netdev - Re: [PATCH net-next 2/2] tcp/dccp: change source port selection at connect() time

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAL+tcoBKjW7WFgf_LEee6r_VnVosgSMMXpH=Y4fxNK3XA6BWTQ@mail.gmail.com>
Date: Fri, 15 Dec 2023 10:26:15 +0800
From: Jason Xing <kerneljasonxing@...il.com>
To: Eric Dumazet <edumazet@...gle.com>
Cc: "David S . Miller" <davem@...emloft.net>, Jakub Kicinski <kuba@...nel.org>, 
	Paolo Abeni <pabeni@...hat.com>, Jakub Sitnicki <jakub@...udflare.com>, netdev@...r.kernel.org, 
	eric.dumazet@...il.com
Subject: Re: [PATCH net-next 2/2] tcp/dccp: change source port selection at
 connect() time

On Fri, Dec 15, 2023 at 3:30 AM Eric Dumazet <edumazet@...gle.com> wrote:
>
> In commit 1580ab63fc9a ("tcp/dccp: better use of ephemeral ports in connect()")
> we added an heuristic to select even ports for connect() and odd ports for bind().
>
> This was nice because no applications changes were needed.
>
[...]
> But it added more costs when all even ports are in use,
> when there are few listeners and many active connections.

Yes, I have encountered this issue several times. So internally adding
a switch to decide which selecting port algorithm the connect() phase
should use can address this issue: go back to the original algo
(without splitting ports range) many years ago.

>
> Since then, IP_LOCAL_PORT_RANGE has been added to permit an application
> to partition ephemeral port range at will.
>
> This patch extends the idea so that if IP_LOCAL_PORT_RANGE is set on
> a socket before accept(), port selection no longer favors even ports.
>
> This means that connect() can find a suitable source port faster,
> and applications can use a different split between connect() and bind()
> users.

Great :)

>
> This should give more entropy to Toeplitz hash used in RSS: Using even
> ports was wasting one bit from the 16bit sport.
>
> A similar change can be done in inet_csk_find_open_port() if needed.
>
> Signed-off-by: Eric Dumazet <edumazet@...gle.com>
> Cc: Jakub Sitnicki <jakub@...udflare.com>

Reviewed-by: Jason Xing <kerneljasonxing@...il.com>

Thanks!

> ---
>  net/ipv4/inet_hashtables.c | 27 ++++++++++++++++-----------
>  1 file changed, 16 insertions(+), 11 deletions(-)
>
> diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c
> index a532f749e47781cc951f2003f621cec4387a2384..9ff201bc4e6d2da04735e8c160d446602e0adde1 100644
> --- a/net/ipv4/inet_hashtables.c
> +++ b/net/ipv4/inet_hashtables.c
> @@ -1012,7 +1012,8 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
>         bool tb_created = false;
>         u32 remaining, offset;
>         int ret, i, low, high;
> -       int l3mdev;
> +       bool local_ports;
> +       int step, l3mdev;
>         u32 index;
>
>         if (port) {
> @@ -1024,10 +1025,12 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
>
>         l3mdev = inet_sk_bound_l3mdev(sk);
>
> -       inet_sk_get_local_port_range(sk, &low, &high);
> +       local_ports = inet_sk_get_local_port_range(sk, &low, &high);
> +       step = local_ports ? 1 : 2;
> +
>         high++; /* [32768, 60999] -> [32768, 61000[ */
>         remaining = high - low;
> -       if (likely(remaining > 1))
> +       if (!local_ports && remaining > 1)
>                 remaining &= ~1U;
>
>         get_random_sleepable_once(table_perturb,
> @@ -1040,10 +1043,11 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
>         /* In first pass we try ports of @low parity.
>          * inet_csk_get_port() does the opposite choice.
>          */
> -       offset &= ~1U;
> +       if (!local_ports)
> +               offset &= ~1U;
>  other_parity_scan:
>         port = low + offset;
> -       for (i = 0; i < remaining; i += 2, port += 2) {
> +       for (i = 0; i < remaining; i += step, port += step) {
>                 if (unlikely(port >= high))
>                         port -= remaining;
>                 if (inet_is_local_reserved_port(net, port))
> @@ -1083,10 +1087,11 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
>                 cond_resched();
>         }
>
> -       offset++;
> -       if ((offset & 1) && remaining > 1)
> -               goto other_parity_scan;
> -
> +       if (!local_ports) {
> +               offset++;
> +               if ((offset & 1) && remaining > 1)
> +                       goto other_parity_scan;
> +       }
>         return -EADDRNOTAVAIL;
>
>  ok:
> @@ -1109,8 +1114,8 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row,
>          * on low contention the randomness is maximal and on high contention
>          * it may be inexistent.
>          */
> -       i = max_t(int, i, get_random_u32_below(8) * 2);
> -       WRITE_ONCE(table_perturb[index], READ_ONCE(table_perturb[index]) + i + 2);
> +       i = max_t(int, i, get_random_u32_below(8) * step);
> +       WRITE_ONCE(table_perturb[index], READ_ONCE(table_perturb[index]) + i + step);
>
>         /* Head lock still held and bh's disabled */
>         inet_bind_hash(sk, tb, tb2, port);
> --
> 2.43.0.472.g3155946c3a-goog
>
>