- 01 Mar, 2018 9 commits
-
-
Yuval Mintz authored
Since commit 8fb472c0 ("ipmr: improve hash scalability") ipmr has been using rhashtable as a basis for its mfc routes, but ip6mr is currently still using the old private MFC hash implementation. Align ip6mr to the current ipmr implementation. Signed-off-by:
Yuval Mintz <yuvalm@mellanox.com> Acked-by:
Nikolay Aleksandrov <nikolay@cumulusnetworks.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Yuval Mintz authored
In ipmr the mr_table socket is handled under RCU. Introduce the same for ip6mr. Signed-off-by:
-
Yuval Mintz authored
The two implementations have almost identical structures - vif_device and mif_device. As a step toward uniforming the mr_tables, eliminate the mif_device and relocate the vif_device definition into a new common header file. Also, introduce a common initializing function for setting most of the vif_device fields in a new common source file. This requires modifying the ipv{4,6] Kconfig and ipv4 makefile as we're introducing a new common config option - CONFIG_IP_MROUTE_COMMON. Signed-off-by: -
David S. Miller authored
Roopa Prabhu says: ==================== fib_rules: support sport, dport and proto match This series extends fib rule match support to include sport, dport and ip proto match (to complete the 5-tuple match support). Common use-cases of Policy based routing in the data center require 5-tuple match. The last 2 patches in the series add a call to flow dissect in the fwd path if required by the installed fib rules (controlled by a flag). v1: - Fix errors reported by kbuild and feedback on RFC - extend port match uapi to accomodate port ranges v2: - address comments from Nikolay, David Ahern and Paolo (Thanks!) Pending things I will submit separate patches for: - extack for fib rules - fib rules test (as requested by david ahern) ==================== Signed-off-by: -
Roopa Prabhu authored
Dissect flow in fwd path if fib rules require it. Controlled by a flag to avoid penatly for the common case. Flag is set when fib rules with sport, dport and proto match that require flow dissect are installed. Also passes the dissected hash keys to the multipath hash function when applicable to avoid dissecting the flow again. icmp packets will continue to use inner header for hash calculations. Signed-off-by:
Roopa Prabhu <roopa@cumulusnetworks.com> Acked-by:
Paolo Abeni <pabeni@redhat.com> Acked-by:
-
Roopa Prabhu authored
Dissect flow in fwd path if fib rules require it. Controlled by a flag to avoid penatly for the common case. Flag is set when fib rules with sport, dport and proto match that require flow dissect are installed. Also passes the dissected hash keys to the multipath hash function when applicable to avoid dissecting the flow again. icmp packets will continue to use inner header for hash calculations (Thanks to Nikolay Aleksandrov for some review here). Signed-off-by:
-
Roopa Prabhu authored
support to match on src port, dst port and ip protocol. Signed-off-by:
-
Roopa Prabhu authored
support to match on src port, dst port and ip protocol. Signed-off-by:
-
Roopa Prabhu authored
uapi for ip_proto, sport and dport range match in fib rules. Signed-off-by:
-
- 28 Feb, 2018 31 commits
-
-
Heiner Kallweit authored
In case of MSI-X the interrupt number may differ from pcidev->irq. Fix this by using pci_irq_vector(). Fixes: 6c6aa15f ("r8169: improve interrupt handling") Signed-off-by:
Heiner Kallweit <hkallweit1@gmail.com> Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queueDavid S. Miller authored
Jeff Kirsher says: ==================== 100GbE Intel Wired LAN Driver Updates 2018-02-28 This series contains updates to fm10k only. Jake provides all the changes in this series, starting with making the function header comments consistent and to align with how the kernel documentation expects it. Also cleaned up code comment as well as bump the driver version. ==================== Signed-off-by: -
David S. Miller authored
Ido Schimmel says: ==================== selftests: forwarding: Add VRF-based tests One of the nice things about network namespaces is that they allow one to easily create and test complex environments. Unfortunately, these namespaces can not be used with actual switching ASICs, as their ports can not be migrated to other network namespaces (NETIF_F_NETNS_LOCAL) and most of them probably do not support the L1-separation provided by namespaces. However, a similar kind of flexibility can be achieved by using VRFs and by looping the switch ports together. For example: br0 + vrf-h1 | vrf-h2 + +---+----+ + | | | | 192.0.2.1/24 + + + + 192.0.2.2/24 swp1 swp2 swp3 swp4 + + + + | | | | +--------+ +--------+ The VRFs act as lightweight namespaces representing hosts connected to the switch. This approach for testing switch ASICs has several advantages over the traditional method that requires multiple physical machines, to name a few: 1. Only the device under test (DUT) is being tested without noise from other system. 2. Ability to easily provision complex topologies. Testing bridging between 4-ports LAGs or 8-way ECMP requires many physical links that are not always available. With the VRF-based approach one merely needs to loopback more ports. These tests are written with switch ASICs in mind, but they can be run on any Linux box using veth pairs to emulate physical loopbacks. v2: * Order local variables declaration according to function arguments order (Petr) v1: * Change location to net/forwarding instead of forwarding/ * Add ability to pause on failure * Add ability to pause on cleanup * Make configuration file optional * Make ping/ping6/mz configurable * Add more tc tests ==================== Signed-off-by: -
Jiri Pirko authored
Test shared block infrastructure. This is a basic test that shares TC block in between 2 clsact qdiscs. Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Signed-off-by:
Ido Schimmel <idosch@mellanox.com> Signed-off-by:
-
Jiri Pirko authored
Tests chains matching and goto chain action. Signed-off-by:
-
Jiri Pirko authored
Add first part of actions tests. This patch only contains tests of gact ok/drop/trap and mirred redirect egress. Signed-off-by:
-
Jiri Pirko authored
Add first part of flower tests. This patch only contains dst/src ip/mac matching. Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Jiri Pirko authored
Signed-off-by:
-
Ido Schimmel authored
Have one host generate 16K IPv6 echo requests with a random flow label and check that they are distributed between both multipath links according to the provided weights. Signed-off-by:
-
Ido Schimmel authored
Use different weights for the multipath route configured on the first router and check that the different flows generated by the first host are distributed according to the provided weights. Signed-off-by:
-
Ido Schimmel authored
Create a topology with two hosts, each directly connected to a different router. Both routers are connected using two links, enabling multipath routing. Test IPv4 and IPv6 ping using default MTU and large MTU. Signed-off-by:
-
Ido Schimmel authored
Configure two hosts which are directly connected to the same router and test IPv4 and IPv6 ping. Use a large MTU and check that ping is unaffected. Signed-off-by:
-
Ido Schimmel authored
Add test cases for unknown unicast and unregistered multicast flooding. For each traffic type, turn off flooding on one bridged port and inject a packet of the specified type through the second bridged port. Make sure the packet was not received by checking the ACL counters on the other end. Later, turn on flooding and make sure the packet was received. Signed-off-by:
-
Ido Schimmel authored
Send a packet with a specific destination MAC, make sure it was learned on the ingress port and then aged-out. Signed-off-by:
-
Ido Schimmel authored
Add initial framework to test packet forwarding functionality. The tests can run on actual devices using loop-backed cables or using veth pairs. Signed-off-by:
-
Paolo Abeni authored
This changeset moves ipvlan address under RCU protection, using a per ipvlan device spinlock to protect list mutation and RCU read access to protect list traversal. Also explicitly use RCU read lock to traverse the per port ipvlans list, so that we can now perform a full address lookup without asserting the RTNL lock. Overall this allows the ipvlan driver to check fully for duplicate addresses - before this commit ipv6 addresses assigned by autoconf via prefix delegation where accepted without any check - and avoid the following rntl assertion failure still in the same code path: RTNL: assertion failed at drivers/net/ipvlan/ipvlan_core.c (124) WARNING: CPU: 15 PID: 0 at drivers/net/ipvlan/ipvlan_core.c:124 ipvlan_addr_busy+0x97/0xa0 [ipvlan] Modules linked in: ipvlan(E) ixgbe CPU: 15 PID: 0 Comm: swapper/15 Tainted: G E 4.16.0-rc2.ipvlan+ #1782 Hardware name: Dell Inc. PowerEdge R730/072T6D, BIOS 2.1.7 06/16/2016 RIP: 0010:ipvlan_addr_busy+0x97/0xa0 [ipvlan] RSP: 0018:ffff881ff9e03768 EFLAGS: 00010286 RAX: 0000000000000000 RBX: ffff881fdf2a9000 RCX: 0000000000000000 RDX: 0000000000000001 RSI: 00000000000000f6 RDI: 0000000000000300 RBP: ffff881fdf2a8000 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000001 R11: ffff881ff9e034c0 R12: ffff881fe07bcc00 R13: 0000000000000001 R14: ffffffffa02002b0 R15: 0000000000000001 FS: 0000000000000000(0000) GS:ffff881ff9e00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fc5c1a4f248 CR3: 000000207e012005 CR4: 00000000001606e0 Call Trace: <IRQ> ipvlan_addr6_event+0x6c/0xd0 [ipvlan] notifier_call_chain+0x49/0x90 atomic_notifier_call_chain+0x6a/0x100 ipv6_add_addr+0x5f9/0x720 addrconf_prefix_rcv_add_addr+0x244/0x3c0 addrconf_prefix_rcv+0x2f3/0x790 ndisc_router_discovery+0x633/0xb70 ndisc_rcv+0x155/0x180 icmpv6_rcv+0x4ac/0x5f0 ip6_input_finish+0x138/0x6a0 ip6_input+0x41/0x1f0 ipv6_rcv+0x4db/0x8d0 __netif_receive_skb_core+0x3d5/0xe40 netif_receive_skb_internal+0x89/0x370 napi_gro_receive+0x14f/0x1e0 ixgbe_clean_rx_irq+0x4ce/0x1020 [ixgbe] ixgbe_poll+0x31a/0x7a0 [ixgbe] net_rx_action+0x296/0x4f0 __do_softirq+0xcf/0x4f5 irq_exit+0xf5/0x110 do_IRQ+0x62/0x110 common_interrupt+0x91/0x91 </IRQ> v1 -> v2: drop unneeded in_softirq check in ipvlan_addr6_validator_event() Fixes: e9997c29 ("ipvlan: fix check for IP addresses in control path") Signed-off-by:
-
Paolo Abeni authored
Currently, if IPv6 is enabled on top of an ipvlan device in l3 mode, the following warning message: Dropped {multi|broad}cast of type= [86dd] is emitted every time that a RS is generated and dmseg is soon filled with irrelevant messages. Replace pr_warn with pr_debug, to preserve debuggability, without scaring the sysadmin. Signed-off-by: -
David S. Miller authored
Jiri Pirko says: ==================== mlxsw: Offload multi-queue RED support Nogah says: Support a two level hierarchy of offloaded qdiscs in mlxsw, with sch_prio being the root qdisc and sch_red as the children. +----------+ | sch_prio | +----+-----+ | | +----------------------------------+ | | | | | | | | | +---v---+ +----v---+ +-----v--+ |sch_red| |sch_red | |sch_red | +-------+ +--------+ +--------+ When setting sch_prio as the root qdisc on a physical port, mlxsw will offload it. When adding it with sch_red as a child qdisc, it will offload it as well. Relocating child qdisc or connecting them to more then one child will result in unoffloading them. Relocating child qdisc more then once is highly unrecommended and might cause a miss match between the kernel configuration and the offloaded one. The offloaded configuration will be aligned with the one shown in the show command. Changing the priomap parameter of sch_prio might cause a band that its configuration was changed and it has offloaded sch_red set on it, to lose some stats data as if sch_red was unoffloaded and offloaded again. However, it won't affect the data on this band that will have sch_red continuously. Patch 1 adds support for setting RED as the child of root qdisc. Patches 2-4 add support for RED bstasts for offloaded child qdiscs. Patches 5-6 handle backlog related changes for offloaded child qdiscs. Patches 7-8 update PRIO in mlxsw to be able to have RED as child on its bands. Patch 9 adds offload handles for PRIO graft operations. In mlxsw it will cause the driver to stop offloading the child in question. ==================== Signed-off-by: -
Nogah Frankel authored
Handle graft command for an offloaded sch_prio. Grafting a qdisc to any place other than under its original parent is not supported by mlxsw and will cause the grafted qdisc to stop being offloaded. Signed-off-by:
Nogah Frankel <nogahf@mellanox.com> Reviewed-by:
-
Nogah Frankel authored
Offload sch_prio graft command for capable drivers. Warn in case of a failure, unless the graft was done as part of a destroy operation (the new qdisc is a noop) or if all the qdiscs (the parent, the old child, and the new one) are not offloaded. Signed-off-by:
-
Nogah Frankel authored
When the number the bands of sch_prio is decreased, child qdiscs on the deleted bands would get deleted as well. This change and deletions are being done under sch_tree_lock of the sch_prio qdisc. Part of the destruction of qdisc is unoffloading it, if it is offloaded. Un-offloading can't be done inside this lock. Move the offload command to be done before reducing the number of bands, so unoffloading of the qdiscs that are about to be deleted could be done outside of the lock. Signed-off-by:
-
Nogah Frankel authored
sch_prio as root qdisc should count all the drops its children have. Since it is possible for it to have sch_red children, it needs to count RED early drops. Signed-off-by:
-
Nogah Frankel authored
Offloaded qdiscs are allowed to expose only parts of their statistics. It means that if backlog is being exposed and qlen is not, it might trigger a warning in qdisc_tree_reduce_backlog. Do not warn in case the qdisc that was removed was an offloaded one. Signed-off-by:
-
Nogah Frankel authored
When removing a child qdisc its backlog will be decreased from the parent backlog. The driver backlog count should do the same. When the parent changes its configuration, the child might need to clean its stats. However, the backlog can't be cleaned with the rest of the stats, because it reflects a momentary value that needs to be synced with the core, not the history of the qdisc. Signed-off-by:
-
Nogah Frankel authored
Priority counters count packets according to their packet priority. Collect the stats for sch_red based on these counters, so the qdisc bstats will be the sum of counters matching the priorities marked in the qdisc priomap. Changing the mapping of the priorities to bands while traffic is running can result in losing the stats of the bands qdiscs from their last dump call to this change, as if the qdisc was unoffloaded and re-offloaded. It will not affect the traffic behaviour according to sch_red. Signed-off-by:
-
Nogah Frankel authored
Add priority map per qdisc, to indicate which priorities are being directed through this qdisc. Signed-off-by:
-
Nogah Frankel authored
Add TX packets and bytes counters per switch priority per port. Signed-off-by:
-
Nogah Frankel authored
Add the option to set a qdisc per tclass. Match the qdisc to the tclass by parent ID. Supported currently for sch_red only. It allows offloading sch_prio as root qdisc and sch_red as its child. (However, doing so might corrupt the stats for both parent and child.) Signed-off-by:
-
Maxime Chevallier authored
Marvell PPv2 controller allows for generic packet filtering. This commit adds entries to implement VLAN filtering. The approach taken is : - Filter entries that would match on the presence of the VLAN tag (existing VLAN detection, DSA / EDSA detection) will set the next lookup ID to be for the VID. - For each VLAN existing on a given port, we add an entry that matches this specific VID. If the incoming packet matches the VID entry, it is set for the next lookup in the chain (LU_L2). - A Guard entry is added for each port, that will match if the incoming packet didn't match any of the above VID entries. This entry tags the packet to be dropped. Due to this design, and the fact that the total 256 filter entries are also used for other purposes, we have a limit of 10 VLANs per port. To accommodate the case where we would need more VLANS on one port, this patch implements the ndo_set_features to allow for disabling of VLAN filtering using ethtool. The default config has VLAN filtering disabled. Signed-off-by:
Maxime Chevallier <maxime.chevallier@bootlin.com> Signed-off-by:
-
