[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <6178131635190015@myt6-af0b0b987ed8.qloud-c.yandex.net>
Date: Mon, 25 Oct 2021 22:26:55 +0300
From: Alexander Azimov <mitradir@...dex-team.ru>
To: Eric Dumazet <eric.dumazet@...il.com>
Cc: zeil@...dex-team.ru, Lawrence Brakmo <brakmo@...com>,
Akhmat Karakotov <hmukos@...dex-team.ru>,
netdev@...r.kernel.org, Neal Cardwell <ncardwell@...gle.com>,
Yuchung Cheng <ycheng@...gle.com>
Subject: Re:[PATCH] tcp: Use BPF timeout setting for SYN ACK RTO
Hi Eric,
Can you please clarify why do you think that SYN RTO should be accessible through BPF and SYN ACK RTO should be bound to TCP_TIMEOUT_INIT constant?
I can't see the reason for such asymmetry, please advise.
I also wonder what kind of existing BPF programs can suffer from these changes. Please give us your insights.
ps: this is a copy of the original message, hope this one will come in a plain text
Powered by blists - more mailing lists