kernel.git/drivers/net/ethernet/intel/igc, branch linux-5.18.y

igc: Reinstate IGC_REMOVED logic and implement it properly

2022-07-29T15:27:59Z

[ Upstream commit 7c1ddcee5311f3315096217881d2dbe47cc683f9 ] The initially merged version of the igc driver code (via commit 146740f9abc4, "igc: Add support for PF") contained the following IGC_REMOVED checks in the igc_rd32/wr32() MMIO accessors: u32 igc_rd32(struct igc_hw *hw, u32 reg) { u8 __iomem *hw_addr = READ_ONCE(hw->hw_addr); u32 value = 0; if (IGC_REMOVED(hw_addr)) return ~value; value = readl(&hw_addr[reg]); /* reads should not return all F's */ if (!(~value) && (!reg || !(~readl(hw_addr)))) hw->hw_addr = NULL; return value; } And: #define wr32(reg, val) \ do { \ u8 __iomem *hw_addr = READ_ONCE((hw)->hw_addr); \ if (!IGC_REMOVED(hw_addr)) \ writel((val), &hw_addr[(reg)]); \ } while (0) E.g. igb has similar checks in its MMIO accessors, and has a similar macro E1000_REMOVED, which is implemented as follows: #define E1000_REMOVED(h) unlikely(!(h)) These checks serve to detect and take note of an 0xffffffff MMIO read return from the device, which can be caused by a PCIe link flap or some other kind of PCI bus error, and to avoid performing MMIO reads and writes from that point onwards. However, the IGC_REMOVED macro was not originally implemented: #ifndef IGC_REMOVED #define IGC_REMOVED(a) (0) #endif /* IGC_REMOVED */ This led to the IGC_REMOVED logic to be removed entirely in a subsequent commit (commit 3c215fb18e70, "igc: remove IGC_REMOVED function"), with the rationale that such checks matter only for virtualization and that igc does not support virtualization -- but a PCIe device can become detached even without virtualization being in use, and without proper checks, a PCIe bus error affecting an igc adapter will lead to various NULL pointer dereferences, as the first access after the error will set hw->hw_addr to NULL, and subsequent accesses will blindly dereference this now-NULL pointer. This patch reinstates the IGC_REMOVED checks in igc_rd32/wr32(), and implements IGC_REMOVED the way it is done for igb, by checking for the unlikely() case of hw_addr being NULL. This change prevents the oopses seen when a PCIe link flap occurs on an igc adapter. Fixes: 146740f9abc4 ("igc: Add support for PF") Signed-off-by: Lennert Buytenhek Tested-by: Naama Meir Acked-by: Sasha Neftin Signed-off-by: Tony Nguyen Signed-off-by: Sasha Levin

igc: Fix suspending when PTM is active

2022-04-13T16:17:58Z

Some mainboard/CPU combinations, in particular, Alder Lake-S with a W680 mainboard, have shown problems (system hangs usually, no kernel logs) with suspend/resume when PCIe PTM is enabled and active. In some cases, it could be reproduced when removing the igc module. The best we can do is to stop PTM dialogs from the downstream/device side before the interface is brought down. PCIe PTM will be re-enabled when the interface is being brought up. Fixes: a90ec8483732 ("igc: Add support for PTP getcrosststamp()") Signed-off-by: Vinicius Costa Gomes Tested-by: Naama Meir Acked-by: Paul Menzel Signed-off-by: Tony Nguyen

igc: Fix BUG: scheduling while atomic

2022-04-13T14:58:32Z

Replace usleep_range() method with udelay() method to allow atomic contexts in low-level MDIO access functions. The following issue can be seen by doing the following: $ modprobe -r bonding $ modprobe -v bonding max_bonds=1 mode=1 miimon=100 use_carrier=0 $ ip link set bond0 up $ ifenslave bond0 eth0 eth1 [ 982.357308] BUG: scheduling while atomic: kworker/u64:0/9/0x00000002 [ 982.364431] INFO: lockdep is turned off. [ 982.368824] Modules linked in: bonding sctp ip6_udp_tunnel udp_tunnel mlx4_ib ib_uverbs ib_core mlx4_en mlx4_core nfp tls sunrpc intel_rapl_msr iTCO_wdt iTCO_vendor_support mxm_wmi dcdbas intel_rapl_common sb_edac x86_pkg_temp_thermal intel_powerclamp coretemp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel rapl intel_cstate intel_uncore pcspkr lpc_ich mei_me ipmi_ssif mei ipmi_si ipmi_devintf ipmi_msghandler wmi acpi_power_meter xfs libcrc32c sr_mod cdrom sd_mod t10_pi sg mgag200 drm_kms_helper syscopyarea sysfillrect sysimgblt fb_sys_fops drm ahci libahci crc32c_intel libata i2c_algo_bit tg3 megaraid_sas igc dm_mirror dm_region_hash dm_log dm_mod [last unloaded: bonding] [ 982.437941] CPU: 25 PID: 9 Comm: kworker/u64:0 Kdump: loaded Tainted: G W --------- - - 4.18.0-348.el8.x86_64+debug #1 [ 982.451333] Hardware name: Dell Inc. PowerEdge R730/0H21J3, BIOS 2.7.0 12/005/2017 [ 982.459791] Workqueue: bond0 bond_mii_monitor [bonding] [ 982.465622] Call Trace: [ 982.468355] dump_stack+0x8e/0xd0 [ 982.472056] __schedule_bug.cold.60+0x3a/0x60 [ 982.476919] __schedule+0x147b/0x1bc0 [ 982.481007] ? firmware_map_remove+0x16b/0x16b [ 982.485967] ? hrtimer_fixup_init+0x40/0x40 [ 982.490625] schedule+0xd9/0x250 [ 982.494227] schedule_hrtimeout_range_clock+0x10d/0x2c0 [ 982.500058] ? hrtimer_nanosleep_restart+0x130/0x130 [ 982.505598] ? hrtimer_init_sleeper_on_stack+0x90/0x90 [ 982.511332] ? usleep_range+0x88/0x130 [ 982.515514] ? recalibrate_cpu_khz+0x10/0x10 [ 982.520279] ? ktime_get+0xab/0x1c0 [ 982.524175] ? usleep_range+0x88/0x130 [ 982.528355] usleep_range+0xdd/0x130 [ 982.532344] ? console_conditional_schedule+0x30/0x30 [ 982.537987] ? igc_put_hw_semaphore+0x17/0x60 [igc] [ 982.543432] igc_read_phy_reg_gpy+0x111/0x2b0 [igc] [ 982.548887] igc_phy_has_link+0xfa/0x260 [igc] [ 982.553847] ? igc_get_phy_id+0x210/0x210 [igc] [ 982.558894] ? lock_acquire+0x34d/0x890 [ 982.563187] ? lock_downgrade+0x710/0x710 [ 982.567659] ? rcu_read_unlock+0x50/0x50 [ 982.572039] igc_check_for_copper_link+0x106/0x210 [igc] [ 982.577970] ? igc_config_fc_after_link_up+0x840/0x840 [igc] [ 982.584286] ? rcu_read_unlock+0x50/0x50 [ 982.588661] ? lock_release+0x591/0xb80 [ 982.592939] ? lock_release+0x591/0xb80 [ 982.597220] igc_has_link+0x113/0x330 [igc] [ 982.601887] ? lock_downgrade+0x710/0x710 [ 982.606362] igc_ethtool_get_link+0x6d/0x90 [igc] [ 982.611614] bond_check_dev_link+0x131/0x2c0 [bonding] [ 982.617350] ? bond_time_in_interval+0xd0/0xd0 [bonding] [ 982.623277] ? rcu_read_lock_held+0x62/0xc0 [ 982.627944] ? rcu_read_lock_sched_held+0xe0/0xe0 [ 982.633198] bond_mii_monitor+0x314/0x2500 [bonding] [ 982.638738] ? lock_contended+0x880/0x880 [ 982.643214] ? bond_miimon_link_change+0xa0/0xa0 [bonding] [ 982.649336] ? lock_acquire+0x34d/0x890 [ 982.653615] ? lock_downgrade+0x710/0x710 [ 982.658089] ? debug_object_deactivate+0x221/0x340 [ 982.663436] ? rcu_read_unlock+0x50/0x50 [ 982.667811] ? debug_print_object+0x2b0/0x2b0 [ 982.672672] ? __switch_to_asm+0x41/0x70 [ 982.677049] ? __switch_to_asm+0x35/0x70 [ 982.681426] ? _raw_spin_unlock_irq+0x24/0x40 [ 982.686288] ? trace_hardirqs_on+0x20/0x195 [ 982.690956] ? _raw_spin_unlock_irq+0x24/0x40 [ 982.695818] process_one_work+0x8f0/0x1770 [ 982.700390] ? pwq_dec_nr_in_flight+0x320/0x320 [ 982.705443] ? debug_show_held_locks+0x50/0x50 [ 982.710403] worker_thread+0x87/0xb40 [ 982.714489] ? process_one_work+0x1770/0x1770 [ 982.719349] kthread+0x344/0x410 [ 982.722950] ? kthread_insert_work_sanity_check+0xd0/0xd0 [ 982.728975] ret_from_fork+0x3a/0x50 Fixes: 5586838fe9ce ("igc: Add code for PHY support") Reported-by: Corinna Vinschen Suggested-by: Dima Ruinskiy Signed-off-by: Sasha Neftin Tested-by: Corinna Vinschen Tested-by: Naama Meir Signed-off-by: Tony Nguyen

igc: Fix infinite loop in release_swfw_sync

2022-04-13T14:58:32Z

An infinite loop may occur if we fail to acquire the HW semaphore, which is needed for resource release. This will typically happen if the hardware is surprise-removed. At this stage there is nothing to do, except log an error and quit. Fixes: c0071c7aa5fe ("igc: Add HW initialization code") Suggested-by: Dima Ruinskiy Signed-off-by: Sasha Neftin Tested-by: Naama Meir Signed-off-by: Tony Nguyen

Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

2022-03-03T19:55:12Z

net/batman-adv/hard-interface.c commit 690bb6fb64f5 ("batman-adv: Request iflink once in batadv-on-batadv check") commit 6ee3c393eeb7 ("batman-adv: Demote batadv-on-batadv skip error message") https://lore.kernel.org/all/20220302163049.101957-1-sw@simonwunderlich.de/ net/smc/af_smc.c commit 4d08b7b57ece ("net/smc: Fix cleanup when register ULP fails") commit 462791bbfa35 ("net/smc: add sysctl interface for SMC") https://lore.kernel.org/all/20220302112209.355def40@canb.auug.org.au/ Signed-off-by: Jakub Kicinski

igc: igc_write_phy_reg_gpy: drop premature return

2022-02-28T16:48:45Z

Similar to "igc_read_phy_reg_gpy: drop premature return" patch. igc_write_phy_reg_gpy checks the return value from igc_write_phy_reg_mdic and if it's not 0, returns immediately. By doing this, it leaves the HW semaphore in the acquired state. Drop this premature return statement, the function returns after releasing the semaphore immediately anyway. Fixes: 5586838fe9ce ("igc: Add code for PHY support") Suggested-by: Dima Ruinskiy Reported-by: Corinna Vinschen Signed-off-by: Sasha Neftin Tested-by: Naama Meir Signed-off-by: Tony Nguyen

igc: igc_read_phy_reg_gpy: drop premature return

2022-02-28T16:48:45Z

igc_read_phy_reg_gpy checks the return value from igc_read_phy_reg_mdic and if it's not 0, returns immediately. By doing this, it leaves the HW semaphore in the acquired state. Drop this premature return statement, the function returns after releasing the semaphore immediately anyway. Fixes: 5586838fe9ce ("igc: Add code for PHY support") Signed-off-by: Corinna Vinschen Acked-by: Sasha Neftin Tested-by: Naama Meir Signed-off-by: Tony Nguyen

igc: avoid kernel warning when changing RX ring parameters

2022-02-07T22:23:01Z

Calling ethtool changing the RX ring parameters like this: $ ethtool -G eth0 rx 1024 on igc triggers kernel warnings like this: [ 225.198467] ------------[ cut here ]------------ [ 225.198473] Missing unregister, handled but fix driver [ 225.198485] WARNING: CPU: 7 PID: 959 at net/core/xdp.c:168 xdp_rxq_info_reg+0x79/0xd0 [...] [ 225.198601] Call Trace: [ 225.198604] [ 225.198609] igc_setup_rx_resources+0x3f/0xe0 [igc] [ 225.198617] igc_ethtool_set_ringparam+0x30e/0x450 [igc] [ 225.198626] ethnl_set_rings+0x18a/0x250 [ 225.198631] genl_family_rcv_msg_doit+0xca/0x110 [ 225.198637] genl_rcv_msg+0xce/0x1c0 [ 225.198640] ? rings_prepare_data+0x60/0x60 [ 225.198644] ? genl_get_cmd+0xd0/0xd0 [ 225.198647] netlink_rcv_skb+0x4e/0xf0 [ 225.198652] genl_rcv+0x24/0x40 [ 225.198655] netlink_unicast+0x20e/0x330 [ 225.198659] netlink_sendmsg+0x23f/0x480 [ 225.198663] sock_sendmsg+0x5b/0x60 [ 225.198667] __sys_sendto+0xf0/0x160 [ 225.198671] ? handle_mm_fault+0xb2/0x280 [ 225.198676] ? do_user_addr_fault+0x1eb/0x690 [ 225.198680] __x64_sys_sendto+0x20/0x30 [ 225.198683] do_syscall_64+0x38/0x90 [ 225.198687] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 225.198693] RIP: 0033:0x7f7ae38ac3aa igc_ethtool_set_ringparam() copies the igc_ring structure but neglects to reset the xdp_rxq_info member before calling igc_setup_rx_resources(). This in turn calls xdp_rxq_info_reg() with an already registered xdp_rxq_info. Make sure to unregister the xdp_rxq_info structure first in igc_setup_rx_resources. Fixes: 73f1071c1d29 ("igc: Add support for XDP_TX action") Reported-by: Lennert Buytenhek Signed-off-by: Corinna Vinschen Acked-by: Vinicius Costa Gomes Tested-by: Dvora Fuxbrumer Signed-off-by: Tony Nguyen

igc: don't reserve excessive XDP_PACKET_HEADROOM on XSK Rx to skb

2022-01-31T17:47:13Z

{__,}napi_alloc_skb() allocates and reserves additional NET_SKB_PAD + NET_IP_ALIGN for any skb. OTOH, igc_construct_skb_zc() currently allocates and reserves additional `xdp->data_meta - xdp->data_hard_start`, which is about XDP_PACKET_HEADROOM for XSK frames. There's no need for that at all as the frame is post-XDP and will go only to the networking stack core. Pass the size of the actual data only (+ meta) to __napi_alloc_skb() and don't reserve anything. This will give enough headroom for stack processing. Also, net_prefetch() xdp->data_meta and align the copy size to speed-up memcpy() a little and better match igc_construct_skb(). Fixes: fc9df2a0b520 ("igc: Enable RX via AF_XDP zero-copy") Signed-off-by: Alexander Lobakin Reviewed-by: Michal Swiatkowski Tested-by: Nechama Kraus Signed-off-by: Tony Nguyen

igc: Remove useless DMA-32 fallback configuration

2022-01-27T16:58:24Z

As stated in [1], dma_set_mask() with a 64-bit mask never fails if dev->dma_mask is non-NULL. So, if it fails, the 32 bits case will also fail for the same reason. So, if dma_set_mask_and_coherent() succeeds, 'pci_using_dac' is known to be 1. Simplify code and remove some dead code accordingly. [1]: https://lkml.org/lkml/2021/6/7/398 Signed-off-by: Christophe JAILLET Signed-off-by: Tony Nguyen