linux/drivers/net, branch v6.0

Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue

2022-09-29T02:04:43Z

Tony Nguyen says: ==================== ice: xsk: ZC changes Maciej Fijalkowski says: This set consists of two fixes to issues that were either pointed out on indirectly (John was reviewing AF_XDP selftests that were testing ice's ZC support) mailing list or were directly reported by customers. First patch allows user space to see done descriptor in CQ even after a single frame being transmitted and second patch removes the need for having HW rings sized to power of 2 number of descriptors when used against AF_XDP. I also forgot to mention that due to the current Tx cleaning algorithm, 4k HW ring was broken and these two patches bring it back to life, so we kill two birds with one stone. * '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue: ice: xsk: drop power of 2 ring size restriction for AF_XDP ice: xsk: change batched Tx descriptor cleaning ==================== Link: https://lore.kernel.org/r/20220927164112.4011983-1-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski

net: ethernet: mtk_eth_soc: fix mask of RX_DMA_GET_SPORT{,_V2}

2022-09-29T02:03:57Z

The bitmasks applied in RX_DMA_GET_SPORT and RX_DMA_GET_SPORT_V2 macros were swapped. Fix that. Reported-by: Chen Minqiang Fixes: 160d3a9b192985 ("net: ethernet: mtk_eth_soc: introduce MTK_NETSYS_V2 support") Acked-by: Lorenzo Bianconi Signed-off-by: Daniel Golle Link: https://lore.kernel.org/r/YzMW+mg9UsaCdKRQ@makrotopia.org Signed-off-by: Jakub Kicinski

net: mscc: ocelot: fix tagged VLAN refusal while under a VLAN-unaware bridge

2022-09-29T02:03:28Z

Currently the following set of commands fails: $ ip link add br0 type bridge # vlan_filtering 0 $ ip link set swp0 master br0 $ bridge vlan port vlan-id swp0 1 PVID Egress Untagged $ bridge vlan add dev swp0 vid 10 Error: mscc_ocelot_switch_lib: Port with more than one egress-untagged VLAN cannot have egress-tagged VLANs. Dumping ocelot->vlans, one can see that the 2 egress-untagged VLANs on swp0 are vid 1 (the bridge PVID) and vid 4094, a PVID used privately by the driver for VLAN-unaware bridging. So this is why bridge vid 10 is refused, despite 'bridge vlan' showing a single egress untagged VLAN. As mentioned in the comment added, having this private VLAN does not impose restrictions to the hardware configuration, yet it is a bookkeeping problem. There are 2 possible solutions. One is to make the functions that operate on VLAN-unaware pvids: - ocelot_add_vlan_unaware_pvid() - ocelot_del_vlan_unaware_pvid() - ocelot_port_setup_dsa_8021q_cpu() - ocelot_port_teardown_dsa_8021q_cpu() call something different than ocelot_vlan_member_(add|del)(), the latter being the real problem, because it allocates a struct ocelot_bridge_vlan *vlan which it adds to ocelot->vlans. We don't really *need* the private VLANs in ocelot->vlans, it's just that we have the extra convenience of having the vlan->portmask cached in software (whereas without these structures, we'd have to create a raw ocelot_vlant_rmw_mask() procedure which reads back the current port mask from hardware). The other solution is to filter out the private VLANs from ocelot_port_num_untagged_vlans(), since they aren't what callers care about. We only need to do this to the mentioned function and not to ocelot_port_num_tagged_vlans(), because private VLANs are never egress-tagged. Nothing else seems to be broken in either solution, but the first one requires more rework which will conflict with the net-next change 36a0bf443585 ("net: mscc: ocelot: set up tag_8021q CPU ports independent of user port affinity"), and I'd like to avoid that. So go with the other one. Fixes: 54c319846086 ("net: mscc: ocelot: enforce FDB isolation when VLAN-unaware") Signed-off-by: Vladimir Oltean Link: https://lore.kernel.org/r/20220927122042.1100231-1-vladimir.oltean@nxp.com Signed-off-by: Jakub Kicinski

can: c_can: don't cache TX messages for C_CAN cores

2022-09-28T08:34:04Z

As Jacob noticed, the optimization introduced in 387da6bc7a82 ("can: c_can: cache frames to operate as a true FIFO") doesn't properly work on C_CAN, but on D_CAN IP cores. The exact reasons are still unknown. For now disable caching if CAN frames in the TX path for C_CAN cores. Fixes: 387da6bc7a82 ("can: c_can: cache frames to operate as a true FIFO") Link: https://lore.kernel.org/all/20220928083354.1062321-1-mkl@pengutronix.de Link: https://lore.kernel.org/all/15a8084b-9617-2da1-6704-d7e39d60643b@gmail.com Reported-by: Jacob Kroon Tested-by: Jacob Kroon Cc: stable@vger.kernel.org # v5.15 Signed-off-by: Marc Kleine-Budde

ice: xsk: drop power of 2 ring size restriction for AF_XDP

2022-09-27T16:01:01Z

We had multiple customers in the past months that reported commit 296f13ff3854 ("ice: xsk: Force rings to be sized to power of 2") makes them unable to use ring size of 8160 in conjunction with AF_XDP. Remove this restriction. Fixes: 296f13ff3854 ("ice: xsk: Force rings to be sized to power of 2") CC: Alasdair McWilliam Signed-off-by: Maciej Fijalkowski Tested-by: George Kuruvinakunnel Signed-off-by: Tony Nguyen

ice: xsk: change batched Tx descriptor cleaning

2022-09-27T15:11:02Z

AF_XDP Tx descriptor cleaning in ice driver currently works in a "lazy" way - descriptors are not cleaned immediately after send. We rather hold on with cleaning until we see that free space in ring drops below particular threshold. This was supposed to reduce the amount of unnecessary work related to cleaning and instead of keeping the ring empty, ring was rather saturated. In AF_XDP realm cleaning Tx descriptors implies producing them to CQ. This is a way of letting know user space that particular descriptor has been sent, as John points out in [0]. We tried to implement serial descriptor cleaning which would be used in conjunction with batched cleaning but it made code base more convoluted and probably harder to maintain in future. Therefore we step away from batched cleaning in a current form in favor of an approach where we set RS bit on every last descriptor from a batch and clean always at the beginning of ice_xmit_zc(). This means that we give up a bit of Tx performance, but this doesn't hurt l2fwd scenario which is way more meaningful than txonly as this can be treaten as AF_XDP based packet generator. l2fwd is not hurt due to the fact that Tx side is much faster than Rx and Rx is the one that has to catch Tx up. FWIW Tx descriptors are still produced in a batched way. [0]: https://lore.kernel.org/bpf/62b0a20232920_3573208ab@john.notmuch/ Fixes: 126cdfe1007a ("ice: xsk: Improve AF_XDP ZC Tx and use batching API") Signed-off-by: Maciej Fijalkowski Tested-by: George Kuruvinakunnel Signed-off-by: Tony Nguyen

net: usb: qmi_wwan: Add new usb-id for Dell branded EM7455

2022-09-27T14:32:11Z

Add support for Dell 5811e (EM7455) with USB-id 0x413c:0x81c2. Signed-off-by: Frank Wunderlich Cc: stable@vger.kernel.org Acked-by: Bjørn Mork Link: https://lore.kernel.org/r/20220926150740.6684-3-linux@fw-web.de Signed-off-by: Jakub Kicinski

net: phy: Don't WARN for PHY_UP state in mdio_bus_phy_resume()

2022-09-27T08:48:38Z

Commit 744d23c71af3 ("net: phy: Warn about incorrect mdio_bus_phy_resume() state") introduced a WARN() on resume from system sleep if a PHY is not in PHY_HALTED state. Commit 6dbe852c379f ("net: phy: Don't WARN for PHY_READY state in mdio_bus_phy_resume()") added an exemption for PHY_READY state from the WARN(). It turns out PHY_UP state needs to be exempted as well because the following may happen on suspend: mdio_bus_phy_suspend() phy_stop_machine() phydev->state = PHY_UP # if (phydev->state >= PHY_UP) Fixes: 744d23c71af3 ("net: phy: Warn about incorrect mdio_bus_phy_resume() state") Reported-by: Marek Szyprowski Tested-by: Marek Szyprowski Link: https://lore.kernel.org/netdev/2b1a1588-505e-dff3-301d-bfc1fb14d685@samsung.com/ Signed-off-by: Lukas Wunner Acked-by: Florian Fainelli Cc: Xiaolei Wang Link: https://lore.kernel.org/r/8128fdb51eeebc9efbf3776a4097363a1317aaf1.1663905575.git.lukas@wunner.de Signed-off-by: Paolo Abeni

net: stmmac: power up/down serdes in stmmac_open/release

2022-09-27T08:38:11Z

This commit fixes DMA engine reset timeout issue in suspend/resume with ADLink I-Pi SMARC Plus board which dmesg shows: ... [ 54.678271] PM: suspend exit [ 54.754066] intel-eth-pci 0000:00:1d.2 enp0s29f2: PHY [stmmac-3:01] driver [Maxlinear Ethernet GPY215B] (irq=POLL) [ 54.755808] intel-eth-pci 0000:00:1d.2 enp0s29f2: Register MEM_TYPE_PAGE_POOL RxQ-0 ... [ 54.780482] intel-eth-pci 0000:00:1d.2 enp0s29f2: Register MEM_TYPE_PAGE_POOL RxQ-7 [ 55.784098] intel-eth-pci 0000:00:1d.2: Failed to reset the dma [ 55.784111] intel-eth-pci 0000:00:1d.2 enp0s29f2: stmmac_hw_setup: DMA engine initialization failed [ 55.784115] intel-eth-pci 0000:00:1d.2 enp0s29f2: stmmac_open: Hw setup failed ... The issue is related with serdes which impacts clock. There is serdes in ADLink I-Pi SMARC board ethernet controller. Please refer to commit b9663b7ca6ff78 ("net: stmmac: Enable SERDES power up/down sequence") for detial. When issue is reproduced, DMA engine clock is not ready because serdes is not powered up. To reproduce DMA engine reset timeout issue with hardware which has serdes in GBE controller, install Ubuntu. In Ubuntu GUI, click "Power Off/Log Out" -> "Suspend" menu, it disables network interface, then goes to sleep mode. When it wakes up, it enables network interface again. Stmmac driver is called in this way: 1. stmmac_release: Stop network interface. In this function, it disables DMA engine and network interface; 2. stmmac_suspend: It is called in kernel suspend flow. But because network interface has been disabled(netif_running(ndev) is false), it does nothing and returns directly; 3. System goes into S3 or S0ix state. Some time later, system is waken up by keyboard or mouse; 4. stmmac_resume: It does nothing because network interface has been disabled; 5. stmmac_open: It is called to enable network interace again. DMA engine is initialized in this API, but serdes is not power on so there will be DMA engine reset timeout issue. Similarly, serdes powerdown should be added in stmmac_release. Network interface might be disabled by cmd "ifconfig eth0 down", DMA engine, phy and mac have been disabled in ndo_stop callback, serdes should be powered down as well. It doesn't make sense that serdes is on while other components have been turned off. If ethernet interface is in enabled state(netif_running(ndev) is true) before suspend/resume, the issue couldn't be reproduced because serdes could be powered up in stmmac_resume. Because serdes_powerup is added in stmmac_open, it doesn't need to be called in probe function. Fixes: b9663b7ca6ff78 ("net: stmmac: Enable SERDES power up/down sequence") Signed-off-by: Junxiao Chang Reviewed-by: Voon Weifeng Tested-by: Jimmy JS Chen Tested-by: Looi, Hong Aun Link: https://lore.kernel.org/r/20220923050448.1220250-1-junxiao.chang@intel.com Signed-off-by: Paolo Abeni

net: hippi: Add missing pci_disable_device() in rr_init_one()

2022-09-26T20:22:39Z

Add missing pci_disable_device() if rr_init_one() fails Signed-off-by: ruanjinjie Link: https://lore.kernel.org/r/20220923094320.3109154-1-ruanjinjie@huawei.com Signed-off-by: Jakub Kicinski