On Wed, Feb 21, 2018 at 02:37:48PM -0500, David Miller wrote:

From: Eric Dumazet <edumazet@google.com>
Date: Mon, 19 Feb 2018 11:56:46 -0800

quoted

Switching TCP to GSO mode, relying on core networking layers
to perform eventual adaptation for dumb devices was overdue.

1) Most TCP developments are done with TSO in mind.
2) Less high-resolution timers needs to be armed for TCP-pacing
3) GSO can benefit of xmit_more hint
4) Receiver GRO is more effective (as if TSO was used for real on sender)
   -> less ACK packets and overhead.
5) Write queues have less overhead (one skb holds about 64KB of payload)
6) SACK coalescing just works. (no payload in skb->head)
7) rtx rb-tree contains less packets, SACK is cheaper.
8) Removal of legacy code. Less maintenance hassles.

Note that I have left the sendpage/zerocopy paths, but they probably can
benefit from the same strategy.

Thanks to Oleksandr Natalenko for reporting a performance issue for BBR/fq_codel,
which was the main reason I worked on this patch series.

Series applied, thanks Eric.

SCTP might want to do something similar, and if so we can get rid
of sk_can_gso() too.

Cc'ing linux-sctp and adding to the ToDo here, although it may be too
soon for SCTP. GSO support was added just a few months ago and
considering that it is not that much widely used as TCP, I fear we may
have some issues that didn't show up yet.

  Marcelo