kernel.git/net/core/net-sysfs.c, branch linux-6.10.y

net: napi: Prevent overflow of napi_defer_hard_irqs

2024-10-10T10:00:22Z

[ Upstream commit 08062af0a52107a243f7608fd972edb54ca5b7f8 ] In commit 6f8b12d661d0 ("net: napi: add hard irqs deferral feature") napi_defer_irqs was added to net_device and napi_defer_irqs_count was added to napi_struct, both as type int. This value never goes below zero, so there is not reason for it to be a signed int. Change the type for both from int to u32, and add an overflow check to sysfs to limit the value to S32_MAX. The limit of S32_MAX was chosen because the practical limit before this patch was S32_MAX (anything larger was an overflow) and thus there are no behavioral changes introduced. If the extra bit is needed in the future, the limit can be raised. Before this patch: $ sudo bash -c 'echo 2147483649 > /sys/class/net/eth4/napi_defer_hard_irqs' $ cat /sys/class/net/eth4/napi_defer_hard_irqs -2147483647 After this patch: $ sudo bash -c 'echo 2147483649 > /sys/class/net/eth4/napi_defer_hard_irqs' bash: line 0: echo: write error: Numerical result out of range Similarly, /sys/class/net/XXXXX/tx_queue_len is defined as unsigned: include/linux/netdevice.h: unsigned int tx_queue_len; And has an overflow check: dev_change_tx_queue_len(..., unsigned long new_len): if (new_len != (unsigned int)new_len) return -ERANGE; Suggested-by: Jakub Kicinski Signed-off-by: Joe Damato Reviewed-by: Eric Dumazet Link: https://patch.msgid.link/20240904153431.307932-1-jdamato@fastly.com Signed-off-by: Jakub Kicinski Signed-off-by: Sasha Levin

net: dqs: Do not use extern for unused dql_group

2024-09-12T09:12:57Z

[ Upstream commit 77461c10819103eaee7b33c744174b32a8c78b40 ] When CONFIG_DQL is not enabled, dql_group should be treated as a dead declaration. However, its current extern declaration assumes the linker will ignore it, which is generally true across most compiler and architecture combinations. But in certain cases, the linker still attempts to resolve the extern struct, even when the associated code is dead, resulting in a linking error. For instance the following error in loongarch64: >> loongarch64-linux-ld: net-sysfs.c:(.text+0x589c): undefined reference to `dql_group' Modify the declaration of the dead object to be an empty declaration instead of an extern. This change will prevent the linker from attempting to resolve an undefined reference. Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/202409012047.eCaOdfQJ-lkp@intel.com/ Fixes: 74293ea1c4db ("net: sysfs: Do not create sysfs for non BQL device") Signed-off-by: Breno Leitao Reviewed-by: Simon Horman Tested-by: Simon Horman # build-tested Link: https://patch.msgid.link/20240902101734.3260455-1-leitao@debian.org Signed-off-by: Jakub Kicinski Signed-off-by: Sasha Levin

ethtool: check device is present when getting link settings

2024-09-04T11:30:09Z

[ Upstream commit a699781c79ecf6cfe67fb00a0331b4088c7c8466 ] A sysfs reader can race with a device reset or removal, attempting to read device state when the device is not actually present. eg: [exception RIP: qed_get_current_link+17] #8 [ffffb9e4f2907c48] qede_get_link_ksettings at ffffffffc07a994a [qede] #9 [ffffb9e4f2907cd8] __rh_call_get_link_ksettings at ffffffff992b01a3 #10 [ffffb9e4f2907d38] __ethtool_get_link_ksettings at ffffffff992b04e4 #11 [ffffb9e4f2907d90] duplex_show at ffffffff99260300 #12 [ffffb9e4f2907e38] dev_attr_show at ffffffff9905a01c #13 [ffffb9e4f2907e50] sysfs_kf_seq_show at ffffffff98e0145b #14 [ffffb9e4f2907e68] seq_read at ffffffff98d902e3 #15 [ffffb9e4f2907ec8] vfs_read at ffffffff98d657d1 #16 [ffffb9e4f2907f00] ksys_read at ffffffff98d65c3f #17 [ffffb9e4f2907f38] do_syscall_64 at ffffffff98a052fb crash> struct net_device.state ffff9a9d21336000 state = 5, state 5 is __LINK_STATE_START (0b1) and __LINK_STATE_NOCARRIER (0b100). The device is not present, note lack of __LINK_STATE_PRESENT (0b10). This is the same sort of panic as observed in commit 4224cfd7fb65 ("net-sysfs: add check for netdevice being present to speed_show"). There are many other callers of __ethtool_get_link_ksettings() which don't have a device presence check. Move this check into ethtool to protect all callers. Fixes: d519e17e2d01 ("net: export device speed and duplex via sysfs") Fixes: 4224cfd7fb65 ("net-sysfs: add check for netdevice being present to speed_show") Signed-off-by: Jamie Bainbridge Link: https://patch.msgid.link/8bae218864beaa44ed01628140475b9bf641c5b0.1724393671.git.jamie.bainbridge@gmail.com Signed-off-by: Jakub Kicinski Signed-off-by: Sasha Levin

net: no longer acquire RTNL in threaded_show()

2024-05-03T22:14:01Z

dev->threaded can be read locklessly, if we add corresponding READ_ONCE()/WRITE_ONCE() annotations. Signed-off-by: Eric Dumazet Link: https://lore.kernel.org/r/20240502173926.2010646-1-edumazet@google.com Signed-off-by: Jakub Kicinski

net: constify net_class

2024-04-15T09:46:01Z

AFAICS all users of net_class take a const struct class * argument. Therefore fully constify net_class. Signed-off-by: Heiner Kallweit Acked-by: Stephen Hemminger Signed-off-by: David S. Miller

net: dqs: use sysfs_emit() in favor of sprintf()

2024-04-08T10:04:08Z

Commit 6025b9135f7a ("net: dqs: add NIC stall detector based on BQL") added three sysfs files. Use the recommended sysfs_emit() helper. Signed-off-by: Eric Dumazet Cc: Jakub Kicinski Cc: Breno Leitao Signed-off-by: David S. Miller

net: dqs: add NIC stall detector based on BQL

2024-03-08T10:23:26Z

softnet_data->time_squeeze is sometimes used as a proxy for host overload or indication of scheduling problems. In practice this statistic is very noisy and has hard to grasp units - e.g. is 10 squeezes a second to be expected, or high? Delaying network (NAPI) processing leads to drops on NIC queues but also RTT bloat, impacting pacing and CA decisions. Stalls are a little hard to detect on the Rx side, because there may simply have not been any packets received in given period of time. Packet timestamps help a little bit, but again we don't know if packets are stale because we're not keeping up or because someone (*cough* cgroups) disabled IRQs for a long time. We can, however, use Tx as a proxy for Rx stalls. Most drivers use combined Rx+Tx NAPIs so if Tx gets starved so will Rx. On the Tx side we know exactly when packets get queued, and completed, so there is no uncertainty. This patch adds stall checks to BQL. Why BQL? Because it's a convenient place to add such checks, already called by most drivers, and it has copious free space in its structures (this patch adds no extra cache references or dirtying to the fast path). The algorithm takes one parameter - max delay AKA stall threshold and increments a counter whenever NAPI got delayed for at least that amount of time. It also records the length of the longest stall. To be precise every time NAPI has not polled for at least stall thrs we check if there were any Tx packets queued between last NAPI run and now - stall_thrs/2. Unlike the classic Tx watchdog this mechanism does not ignore stalls caused by Tx being disabled, or loss of link. I don't think the check is worth the complexity, and stall is a stall, whether due to host overload, flow control, link down... doesn't matter much to the application. We have been running this detector in production at Meta for 2 years, with the threshold of 8ms. It's the lowest value where false positives become rare. There's still a constant stream of reported stalls (especially without the ksoftirqd deferral patches reverted), those who like their stall metrics to be 0 may prefer higher value. Signed-off-by: Jakub Kicinski Signed-off-by: Breno Leitao Signed-off-by: David S. Miller

net: introduce include/net/rps.h

2024-03-08T05:12:43Z

Move RPS related structures and helpers from include/linux/netdevice.h and include/net/sock.h to a new include file. Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Reviewed-by: David Ahern Link: https://lore.kernel.org/r/20240306160031.874438-18-edumazet@google.com Signed-off-by: Jakub Kicinski

net: sysfs: Do not create sysfs for non BQL device

2024-02-19T20:30:44Z

Creation of sysfs entries is expensive, mainly for workloads that constantly creates netdev and netns often. Do not create BQL sysfs entries for devices that don't need, basically those that do not have a real queue, i.e, devices that has NETIF_F_LLTX and IFF_NO_QUEUE, such as `lo` interface. This will remove the /sys/class/net/eth0/queues/tx-X/byte_queue_limits/ directory for these devices. In the example below, eth0 has the `byte_queue_limits` directory but not `lo`. # ls /sys/class/net/lo/queues/tx-0/ traffic_class tx_maxrate tx_timeout xps_cpus xps_rxqs # ls /sys/class/net/eth0/queues/tx-0/byte_queue_limits/ hold_time inflight limit limit_max limit_min This also removes the #ifdefs, since we can also use netdev_uses_bql() to check if the config is enabled. (as suggested by Jakub). Suggested-by: Eric Dumazet Signed-off-by: Breno Leitao Link: https://lore.kernel.org/r/20240216094154.3263843-1-leitao@debian.org Signed-off-by: Jakub Kicinski

net-sysfs: convert netstat_show() to RCU

2024-02-14T11:20:13Z

dev_get_stats() can be called from RCU, there is no need to acquire dev_base_lock. Change dev_isalive() comment to reflect we no longer use dev_base_lock from net/core/net-sysfs.c Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller