Content-Type: multipart/mixed; boundary="===============3892372081293502453==" MIME-Version: 1.0 From: Florian Westphal To: mptcp at lists.01.org Subject: Re: [MPTCP] [PATCH RFC 0/6] mptcp_poll rework preview Date: Tue, 17 Sep 2019 08:42:07 +0200 Message-ID: <20190917064207.GB6961@breakpoint.cc> In-Reply-To: 90fe1f1df1eb0d88beaae8e9305ee3db27b53428.camel@redhat.com X-Status: X-Keywords: X-UID: 1858 --===============3892372081293502453== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Paolo Abeni wrote: > On Mon, 2019-09-09 at 17:53 +0200, Florian Westphal wrote: > > This patch series is in need of a bit more work, f.e. there is too much > > copypastry in patch 3. > > = > > However, it is functional as-is so I am sharing it to get earlier revie= ws. > > = > > Problem with current mptcp_poll() is that we can't block the task on ea= ch > > subflow: subflow we're polling on might be idle while another one has d= ata > > ready for processing. > = > If I read the series correctly, patch 3 is not strictly needed for the > above goal: AFAICS it should help to avoid empty/blocking recvmsg() > after successful poll(), right? Thats right. EPOLLIN means next recv won't block, which isn't the case at this time because we do mptcp processing in recv only. > I think we should rework a bit recvmsg() before going on with this - > see "mptcp: partial recvmsg() refactor". Agree. > Perhaps even something more > drastic e.g. subflow could try move skbs from ssk->sk_receive_queue to > a per msk queue, ordered by mptcp seq in subflow_data_ready(). WDYT? Yes, that might be a good idea/needed. We shouldn't depend on regular recv calls for mptcp processing. --===============3892372081293502453==--