tree 9d2349b63f328c9cb79fe6580215f823f8765b78
parent 0b93aed2842d950e8d2625e975e5a57febeff33d
author Eric Dumazet <edumazet@google.com> 1634218886 -0700
committer David S. Miller <davem@davemloft.net> 1634293714 +0100

tcp: switch orphan_count to bare per-cpu counters

Use of percpu_counter structure to track count of orphaned
sockets is causing problems on modern hosts with 256 cpus
or more.

Stefan Bach reported a serious spinlock contention in real workloads,
that I was able to reproduce with a netfilter rule dropping
incoming FIN packets.

    53.56%  server  [kernel.kallsyms]      [k] queued_spin_lock_slowpath
            |
            ---queued_spin_lock_slowpath
               |
                --53.51%--_raw_spin_lock_irqsave
                          |
                           --53.51%--__percpu_counter_sum
                                     tcp_check_oom
                                     |
                                     |--39.03%--__tcp_close
                                     |          tcp_close
                                     |          inet_release
                                     |          inet6_release
                                     |          sock_close
                                     |          __fput
                                     |          ____fput
                                     |          task_work_run
                                     |          exit_to_usermode_loop
                                     |          do_syscall_64
                                     |          entry_SYSCALL_64_after_hwframe
                                     |          __GI___libc_close
                                     |
                                      --14.48%--tcp_out_of_resources
                                                tcp_write_timeout
                                                tcp_retransmit_timer
                                                tcp_write_timer_handler
                                                tcp_write_timer
                                                call_timer_fn
                                                expire_timers
                                                __run_timers
                                                run_timer_softirq
                                                __softirqentry_text_start

As explained in commit cf86a086a180 ("net/dst: use a smaller percpu_counter
batch for dst entries accounting"), default batch size is too big
for the default value of tcp_max_orphans (262144).

But even if we reduce batch sizes, there would still be cases
where the estimated count of orphans is beyond the limit,
and where tcp_too_many_orphans() has to call the expensive
percpu_counter_sum_positive().

One solution is to use plain per-cpu counters, and have
a timer to periodically refresh this cache.

Updating this cache every 100ms seems about right, tcp pressure
state is not radically changing over shorter periods.

percpu_counter was nice 15 years ago while hosts had less
than 16 cpus, not anymore by current standards.

v2: Fix the build issue for CONFIG_CRYPTO_DEV_CHELSIO_TLS=m,
    reported by kernel test robot <lkp@intel.com>
    Remove unused socket argument from tcp_too_many_orphans()

Fixes: dd24c00191d5 ("net: Use a percpu_counter for orphan_count")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Stefan Bach <sfb@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>