- 16 Aug, 2021 15 commits
-
-
Vladimir Oltean authored
The felix DSA driver, which is a wrapper over the same hardware class as ocelot, is integrated with phylink, but ocelot is using the plain PHY library. It makes sense to bring together the two implementations, which is what this patch achieves. This is a large patch and hard to break up, but it does the following: The existing ocelot_adjust_link writes some registers, and felix_phylink_mac_link_up writes some registers, some of them are common, but both functions write to some registers to which the other doesn't. The main reasons for this are: - Felix switches so far have used an NXP PCS so they had no need to write the PCS1G registers that ocelot_adjust_link writes - Felix switches have the MAC fixed at 1G, so some of the MAC speed changes actually break the link and must be avoided. The naming conventions for the functions introduced in this patch are: - vsc7514_phylink_{mac_config,validate} are specific to the Ocelot instantiations and placed in ocelot_net.c which is built only for the ocelot switchdev driver. - ocelot_phylink_mac_link_{up,down} are shared between the ocelot switchdev driver and the felix DSA driver (they are put in the common lib). One by one, the registers written by ocelot_adjust_link are: DEV_MAC_MODE_CFG - felix_phylink_mac_link_up had no need to write this register since its out-of-reset value was fine and did not need changing. The write is moved to the common ocelot_phylink_mac_link_up and on felix it is guarded by a quirk bit that makes the written value identical with the out-of-reset one DEV_PORT_MISC - runtime invariant, was moved to vsc7514_phylink_mac_config PCS1G_MODE_CFG - same as above PCS1G_SD_CFG - same as above PCS1G_CFG - same as above PCS1G_ANEG_CFG - same as above PCS1G_LB_CFG - same as above DEV_MAC_ENA_CFG - both ocelot_adjust_link and ocelot_port_disable touched this. felix_phylink_mac_link_{up,down} also do. We go with what felix does and put it in ocelot_phylink_mac_link_up. DEV_CLOCK_CFG - ocelot_adjust_link and felix_phylink_mac_link_up both write this, but to different values. Move to the common ocelot_phylink_mac_link_up and make sure via the quirk that the old values are preserved for both. ANA_PFC_PFC_CFG - ocelot_adjust_link wrote this, felix_phylink_mac_link_up did not. Runtime invariant, speed does not matter since PFC is disabled via the RX_PFC_ENA bits which are cleared. Move to vsc7514_phylink_mac_config. QSYS_SWITCH_PORT_MODE_PORT_ENA - both ocelot_adjust_link and felix_phylink_mac_link_{up,down} wrote this. Ocelot also wrote this register from ocelot_port_disable. Keep what felix did, move in ocelot_phylink_mac_link_{up,down} and delete ocelot_port_disable. ANA_POL_FLOWC - same as above SYS_MAC_FC_CFG - same as above, except slight behavior change. Whereas ocelot always enabled RX and TX flow control, felix listened to phylink (for the most part, at least - see the 2500base-X comment). The registers which only felix_phylink_mac_link_up wrote are: SYS_PAUSE_CFG_PAUSE_ENA - this is why I am not sure that flow control worked on ocelot. Not it should, since the code is shared with felix where it does. ANA_PORT_PORT_CFG - this is a Frame Analyzer block register, phylink should be the one touching them, deleted. Other changes: - The old phylib registration code was in mscc_ocelot_init_ports. It is hard to work with 2 levels of indentation already in, and with hard to follow teardown logic. The new phylink registration code was moved inside ocelot_probe_port(), right between alloc_etherdev() and register_netdev(). It could not be done before (=> outside of) ocelot_probe_port() because ocelot_probe_port() allocates the struct ocelot_port which we then use to assign ocelot_port->phy_mode to. It is more preferable to me to have all PHY handling logic inside the same function. - On the same topic: struct ocelot_port_private :: serdes is only used in ocelot_port_open to set the SERDES protocol to Ethernet. This is logically a runtime invariant and can be done just once, when the port registers with phylink. We therefore don't even need to keep the serdes reference inside struct ocelot_port_private, or to use the devm variant of of_phy_get(). - Phylink needs a valid phy-mode for phylink_create() to succeed, and the existing device tree bindings in arch/mips/boot/dts/mscc/ocelot_pcb120.dts don't define one for the internal PHY ports. So we patch PHY_INTERFACE_MODE_NA into PHY_INTERFACE_MODE_INTERNAL. - There was a strategically placed: switch (priv->phy_mode) { case PHY_INTERFACE_MODE_NA: continue; which made the code skip the serdes initialization for the internal PHY ports. Frankly that is not all that obvious, so now we explicitly initialize the serdes under an "if" condition and not rely on code jumps, so everything is clearer. - There was a write of OCELOT_SPEED_1000 to DEV_CLOCK_CFG for QSGMII ports. Since that is in fact the default value for the register field DEV_CLOCK_CFG_LINK_SPEED, I can only guess the intention was to clear the adjacent fields, MAC_TX_RST and MAC_RX_RST, aka take the port out of reset, which does match the comment. I don't even want to know why this code is placed there, but if there is indeed an issue that all ports that share a QSGMII lane must all be up, then this logic is already buggy, since mscc_ocelot_init_ports iterates using for_each_available_child_of_node, so nobody prevents the user from putting a 'status = "disabled";' for some QSGMII ports which would break the driver's assumption. In any case, in the eventuality that I'm right, we would have yet another issue if ocelot_phylink_mac_link_down would reset those ports and that would be forbidden, so since the ocelot_adjust_link logic did not do that (maybe for a reason), add another quirk to preserve the old logic. The ocelot driver teardown goes through all ports in one fell swoop. When initialization of one port fails, the ocelot->ports[port] pointer for that is reset to NULL, and teardown is done only for non-NULL ports, so there is no reason to do partial teardowns, let the central mscc_ocelot_release_ports() do its job. Tested bind, unbind, rebind, link up, link down, speed change on mock-up hardware (modified the driver to probe on Felix VSC9959). Also regression tested the felix DSA driver. Could not test the Ocelot specific bits (PCS1G, SERDES, device tree bindings). Signed-off-by:Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Vladimir Oltean authored
ocelot_port_enable touches ANA_PORT_PORT_CFG, which has the following fields: - LOCKED_PORTMOVE_CPU, LEARNDROP, LEARNCPU, LEARNAUTO, RECV_ENA, all of which are written with their hardware default values, also runtime invariants. So it makes no sense to write these during every .ndo_open. - PORTID_VAL: this field has an out-of-reset value of zero for all ports and must be initialized by software. Additionally, the ocelot_setup_logical_port_ids() code path sets up different logical port IDs for the ports in a hardware LAG, and we absolutely don't want .ndo_open to interfere there and reset those values. So in fact the write from ocelot_port_enable can better be moved to ocelot_init_port, and the .ndo_open hook deleted. ocelot_port_disable touches DEV_MAC_ENA_CFG and QSYS_SWITCH_PORT_MODE_PORT_ENA, in an attempt to undo what ocelot_adjust_link did. But since .ndo_stop does not get called each time the link falls (i.e. this isn't a substitute for .phylink_mac_link_down), felix already does better at this by writing those registers already in felix_phylink_mac_link_down. So keep ocelot_port_disable (for now, until ocelot is converted to phylink too), and just delete the felix call to it, which is not necessary. Signed-off-by:
-
Changbin Du authored
Replace the obsolete and ambiguos macro in_irq() with new macro in_hardirq(). Signed-off-by:
Changbin Du <changbin.du@gmail.com> Signed-off-by:
-
Vladimir Oltean authored
During the development of the blamed patch, the "bool broadcast" argument of dsa_port_tag_8021q_vlan_{add,del} was originally called "bool local", and the meaning was the exact opposite. Due to a rookie mistake where the patch was modified at the last minute without retesting, the instances of dsa_port_tag_8021q_vlan_{add,del} are called with the wrong values. During setup and teardown, cross-chip notifiers should not be broadcast to all DSA trees, while during bridging, they should. Fixes: 724395f4 ("net: dsa: tag_8021q: don't broadcast during setup/teardown") Signed-off-by: -
Randy Dunlap authored
Fix kconfig warning on arch/s390/: WARNING: unmet direct dependencies detected for SERIAL_8250 Depends on [n]: TTY [=y] && HAS_IOMEM [=y] && !S390 [=y] Selected by [m]: - PTP_1588_CLOCK_OCP [=m] && PTP_1588_CLOCK [=m] && HAS_IOMEM [=y] && PCI [=y] && SPI [=y] && I2C [=m] && MTD [=m] Signed-off-by:
Randy Dunlap <rdunlap@infradead.org> Cc: Richard Cochran <richardcochran@gmail.com> Cc: Jonathan Lemon <jonathan.lemon@gmail.com> Signed-off-by:
-
Rao Shoaib authored
edumazet@google.com pointed out that queue_oob does not check socket state after acquiring the lock. He also pointed to an incorrect usage of kfree_skb and an unnecessary setting of skb length. This patch addresses those issue. Signed-off-by:
Rao Shoaib <Rao.Shoaib@oracle.com> Signed-off-by:
-
Song Yoong Siang authored
Add Wake-on-PHY feature support by enabling the Link Up Event. Signed-off-by:
Song Yoong Siang <yoong.siang.song@intel.com> Reviewed-by:
Andrew Lunn <andrew@lunn.ch> Signed-off-by:
-
Wong Vee Khee authored
SGMII/2500BaseX supports Pause frame as defined in the IEEE802.3x Flow Control standardization. Add this as a supported feature under the xpcs_sgmii_features struct. Cc: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by:
Wong Vee Khee <vee.khee.wong@linux.intel.com> Reviewed-by:
-
David S. Miller authored
samples: pktgen: enhance the usability of pktgen samples This patchset improves the usability of pktgen samples by adding an option for propagating the environment variable of normal user to sudo. And also adds the missing IPv6 option to pktgen scripts. Currently, all pktgen samples are able to use the environment variable instead of optional parameters. However, it doesn't work appropriately when running samples as normal user. This is results of running sample as root and user: // running as root # DEV=eth0 DEST_IP=10.1.0.1 DST_MAC=00:11:22:33:44:55 ./pktgen_sample01_simple.sh -v -n 1 Running... ctrl^C to stop // running as normal user $ DEV=eth0 DEST_IP=10.1.0.1 DST_MAC=00:11:22:33:44:55 ./pktgen_sample01_simple.sh -v -n 1 [...] ERROR: Please specify output device The reason why passing the environment varaible doesn't work properly when running samples as normal user is that the environment variable of normal user doesn't propagate to sudo (root_check_run_with_sudo)). So the first commit solves this issue by using "-E" (--preserve-env) option of "sudo", which passes normal user's existing environment variables. Also, "sample04" and "sample05" are not working properly when running with IPv6 option parameter("-6"). Because the commit 0f06a678 ("samples: Add an IPv6 "-6" option to the pktgen scripts") has omitted the addition of this option at these samples. So the second commit adds missing IPv6 option to pktgen scripts. ==================== Fixes: 0f06a678 ("samples: Add an IPv6 "-6" option to the pktgen scripts") Signed-off-by: -
Juhee Kang authored
Currently, "sample04" and "sample05" are not working properly when running with an IPv6 option("-6"). The commit 0f06a678 ("samples: Add an IPv6 "-6" option to the pktgen scripts") has omitted the addition of this option at "sample04" and "sample05". In order to support IPv6 option, this commit adds logic related to IPv6 option. Fixes: 0f06a678 ("samples: Add an IPv6 "-6" option to the pktgen scripts") Signed-off-by:Juhee Kang <claudiajkang@gmail.com> Signed-off-by:
-
Juhee Kang authored
All pktgen samples can use the environment variable instead of option parameters(eg. $DEV is able to use instead of '-i' option). This is results of running sample as root and user: // running as root # DEV=eth0 DEST_IP=10.1.0.1 DST_MAC=00:11:22:33:44:55 ./pktgen_sample01_simple.sh -v -n 1 Running... ctrl^C to stop // running as normal user $ DEV=eth0 DEST_IP=10.1.0.1 DST_MAC=00:11:22:33:44:55 ./pktgen_sample01_simple.sh -v -n 1 [...] ERROR: Please specify output device This results show the sample doesn't work properly when the sample runs as normal user. Because the sample is restarted by the function (root_check_run_with_sudo) to run with sudo. In this process, the environment variable of normal user doesn't propagate to sudo. It can be solved by using "-E"(--preserve-env) option of "sudo", which preserve normal user's existing environment variables. So this commit adds "-E" option in the function (root_check_run_with_sudo). Signed-off-by: -
David S. Miller authored
Luo Jie says: ==================== net: mdio: Add IPQ MDIO reset related function This patch series add the MDIO reset features, which includes configuring MDIO clock source frequency and indicating CMN_PLL that ethernet LDO has been ready, this ethernet LDO is dedicated in the IPQ5018 platform. Specify more chipset IPQ40xx, IPQ807x, IPQ60xx and IPQ50xx supported by this MDIO driver. Changes in v3: * simplify the function ipq_mdio_reset. Changes in v2: * Addressed review comments (Andrew Lunn). * Remove the IS_ERR(). * make binding patch part of series. * document the property 'reg' and 'clock'. Changes in v1: * make MDIO_IPQ4019 unchanged for backwards compatibility. * remove the PHY reset functions ==================== Signed-off-by: -
Luo Jie authored
The new added properties resource "reg" is for configuring ethernet LDO in the IPQ5018 chipset, the property "clocks" is for configuring the MDIO clock source frequency. Signed-off-by:
Luo Jie <luoj@codeaurora.org> Reviewed-by:
-
Luo Jie authored
The IPQ MDIO driver currently supports the chipset IPQ40xx, IPQ807x, IPQ60xx and IPQ50xx. Add the compatible 'qcom,ipq5018-mdio' because of ethernet LDO dedicated to the IPQ5018 platform. Signed-off-by:
-
Luo Jie authored
1. configure the MDIO clock source frequency. 2. the LDO resource is needed to configure the ethernet LDO available for CMN_PLL. Signed-off-by:
-
- 14 Aug, 2021 25 commits
-
-
Cai Huoqing authored
commit <47595e32> ("<MAINTAINERS: Mark some staging directories>") indicated the ipx network layer as obsolete in Jan 2018, updated in the MAINTAINERS file. now, after being exposed for 3 years to refactoring, so to remove the ipx network layer info from MAINTAINERS. additionally, there is no module that depends on ipx.h except a broken staging driver(r8188eu) Signed-off-by:
Cai Huoqing <caihuoqing@baidu.com> Signed-off-by:
-
Cai Huoqing authored
commit <47595e32> ("<MAINTAINERS: Mark some staging directories>") indicated the ipx network layer as obsolete in Jan 2018, updated in the MAINTAINERS file now, after being exposed for 3 years to refactoring, so to delete uapi/linux/ipx.h and net/ipx.h header files for good. additionally, there is no module that depends on ipx.h except a broken staging driver(r8188eu) Signed-off-by:
-
David S. Miller authored
Alex Elder says: ==================== net: ipa: last things before PM conversion This series contains a few remaining changes needed before fully switching over to using runtime power management rather than the previous "IPA clock" mechanism. The first patch moves the calls to enable and disable the IPA interrupt as a system wakeup interrupt into "ipa_clock.c" with the rest of the power-related code. The second adds a flag to make it possible to distinguish runtime suspend from system suspend. The third and fourth patches arrange for the ->start_xmit path to resume hardware if necessary, to ensure it is powered. If power is not active, the TX queue is stopped, and arrangements are made for the queue to be restarted once hardware power is active again. The fifth patch keeps the TX queue active during suspend. This isn't necessary for system suspend but it's important for runtime suspend. And the last patch makes it so we don't hold the hardware active while the modem network device is open. ==================== Signed-off-by: -
Alex Elder authored
Currently a clock reference is taken whenever the ->ndo_open callback for the modem netdev is called. That reference is dropped when the device is closed, in ipa_stop(). We no longer need this, because ipa_start_xmit() now handles the situation where the hardware power state is not active. Drop the clock reference in ipa_open() when we're done, and take a new reference in ipa_stop() before we begin closing the interface. Finally (and unrelated, but trivial), change the return type of ipa_start_xmit() to be netdev_tx_t instead of int. Signed-off-by:
Alex Elder <elder@linaro.org> Signed-off-by:
-
Alex Elder authored
Currently we stop the modem netdev transmit queue when suspending the hardware. For system suspend this ensured we'd never attempt to transmit while attempting to suspend the modem endpoints. For runtime suspend, the IPA hardware might get suspended while the system is operating. In that case we want an attempt to transmit a packet to cause the hardware to resume if necessary. But if we disable the queue this cannot happen. So stop disabling the queue on suspend. In case we end up disabling it in ipa_start_xmit() (see the previous commit), we still arrange to start the TX queue on resume. Signed-off-by:
-
Alex Elder authored
We need to ensure the hardware is powered when we transmit a packet. But if it's not, we can't block to wait for it. So asynchronously request power in ipa_start_xmit(), and only proceed if the return value indicates the power state is active. If the hardware is not active, a runtime resume request will have been initiated. In that case, stop the network stack from further transmit attempts until the resume completes. Return NETDEV_TX_BUSY, to retry sending the packet once the queue is restarted. If the power request returns an error (other than -EINPROGRESS, which just means a resume requested elsewhere isn't complete), just drop the packet. Signed-off-by:
-
Alex Elder authored
Create a new work structure in the modem private data, and use it to re-enable the modem network device transmit queue when resuming. This is needed by the next patch, which stops the TX queue if IPA power isn't active when a transmit request arrives. Packets will start arriving the instant the TX queue is enabled, but resuming isn't complete until ipa_modem_resume() returns. This way we're sure to be resumed before transmits are allowed again. Cancel it before calling ipa_stop() in ipa_modem_stop() to ensure the transmit queue restart completes before it gets stopped there. Signed-off-by:
-
Alex Elder authored
Add a new flag that is set when the hardware is suspended due to a system suspend operation, distingishing it from runtime suspend. Use it in the SUSPEND IPA interrupt handler to determine whether to trigger a system resume because of the event. Define new suspend and resume power management callback functions to set and clear the new flag, respectively. Signed-off-by:
-
Alex Elder authored
Move the call to enable the IPA interrupt as a wakeup interrupt into ipa_power_setup(), disable it in ipa_power_teardown(). Signed-off-by:
-
David S. Miller authored
Nikolay Aleksandrov says: ==================== net: bridge: mcast: dump querier state This set adds the ability to dump the current multicast querier state. This is extremely useful when debugging multicast issues, we've had many cases of unexpected queriers causing strange behaviour and mcast test failures. The first patch changes the querier struct to record a port device's ifindex instead of a pointer to the port itself so we can later retrieve it, I chose this way because it's much simpler and doesn't require us to do querier port ref counting, it is best effort anyway. Then patch 02 makes the querier address/port updates consistent via a combination of multicast_lock and seqcount, so readers can only use seqcount to get a consistent snapshot of address and port. Patch 03 is a minor cleanup in preparation for the dump support, it consolidates IPv4 and IPv6 querier selection paths as they share most of the logic (except address comparisons of course). Finally the last three patches add the new querier state dumping support, for the bridge's global multicast context we embed the BRIDGE_QUERIER_xxx attributes into IFLA_BR_MCAST_QUERIER_STATE and for the per-vlan global mcast contexts we embed them into BRIDGE_VLANDB_GOPTS_MCAST_QUERIER_STATE. The structure is: [IFLA_BR_MCAST_QUERIER_STATE / BRIDGE_VLANDB_GOPTS_MCAST_QUERIER_STATE] `[BRIDGE_QUERIER_IP_ADDRESS] - ip address of the querier `[BRIDGE_QUERIER_IP_PORT] - bridge port ifindex where the querier was seen (set only if external querier) `[BRIDGE_QUERIER_IP_OTHER_TIMER] - other querier timeout `[BRIDGE_QUERIER_IPV6_ADDRESS] - ip address of the querier `[BRIDGE_QUERIER_IPV6_PORT] - bridge port ifindex where the querier was seen (set only if external querier) `[BRIDGE_QUERIER_IPV6_OTHER_TIMER] - other querier timeout Later we can also add IGMP version of seen queriers and last seen values from the queries. ==================== -
Nikolay Aleksandrov authored
Use the new mcast querier state dump infrastructure and export vlans' mcast context querier state embedded in attribute BRIDGE_VLANDB_GOPTS_MCAST_QUERIER_STATE. Signed-off-by:
Nikolay Aleksandrov <nikolay@nvidia.com> Signed-off-by:
-
Nikolay Aleksandrov authored
Add support for dumping global IPv6 querier state, we dump the state only if our own querier is enabled or there has been another external querier which has won the election. For the bridge global state we use a new attribute IFLA_BR_MCAST_QUERIER_STATE and embed the state inside. The structure is: [IFLA_BR_MCAST_QUERIER_STATE] `[BRIDGE_QUERIER_IPV6_ADDRESS] - ip address of the querier `[BRIDGE_QUERIER_IPV6_PORT] - bridge port ifindex where the querier was seen (set only if external querier) `[BRIDGE_QUERIER_IPV6_OTHER_TIMER] - other querier timeout IPv4 and IPv6 attributes are embedded at the same level of IFLA_BR_MCAST_QUERIER_STATE. If we didn't dump anything we cancel the nest and return. Signed-off-by: -
Nikolay Aleksandrov authored
Add support for dumping global IPv4 querier state, we dump the state only if our own querier is enabled or there has been another external querier which has won the election. For the bridge global state we use a new attribute IFLA_BR_MCAST_QUERIER_STATE and embed the state inside. The structure is: [IFLA_BR_MCAST_QUERIER_STATE] `[BRIDGE_QUERIER_IP_ADDRESS] - ip address of the querier `[BRIDGE_QUERIER_IP_PORT] - bridge port ifindex where the querier was seen (set only if external querier) `[BRIDGE_QUERIER_IP_OTHER_TIMER] - other querier timeout Signed-off-by: -
Nikolay Aleksandrov authored
We can consolidate both functions as they share almost the same logic. This is easier to maintain and we have a single querier update function. Signed-off-by:
-
Nikolay Aleksandrov authored
Use a sequence counter to make sure port/address updates can be read consistently without requiring the bridge multicast_lock. We need to zero out the port and address when the other querier has expired and we're about to select ourselves as querier. br_multicast_read_querier will be used later when dumping querier state. Updates are done only with the multicast spinlock and softirqs disabled, while reads are done from process context and from softirqs (due to notifications). Signed-off-by:
-
Nikolay Aleksandrov authored
Currently when a querier port is detected its net_bridge_port pointer is recorded, but it's used only for comparisons so it's fine to have stale pointer, in order to dereference and use the port pointer a proper accounting of its usage must be implemented adding unnecessary complexity. To solve the problem we can just store the netdevice ifindex instead of the port pointer and retrieve the bridge port. It is a best effort and the device needs to be validated that is still part of that bridge before use, but that is small price to pay for avoiding querier reference counting for each port/vlan. Signed-off-by:
-
David S. Miller authored
Leon Romanovsky says: ==================== Devlink cleanup for delay event series Jakub's request to make sure that devlink events are delayed and not printed till they fully accessible [1] requires us to implement delayed event notification system in the devlink. In order to do it, I moved some of my patches (xarray e.t.c) from the future series to be before "Move devlink_register to be near devlink_reload_enable" [2]. That allows us to rely on DEVLINK_REGISTERED xarray mark to decide if to print event or not. Other patches are simple cleanup which is needed anyway. [1] https://lore.kernel.org/lkml/20210811071817.4af5ab34@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com [2] https://lore.kernel.org/lkml/cover.1628599239.git.leonro@nvidia.com Next in the queue: * Delay event series * Move devlink_register to be near devlink_reload_enable" * Extension of devlink_ops to be set dynamically * devlink_reload_* delete * Devlink locks rework to user xarray and reference counting * ???? ==================== Signed-off-by:
-
Leon Romanovsky authored
The devlink pointer always exists after hclge_devlink_init() succeed. Remove that check together with NULL setting after release and ensure that devlink_register is last command prior to call to devlink_reload_enable(). Fixes: b741269b ("net: hns3: add support for registering devlink for PF") Signed-off-by:
Leon Romanovsky <leonro@nvidia.com> Signed-off-by:
-
Leon Romanovsky authored
The { 0 } doesn't clear all fields in the struct, but tells to the compiler to set all fields to zero and doesn't touch any sub-fields if they exists. The {} is an empty initialiser that instructs to fully initialize whole struct including sub-fields, which is error-prone for future devlink_flash_notify extensions. Fixes: 6700acc5 ("devlink: collect flash notify params into a struct") Signed-off-by: -
Leon Romanovsky authored
We can use xarray instead of linearly organized linked lists for the devlink instances. This will let us revise the locking scheme in favour of internal xarray locking that protects database. Signed-off-by:
-
Leon Romanovsky authored
The struct devlink itself is protected by internal lock and doesn't need global lock during operation. That global lock is used to protect addition/removal new devlink instances from the global list in use by all devlink consumers in the system. The future conversion of linked list to be xarray will allow us to actually delete that lock, but first we need to count all struct devlink users. The reference counting provides us a way to ensure that no new user space commands success to grab devlink instance which is going to be destroyed makes it is safe to access it without lock. Signed-off-by:
-
Leon Romanovsky authored
Devlink objects are accessible only after they were registered and have valid devlink_*->devlink pointers. Remove that check and simplify respective fill functions as an outcome of such change. Signed-off-by:
-
Leon Romanovsky authored
The devlink_pernet_pre_exit() will be called if net namespace exits. That routine is relevant for devlink instances that were assigned to that namespaces first. This assignment is possible only with the following command: "devlink reload DEV netns ...", which already checks reload support. Signed-off-by:
-
David S. Miller authored
Mat Martineau says: ==================== mptcp: Improve use of backup subflows Multipath TCP combines multiple TCP subflows in to one stream, and the MPTCP-level socket must decide which subflow to use when sending (or resending) chunks of data. The choice of the "best" subflow to transmit on can vary depending on the priority (normal or backup) for each subflow and how well the subflow is performing. In order to improve MPTCP performance when some subflows are failing, this patch set changes how backup subflows are utilized and introduces tracking of "stale" subflows that are still connected but not making progress. Patch 1 adjusts MPTCP-level retransmit timeouts to use data from all subflows. Patch 2 makes MPTCP-level retransmissions less aggressive to avoid resending data that's still queued at the TCP level. Patch 3 changes the way pending data is handled when subflows are closed. Unacked MPTCP-level data still in the subflow tx queue is immediately moved to another subflow for transmission instead of waiting for MPTCP-level timeouts to trigger retransmission. Patch 4 has some sysctl code cleanup. Patches 5 and 6 add tracking of "stale" subflows, so only underlying TCP subflow connections that appear to be making progress are considered when selecting a subflow to (re)transmit data. How fast a subflow goes stale is configurable with a per-namespace sysctl. Related MIBS are added too. Patch 7 makes sure the backup flag is always correctly recorded when the MP_JOIN SYN/ACK is received for an added subflow. Patch 8 adds more test cases for backup subflows and stale subflows. ==================== Signed-off-by: -
Paolo Abeni authored
Add more test-case for link failures scenario, including recovery from link failure using only backup subflows and bi-directional transfer. Additionally explicitly check for stale count Co-developed-by:
Geliang Tang <geliangtang@gmail.com> Signed-off-by:
Paolo Abeni <pabeni@redhat.com> Signed-off-by:
Mat Martineau <mathew.j.martineau@linux.intel.com> Signed-off-by:
-
