kernel/linux.git/net/core/sysctl_net_core.c, branch v6.1.168

net: set the minimum for net_hotdata.netdev_budget_usecs

2025-03-07T15:56:43+00:00

[ Upstream commit c180188ec02281126045414e90d08422a80f75b4 ] Commit 7acf8a1e8a28 ("Replace 2 jiffies with sysctl netdev_budget_usecs to enable softirq tuning") added a possibility to set net_hotdata.netdev_budget_usecs, but added no lower bound checking. Commit a4837980fd9f ("net: revert default NAPI poll timeout to 2 jiffies") made the *initial* value HZ-dependent, so the initial value is at least 2 jiffies even for lower HZ values (2 ms for 1000 Hz, 8ms for 250 Hz, 20 ms for 100 Hz). But a user still can set improper values by a sysctl. Set .extra1 (the lower bound) for net_hotdata.netdev_budget_usecs to the same value as in the latter commit. That is to 2 jiffies. Fixes: a4837980fd9f ("net: revert default NAPI poll timeout to 2 jiffies") Fixes: 7acf8a1e8a28 ("Replace 2 jiffies with sysctl netdev_budget_usecs to enable softirq tuning") Signed-off-by: Jiri Slaby (SUSE) Cc: Dmitry Yakunin Cc: Konstantin Khlebnikov Link: https://patch.msgid.link/20250220110752.137639-1-jirislaby@kernel.org Signed-off-by: Jakub Kicinski Signed-off-by: Sasha Levin

net: let net.core.dev_weight always be non-zero

2025-02-21T12:49:05+00:00

[ Upstream commit d1f9f79fa2af8e3b45cffdeef66e05833480148a ] The following problem was encountered during stability test: (NULL net_device): NAPI poll function process_backlog+0x0/0x530 \ returned 1, exceeding its budget of 0. ------------[ cut here ]------------ list_add double add: new=ffff88905f746f48, prev=ffff88905f746f48, \ next=ffff88905f746e40. WARNING: CPU: 18 PID: 5462 at lib/list_debug.c:35 \ __list_add_valid_or_report+0xf3/0x130 CPU: 18 UID: 0 PID: 5462 Comm: ping Kdump: loaded Not tainted 6.13.0-rc7+ RIP: 0010:__list_add_valid_or_report+0xf3/0x130 Call Trace: ? __warn+0xcd/0x250 ? __list_add_valid_or_report+0xf3/0x130 enqueue_to_backlog+0x923/0x1070 netif_rx_internal+0x92/0x2b0 __netif_rx+0x15/0x170 loopback_xmit+0x2ef/0x450 dev_hard_start_xmit+0x103/0x490 __dev_queue_xmit+0xeac/0x1950 ip_finish_output2+0x6cc/0x1620 ip_output+0x161/0x270 ip_push_pending_frames+0x155/0x1a0 raw_sendmsg+0xe13/0x1550 __sys_sendto+0x3bf/0x4e0 __x64_sys_sendto+0xdc/0x1b0 do_syscall_64+0x5b/0x170 entry_SYSCALL_64_after_hwframe+0x76/0x7e The reproduction command is as follows: sysctl -w net.core.dev_weight=0 ping 127.0.0.1 This is because when the napi's weight is set to 0, process_backlog() may return 0 and clear the NAPI_STATE_SCHED bit of napi->state, causing this napi to be re-polled in net_rx_action() until __do_softirq() times out. Since the NAPI_STATE_SCHED bit has been cleared, napi_schedule_rps() can be retriggered in enqueue_to_backlog(), causing this issue. Making the napi's weight always non-zero solves this problem. Triggering this issue requires system-wide admin (setting is not namespaced). Fixes: e38766054509 ("[NET]: Fix sysctl net.core.dev_weight") Fixes: 3d48b53fb2ae ("net: dev_weight: TX/RX orthogonality") Signed-off-by: Liu Jian Link: https://patch.msgid.link/20250116143053.4146855-1-liujian56@huawei.com Signed-off-by: Jakub Kicinski Signed-off-by: Sasha Levin

net: make SK_MEMORY_PCPU_RESERV tunable

2024-05-02T14:29:24+00:00

[ Upstream commit 12a686c2e761f1f1f6e6e2117a9ab9c6de2ac8a7 ] This patch adds /proc/sys/net/core/mem_pcpu_rsv sysctl file, to make SK_MEMORY_PCPU_RESERV tunable. Commit 3cd3399dd7a8 ("net: implement per-cpu reserves for memory_allocated") introduced per-cpu forward alloc cache: "Implement a per-cpu cache of +1/-1 MB, to reduce number of changes to sk->sk_prot->memory_allocated, which would otherwise be cause of false sharing." sk_prot->memory_allocated points to global atomic variable: atomic_long_t tcp_memory_allocated ____cacheline_aligned_in_smp; If increasing the per-cpu cache size from 1MB to e.g. 16MB, changes to sk->sk_prot->memory_allocated can be further reduced. Performance may be improved on system with many cores. Signed-off-by: Adam Li Reviewed-by: Christoph Lameter (Ampere) Signed-off-by: David S. Miller Stable-dep-of: 3584718cf2ec ("net: fix sk_memory_allocated_{add|sub} vs softirqs") Signed-off-by: Sasha Levin

net: sysctl: remove unused variable long_max

2022-09-07T14:31:19+00:00

The variable long_max is replaced by bpf_jit_limit_max and no longer be used. So remove it. No functional change. Signed-off-by: Liu Shixin Signed-off-by: David S. Miller

net: Fix data-races around weight_p and dev_weight_[rt]x_bias.

2022-08-24T12:46:57+00:00

While reading weight_p, it can be changed concurrently. Thus, we need to add READ_ONCE() to its reader. Also, dev_[rt]x_weight can be read/written at the same time. So, we need to use READ_ONCE() and WRITE_ONCE() for its access. Moreover, to use the same weight_p while changing dev_[rt]x_weight, we add a mutex in proc_do_dev_weight(). Fixes: 3d48b53fb2ae ("net: dev_weight: TX/RX orthogonality") Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Kuniyuki Iwashima Signed-off-by: David S. Miller

Merge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next

2022-05-23T23:07:14+00:00

Daniel Borkmann says: ==================== pull-request: bpf-next 2022-05-23 We've added 113 non-merge commits during the last 26 day(s) which contain a total of 121 files changed, 7425 insertions(+), 1586 deletions(-). The main changes are: 1) Speed up symbol resolution for kprobes multi-link attachments, from Jiri Olsa. 2) Add BPF dynamic pointer infrastructure e.g. to allow for dynamically sized ringbuf reservations without extra memory copies, from Joanne Koong. 3) Big batch of libbpf improvements towards libbpf 1.0 release, from Andrii Nakryiko. 4) Add BPF link iterator to traverse links via seq_file ops, from Dmitrii Dolgov. 5) Add source IP address to BPF tunnel key infrastructure, from Kaixi Fan. 6) Refine unprivileged BPF to disable only object-creating commands, from Alan Maguire. 7) Fix JIT blinding of ld_imm64 when they point to subprogs, from Alexei Starovoitov. 8) Add BPF access to mptcp_sock structures and their meta data, from Geliang Tang. 9) Add new BPF helper for access to remote CPU's BPF map elements, from Feng Zhou. 10) Allow attaching 64-bit cookie to BPF link of fentry/fexit/fmod_ret, from Kui-Feng Lee. 11) Follow-ups to typed pointer support in BPF maps, from Kumar Kartikeya Dwivedi. 12) Add busy-poll test cases to the XSK selftest suite, from Magnus Karlsson. 13) Improvements in BPF selftest test_progs subtest output, from Mykola Lysenko. 14) Fill bpf_prog_pack allocator areas with illegal instructions, from Song Liu. 15) Add generic batch operations for BPF map-in-map cases, from Takshak Chahande. 16) Make bpf_jit_enable more user friendly when permanently on 1, from Tiezhu Yang. 17) Fix an array overflow in bpf_trampoline_get_progs(), from Yuntao Wang. ==================== Link: https://lore.kernel.org/r/20220523223805.27931-1-daniel@iogearbox.net Signed-off-by: Jakub Kicinski

net: add skb_defer_max sysctl

2022-05-16T10:33:59+00:00

commit 68822bdf76f1 ("net: generalize skb freeing deferral to per-cpu lists") added another per-cpu cache of skbs. It was expected to be small, and an IPI was forced whenever the list reached 128 skbs. We might need to be able to control more precisely queue capacity and added latency. An IPI is generated whenever queue reaches half capacity. Default value of the new limit is 64. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller

bpf: Print some info if disable bpf_jit_enable failed

2022-05-10T17:13:06+00:00

A user told me that bpf_jit_enable can be disabled on one system, but he failed to disable bpf_jit_enable on the other system: # echo 0 > /proc/sys/net/core/bpf_jit_enable bash: echo: write error: Invalid argument No useful info is available through the dmesg log, a quick analysis shows that the issue is related with CONFIG_BPF_JIT_ALWAYS_ON. When CONFIG_BPF_JIT_ALWAYS_ON is enabled, bpf_jit_enable is permanently set to 1 and setting any other value than that will return failure. It is better to print some info to tell the user if disable bpf_jit_enable failed. Signed-off-by: Tiezhu Yang Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/1652153703-22729-3-git-send-email-yangtiezhu@loongson.cn

net: sysctl: Use SYSCTL_TWO instead of &two

2022-05-10T17:13:06+00:00

It is better to use SYSCTL_TWO instead of &two, and then we can remove the variable "two" in net/core/sysctl_net_core.c. Signed-off-by: Tiezhu Yang Signed-off-by: Alexei Starovoitov Link: https://lore.kernel.org/bpf/1652153703-22729-2-git-send-email-yangtiezhu@loongson.cn

net: sysctl: introduce sysctl SYSCTL_THREE

2022-05-03T08:15:06+00:00

This patch introdues the SYSCTL_THREE. KUnit: [00:10:14] ================ sysctl_test (10 subtests) ================= [00:10:14] [PASSED] sysctl_test_api_dointvec_null_tbl_data [00:10:14] [PASSED] sysctl_test_api_dointvec_table_maxlen_unset [00:10:14] [PASSED] sysctl_test_api_dointvec_table_len_is_zero [00:10:14] [PASSED] sysctl_test_api_dointvec_table_read_but_position_set [00:10:14] [PASSED] sysctl_test_dointvec_read_happy_single_positive [00:10:14] [PASSED] sysctl_test_dointvec_read_happy_single_negative [00:10:14] [PASSED] sysctl_test_dointvec_write_happy_single_positive [00:10:14] [PASSED] sysctl_test_dointvec_write_happy_single_negative [00:10:14] [PASSED] sysctl_test_api_dointvec_write_single_less_int_min [00:10:14] [PASSED] sysctl_test_api_dointvec_write_single_greater_int_max [00:10:14] =================== [PASSED] sysctl_test =================== ./run_kselftest.sh -c sysctl ... ok 1 selftests: sysctl: sysctl.sh Cc: Luis Chamberlain Cc: Kees Cook Cc: Iurii Zaikin Cc: "David S. Miller" Cc: Jakub Kicinski Cc: Paolo Abeni Cc: Hideaki YOSHIFUJI Cc: David Ahern Cc: Simon Horman Cc: Julian Anastasov Cc: Pablo Neira Ayuso Cc: Jozsef Kadlecsik Cc: Florian Westphal Cc: Shuah Khan Cc: Andrew Morton Cc: Alexei Starovoitov Cc: Eric Dumazet Cc: Lorenz Bauer Cc: Akhmat Karakotov Signed-off-by: Tonghao Zhang Reviewed-by: Simon Horman Signed-off-by: Paolo Abeni