- 23 Apr, 2020 25 commits
-
-
Sudarsana Reddy Kalluru authored
The patch enables the device to send error messages to root port when an error is detected. Signed-off-by:
Sudarsana Reddy Kalluru <skalluru@marvell.com> Signed-off-by:
Ariel Elior <aelior@marvell.com> Signed-off-by:
Igor Russkikh <irusskikh@marvell.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Alexander Lobakin authored
gro_cells lib is used by different encapsulating netdevices, such as geneve, macsec, vxlan etc. to speed up decapsulated traffic processing. CPU tag is a sort of "encapsulation", and we can use the same mechs to greatly improve overall DSA performance. skbs are passed to the GRO layer after removing CPU tags, so we don't need any new packet offload types as it was firstly proposed by me in the first GRO-over-DSA variant [1]. The size of struct gro_cells is sizeof(void *), so hot struct dsa_slave_priv becomes only 4/8 bytes bigger, and all critical fields remain in one 32-byte cacheline. The other positive side effect is that drivers for network devices that can be shipped as CPU ports of DSA-driven switches can now use napi_gro_frags() to pass skbs to kernel. Packets built that way are completely non-linear and are likely being dropped without GRO. This was tested on to-be-mainlined-soon Ethernet driver that uses napi_gro_frags(), and the overall performance was on par with the variant from [1], sometimes even better due to minimal overhead. net.core.gro_normal_batch tuning may help to push it to the limit on particular setups and platforms. iperf3 IPoE VLAN NAT TCP forwarding (port1.218 -> port0) setup on 1.2 GHz MIPS board: 5.7-rc2 baseline: [ID] Interval Transfer Bitrate Retr [ 5] 0.00-120.01 sec 9.00 GBytes 644 Mbits/sec 413 sender [ 5] 0.00-120.00 sec 8.99 GBytes 644 Mbits/sec receiver Iface RX packets TX packets eth0 7097731 7097702 port0 426050 6671829 port1 6671681 425862 port1.218 6671677 425851 With this patch: [ID] Interval Transfer Bitrate Retr [ 5] 0.00-120.01 sec 12.2 GBytes 870 Mbits/sec 122 sender [ 5] 0.00-120.00 sec 12.2 GBytes 870 Mbits/sec receiver Iface RX packets TX packets eth0 9474792 9474777 port0 455200 353288 port1 9019592 455035 port1.218 353144 455024 v2: - Add some performance examples in the commit message; - No functional changes. [1] https://lore.kernel.org/netdev/20191230143028.27313-1-alobakin@dlink.ru/Signed-off-by:
Alexander Lobakin <bloodyreaper@yandex.ru> Signed-off-by:
-
Fernando Gont authored
RFC4862 5.5.3 e) prevents received Router Advertisements from reducing the Valid Lifetime of configured addresses to less than two hours, thus preventing hosts from reacting to the information provided by a router that has positive knowledge that a prefix has become invalid. This patch makes hosts honor all Valid Lifetime values, as per draft-gont-6man-slaac-renum-06, Section 4.2. This is meant to help mitigate the problem discussed in draft-ietf-v6ops-slaac-renum. Note: Attacks aiming at disabling an advertised prefix via a Valid Lifetime of 0 are not really more harmful than other attacks that can be performed via forged RA messages, such as those aiming at completely disabling a next-hop router via an RA that advertises a Router Lifetime of 0, or performing a Denial of Service (DoS) attack by advertising illegitimate prefixes via forged PIOs. In scenarios where RA-based attacks are of concern, proper mitigations such as RA-Guard [RFC6105] [RFC7113] should be implemented. Signed-off-by:
Fernando Gont <fgont@si6networks.com> Signed-off-by:
-
David S. Miller authored
Ioana Ciornei says: ==================== dpaa2-eth: add support for xdp bulk enqueue The first patch moves the DEV_MAP_BULK_SIZE macro into the xdp.h header file so that drivers can take advantage of it and use it. The following 3 patches are there to setup the scene for using the bulk enqueue feature. First of all, the prototype of the enqueue function is changed so that it returns the number of enqueued frames. Second, the bulk enqueue interface is used but without any functional changes, still one frame at a time is enqueued. Third, the .ndo_xdp_xmit callback is split into two stages, create all FDs for the xdp_frames received and then enqueue them. The last patch of the series builds on top of the others and instead of issuing an enqueue operation for each FD it issues a bulk enqueue call for as many frames as possible. This is repeated until all frames are enqueued or the maximum number of retries is hit. We do not use the XDP_XMIT_FLUSH flag since the architecture is not capable to store all frames dequeued in a NAPI cycle, instead we send out right away all frames received in a .ndo_xdp_xmit call. Changes in v2: - statically allocate an array of dpaa2_fd by frame queue - use the DEV_MAP_BULK_SIZE as the maximum number of xdp_frames received in .ndo_xdp_xmit() ==================== Signed-off-by: -
Ioana Ciornei authored
Take advantage of the bulk enqueue feature in .ndo_xdp_xmit. We cannot use the XDP_XMIT_FLUSH since the architecture is not capable to store all the frames dequeued in a NAPI cycle so we instead are enqueueing all the frames received in a ndo_xdp_xmit call right away. After setting up all FDs for the xdp_frames received, enqueue multiple frames at a time until all are sent or the maximum number of retries is hit. Signed-off-by:
Ioana Ciornei <ioana.ciornei@nxp.com> Acked-by:
Jesper Dangaard Brouer <brouer@redhat.com> Signed-off-by:
-
Ioana Ciornei authored
Instead of having a function that both creates a frame descriptor from an xdp_frame and enqueues it, split this into two stages. Add the dpaa2_eth_xdp_create_fd that just transforms an xdp_frame into a FD while the actual enqueue callback is called directly from the ndo for each frame. This is particulary useful in conjunction with bulk enqueue. Signed-off-by:
-
Ioana Ciornei authored
Update the dpaa2-eth driver to use the bulk enqueue function introduced with the change to QBMAN ring mode. At the moment, no functional changes are made but rather the driver just transitions to the new interface while still enqueuing just one frame at a time. Signed-off-by:
-
Ioana Ciornei authored
The enqueue dpaa2-eth callback now returns the number of successfully enqueued frames. This is a preliminary patch necessary for adding support for bulk ring mode enqueue. Signed-off-by:
-
Ioana Ciornei authored
Export the DEV_MAP_BULK_SIZE macro to the header file so that drivers can directly use it as the maximum number of xdp_frames received in the .ndo_xdp_xmit() callback. Signed-off-by:
-
David Ahern authored
Add nodad when adding IPv6 addresses and remove the sleep. A recent change to iproute2 moved the 'pref medium' to the prefix (where it belongs). Change the expected route check to strip 'pref medium' to be compatible with old and new iproute2. Add IPv4 runtime test with an IPv6 address as the gateway in the default route. Signed-off-by:
David Ahern <dsahern@gmail.com> Signed-off-by:
-
David S. Miller authored
Petr Machata says: ==================== Add selftests for pedit ex munge ip6 dsfield Patch #1 extends the existing generic forwarding selftests to cover pedit ex munge ip6 traffic_class as well. Patch #2 adds TDC test coverage. ==================== Signed-off-by: -
Petr Machata authored
Add a self-test for the IPv6 dsfield munge that iproute2 will support. Signed-off-by:
Petr Machata <petrm@mellanox.com> Signed-off-by:
-
Petr Machata authored
Extend the pedit_dsfield forwarding selftest with coverage of "pedit ex munge ip6 dsfield set". Signed-off-by:
-
David S. Miller authored
Oleksij Rempel says: ==================== add TJA1102 support changes v5: - rename __of_mdiobus_register_phy() to of_mdiobus_phy_device_register() changes v4: - remove unused phy_id variable changes v3: - export part of of_mdiobus_register_phy() and reuse it in tja11xx driver - coding style fixes changes v2: - use .match_phy_device - add irq support - add add delayed registration for PHY1 ==================== Signed-off-by: -
Oleksij Rempel authored
TJA1102 is a dual PHY package with PHY0 having proper PHYID and PHY1 having no ID. On one hand it is possible to for PHY detection by compatible, on other hand we should be able to reset complete chip before PHY1 configured it, and we need to define dependencies for proper power management. We can solve it by defining PHY1 as child of PHY0: tja1102_phy0: ethernet-phy@4 { reg = <0x4>; interrupts-extended = <&gpio5 8 IRQ_TYPE_LEVEL_LOW>; reset-gpios = <&gpio5 9 GPIO_ACTIVE_LOW>; reset-assert-us = <20>; reset-deassert-us = <2000>; tja1102_phy1: ethernet-phy@5 { reg = <0x5>; interrupts-extended = <&gpio5 8 IRQ_TYPE_LEVEL_LOW>; }; }; The PHY1 should be a subnode of PHY0 and registered only after PHY0 was completely reset and initialized. Signed-off-by:Oleksij Rempel <o.rempel@pengutronix.de> Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Oleksij Rempel authored
This function will be needed in tja11xx driver for secondary PHY support. Signed-off-by:
-
Oleksij Rempel authored
TJA1102 is an dual T1 PHY chip. Both PHYs are separately addressable. Both PHYs are similar but have different amount of functionality. For example PHY 1 has no PHY ID and no health monitor. Signed-off-by:
-
Oleksij Rempel authored
Document the NXP TJA11xx PHY bindings. Signed-off-by:
-
Tang Bin authored
Use IS_ERR() and PTR_ERR() instead of PTR_ZRR_OR_ZERO() to simplify code, avoid redundant paramenter definitions and judgements. Signed-off-by:
Zhang Shengju <zhangshengju@cmss.chinamobile.com> Signed-off-by:
Tang Bin <tangbin@cmss.chinamobile.com> Signed-off-by:
-
Oleksij Rempel authored
Add support for following phy-modes: rgmii, rgmii-id, rgmii-txid, rgmii-rxid. This PHY has an internal RX delay of 1.2ns and no delay for TX. The pad skew registers allow to set the total TX delay to max 1.38ns and the total RX delay to max of 2.58ns (configurable 1.38ns + build in 1.2ns) and a minimal delay of 0ns. According to the RGMII v1.3 specification the delay provided by PCB traces should be between 1.5ns and 2.0ns. The RGMII v2.0 allows to provide this delay by MAC or PHY. So, we configure this PHY to the best values we can get by this HW: TX delay to 1.38ns (max supported value) and RX delay to 1.80ns (best calculated delay) The phy-modes can still be fine tuned/overwritten by *-skew-ps device tree properties described in: Documentation/devicetree/bindings/net/micrel-ksz90x1.txt Signed-off-by:
Philippe Schenker <philippe.schenker@toradex.com> Reviewed-by:
-
Jason Yan authored
Fix the following coccicheck warning: net/caif/caif_dev.c:410:2-13: WARNING: Assignment of 0/1 to bool variable net/caif/caif_dev.c:445:2-13: WARNING: Assignment of 0/1 to bool variable net/caif/caif_dev.c:145:1-12: WARNING: Assignment of 0/1 to bool variable net/caif/caif_dev.c:223:1-12: WARNING: Assignment of 0/1 to bool variable Signed-off-by:
Jason Yan <yanaijie@huawei.com> Signed-off-by:
-
Chuah, Kim Tatt authored
For dwmac4, enable VLAN promiscuity when MAC controller is requested to enter promiscuous mode. Signed-off-by:
Chuah, Kim Tatt <kim.tatt.chuah@intel.com> Signed-off-by:
Ong Boon Leong <boon.leong.ong@intel.com> Signed-off-by:
Tan, Tee Min <tee.min.tan@intel.com> Signed-off-by:
Wong Vee Khee <vee.khee.wong@intel.com> Signed-off-by:
-
Wei Yongjun authored
macvlan_hash_lookup() uses list_for_each_entry_rcu() for traversing should either under RCU in fast path or the protection of rtnl_mutex. In the case of holding RTNL, we should add the corresponding lockdep expression to silence the following false-positive warning: ============================= WARNING: suspicious RCU usage 5.7.0-rc1-next-20200416-00003-ga3b8d28bc #1 Not tainted ----------------------------- drivers/net/macvlan.c:126 RCU-list traversed in non-reader section!! Signed-off-by:
Wei Yongjun <weiyongjun1@huawei.com> Signed-off-by:
-
David Ahern authored
Add tests for vrf and xfrms with a second round after adding a qdisc. There are a few known problems documented with the test cases that fail. The fix is non-trivial; will come back to it when time allows. Signed-off-by:
-
Xiaoliang Yang authored
Right now it can be seen that the VSC9959 (Felix) switch will not flood frames if they have a VLAN tag with a PCP of 1-7 (nonzero). It turns out that Felix is quite different from its cousin, Ocelot, in that frame flooding can be allowed/denied per traffic class. Where Ocelot has 1 instance of the ANA_FLOODING register, Felix has 8. The approach that this driver is going to take is "thanks, but no thanks". We have no use case of limiting the flooding domain based on traffic class, so we just want to allow packets to be flooded, no matter what traffic class they have. So we copy the line of code from ocelot.c which does the one-shot initialization of the flooding PGIDs, and we add it to felix.c as well - except replicated 8 times. Signed-off-by:
Xiaoliang Yang <xiaoliang.yang_1@nxp.com> Signed-off-by:
Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by:
-
- 22 Apr, 2020 15 commits
-
-
Manivannan Sadhasivam authored
Add tracepoint support for QRTR with NS as the first candidate. Later on this can be extended to core QRTR and transport drivers. The trace_printk() used in NS has been replaced by tracepoints. Signed-off-by:
Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Signed-off-by:
-
YueHaibing authored
net/ipv6/ila/ila_xlat.c:604:0: warning: macro "ILA_HASH_TABLE_SIZE" is not used [-Wunused-macros] Signed-off-by:
YueHaibing <yuehaibing@huawei.com> Signed-off-by:
-
wenxu authored
When the act_ct SW offload in flowtable, The counter of the conntrack entry will never update. So update the nf_conn_acct conuter in act_ct flowtable software offload. Signed-off-by:
wenxu <wenxu@ucloud.cn> Reviewed-by:
Roi Dayan <roid@mellanox.com> Signed-off-by:
-
David S. Miller authored
Heiner Kallweit says: ==================== net: phy: add device-managed devm_mdiobus_register If there's no special ordering requirement for mdiobus_unregister(), then driver code can be simplified by using a device-managed version of mdiobus_register(). Prerequisite is that bus allocation has been done device-managed too. Else mdiobus_free() may be called whilst bus is still registered, resulting in a BUG_ON(). Therefore let devm_mdiobus_register() return -EPERM if bus was allocated non-managed. First user of the new functionality is r8169 driver. ==================== Signed-off-by: -
Heiner Kallweit authored
Use new function devm_mdiobus_register() to simplify the driver. Signed-off-by:
Heiner Kallweit <hkallweit1@gmail.com> Signed-off-by:
-
Heiner Kallweit authored
If there's no special ordering requirement for mdiobus_unregister(), then driver code can be simplified by using a device-managed version of mdiobus_register(). Prerequisite is that bus allocation has been done device-managed too. Else mdiobus_free() may be called whilst bus is still registered, resulting in a BUG_ON(). Therefore let devm_mdiobus_register() return -EPERM if bus was allocated non-managed. Signed-off-by:
-
Michael Walle authored
The PHY supports monitoring its die temperature as well as two analog voltages. Add support for it. Signed-off-by:
Michael Walle <michael@walle.cc> Acked-by:
Guenter Roeck <linux@roeck-us.net> Reviewed-by:
-
Michael Walle authored
The Broadcom BCM54140 is a Quad SGMII/QSGMII Copper/Fiber Gigabit Ethernet transceiver. This also adds support for tunables to set and get downshift and energy detect auto power-down. The PHY has four ports and each port has its own PHY address. There are per-port registers as well as global registers. Unfortunately, the global registers can only be accessed by reading and writing from/to the PHY address of the first port. Further, there is no way to find out what port you actually are by just reading the per-port registers. We therefore, have to scan the bus on the PHY probe to determine the port and thus what address we need to access the global registers. Signed-off-by:
-
Michael Walle authored
RDB (Register Data Base) registers are used on newer Broadcom PHYs. Add helper to read, write and modify these registers. Signed-off-by:
Florian Fainelli <f.fainelli@gmail.com> Signed-off-by:
-
David S. Miller authored
Florian Fainelli says: ==================== dt-bindings: net: mdio.yaml fixes This patch series documents some common MDIO devices properties such as resets (and delays) and broken-turn-around. The second patch also rephrases some descriptions to be more general towards MDIO devices and not specific towards Ethernet PHYs. Changes in v3: - corrected wording of 'broken-turn-around' in ethernet-phy.yaml and mdio.yaml, add Andrew's R-b tag to patch #3 ==================== Signed-off-by: -
Florian Fainelli authored
A number of descriptions assume a PHY device, but since this binding describes a MDIO bus which can have different kinds of MDIO devices attached to it, rephrase some descriptions to be more general in that regard. Reviewed-by:
-
Florian Fainelli authored
Some of the properties pertaining to the broken turn around or resets were only documented in ethernet-phy.yaml while they are applicable across all MDIO devices and not Ethernet PHYs specifically which are a superset. Signed-off-by:
-
Florian Fainelli authored
The turn around bytes (2) are placed between the control phase of the MDIO transaction and the data phase, correct the wording to be more exact. Signed-off-by:
-
David S. Miller authored
Vladimir Oltean says: ==================== Ocelot MAC_ETYPE tc-flower key improvements As discussed in the comments surrounding this patch: https://patchwork.ozlabs.org/project/netdev/patch/20200417190308.32598-1-olteanv@gmail.com/ the restrictions imposed on non-MAC_ETYPE rules were harsher than they needed to be. IP, IPv6, ARP rules can still be added concurrently with src_mac and dst_mac rules, as long as those MAC address rules do not ask for an offending EtherType. For that to actually be supported, we need to parse the EtherType from the flower classification rule first. ==================== Signed-off-by:
-
Vladimir Oltean authored
An attempt was made in commit fe3490e6 ("net: mscc: ocelot: Hardware ofload for tc flower filter") to avoid clashes between MAC_ETYPE rules and IP rules. Because the protocol blacklist should have included ETH_P_ALL too, it created some confusion, but now the situation should be dealt with a bit better by the patch immediately previous to this one ("net: mscc: ocelot: refine the ocelot_ace_is_problematic_mac_etype function"). So now we can remove that check. MAC_ETYPE rules with a protocol of ETH_P_IP, ETH_P_IPV6, ETH_P_ARP and ETH_P_ALL _are_ supported, with some restrictions regarding per-port exclusivity which are enforced now. Signed-off-by:
-
