PHY devices with a shared IRQ line and used by phylib to discern which PHY from the package was the one that actually fired the interrupt. - .handle_interrupt() - Completely overrides the default interrupt handling logic from phylib. The PHY driver is responsible for checking if any interrupt was fired by the respective PHY and choose accordingly if it's the one that should trigger the link state machine. From my point of view, the interrupt handling in phylib has become somewhat confusing with all these callbacks that actually read the same PHY register - the interrupt status. A more streamlined approach would be to just move the responsibility to write an interrupt handler to the driver (as any other device driver does) and make .handle_interrupt() the only way to deal with interrupts. Another advantage with this approach would be that phylib would gain support for shared IRQs between different PHY (not just multi-PHY devices), something which at the moment would require extending every PHY driver anyway in order to implement their .did_interrupt() callback and duplicate the same logic as in .ack_interrupt(). The disadvantage of making .did_interrupt() mandatory would be that we are slightly changing the semantics of the phylib API and that would increase confusion instead of reducing it. What I am proposing is the following: - As a first step, make the .ack_interrupt() callback optional so that we do not break any PHY driver amid the transition. - Every PHY driver gains a .handle_interrupt() implementation that, for the most part, would look like below: irq_status = phy_read(phydev, INTR_STATUS); if (irq_status < 0) { phy_error(phydev); return IRQ_NONE; } if (!(irq_status & irq_mask)) return IRQ_NONE; phy_trigger_machine(phydev); return IRQ_HANDLED; - Remove each PHY driver's implementation of the .ack_interrupt() by actually taking care of quiescing any pending interrupts before enabling/after disabling the interrupt line. - Finally, after all drivers have been ported, remove the .ack_interrupt() and .did_interrupt() callbacks from phy_driver. This patch set is part 2 of the entire change set and it addresses the changes needed in 9 PHY drivers. The rest can be found on my Github branch here: https://github.com/IoanaCiornei/linux/commits/phylib-shared-irq ==================== Link: https://lore.kernel.org/r/20201113165226.561153-1-ciorneiioana@gmail.comSigned-off-by: -
Ioana Ciornei authored
In preparation of removing the .ack_interrupt() callback, we must replace its occurrences (aka phy_clear_interrupt), from the 2 places where it is called from (phy_enable_interrupts and phy_disable_interrupts), with equivalent functionality. This means that clearing interrupts now becomes something that the PHY driver is responsible of doing, before enabling interrupts and after clearing them. Make this driver follow the new contract. Signed-off-by:
Ioana Ciornei <ioana.ciornei@nxp.com> Acked-by:
Alexandru Ardelean <alexandru.ardelean@analog.com> Signed-off-by:
-
Ioana Ciornei authored
In an attempt to actually support shared IRQs in phylib, we now move the responsibility of triggering the phylib state machine or just returning IRQ_NONE, based on the IRQ status register, to the PHY driver. Having 3 different IRQ handling callbacks (.handle_interrupt(), .did_interrupt() and .ack_interrupt() ) is confusing so let the PHY driver implement directly an IRQ handler like any other device driver. Make this driver follow the new convention. Signed-off-by:
-
Ioana Ciornei authored
In preparation of removing the .ack_interrupt() callback, we must replace its occurrences (aka phy_clear_interrupt), from the 2 places where it is called from (phy_enable_interrupts and phy_disable_interrupts), with equivalent functionality. This means that clearing interrupts now becomes something that the PHY driver is responsible of doing, before enabling interrupts and after clearing them. Make this driver follow the new contract. Signed-off-by:
-
