Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1605154 > unrolled thread

[net-next PATCH 0/2] NAPI ID fixups related to busy polling

Started byAlexander Duyck <alexander.duyck@gmail.com>
First post2017-03-20 22:50 +0100
Last post2017-03-22 19:30 +0100
Articles 2 — 2 participants

Back to article view | Back to linux.kernel


Contents

  [net-next PATCH 0/2] NAPI ID fixups related to busy polling Alexander Duyck <alexander.duyck@gmail.com> - 2017-03-20 22:50 +0100
    Re: [net-next PATCH 0/2] NAPI ID fixups related to busy polling David Miller <davem@davemloft.net> - 2017-03-22 19:30 +0100

#1605154 — [net-next PATCH 0/2] NAPI ID fixups related to busy polling

FromAlexander Duyck <alexander.duyck@gmail.com>
Date2017-03-20 22:50 +0100
Subject[net-next PATCH 0/2] NAPI ID fixups related to busy polling
Message-ID<tndlD-2eT-5@gated-at.bofh.it>
These two patches are a couple of minor clean-ups related to busy polling.
The first one addresses the fact that we were trying to busy poll on
sender_cpu values instead of true NAPI IDs.  The second addresses the fact
that there were a few paths where TCP sockets were being instanciated based
on a received patcket, but not recording the hash or NAPI ID of the packet
that was used to instanciate them.

---

Alexander Duyck (2):
      net: Busy polling should ignore sender CPUs
      tcp: Record Rx hash and NAPI ID in tcp_child_process


 include/net/busy_poll.h  |   11 +++++++++--
 net/core/dev.c           |    6 +++---
 net/ipv4/tcp_ipv4.c      |    2 --
 net/ipv4/tcp_minisocks.c |    5 +++++
 net/ipv6/tcp_ipv6.c      |    2 --
 5 files changed, 17 insertions(+), 9 deletions(-)

--

[toc] | [next] | [standalone]


#1606892

FromDavid Miller <davem@davemloft.net>
Date2017-03-22 19:30 +0100
Message-ID<tnTbc-6LY-27@gated-at.bofh.it>
In reply to#1605154
From: Alexander Duyck <alexander.duyck@gmail.com>
Date: Mon, 20 Mar 2017 14:48:41 -0700

> These two patches are a couple of minor clean-ups related to busy polling.
> The first one addresses the fact that we were trying to busy poll on
> sender_cpu values instead of true NAPI IDs.  The second addresses the fact
> that there were a few paths where TCP sockets were being instanciated based
> on a received patcket, but not recording the hash or NAPI ID of the packet
> that was used to instanciate them.

I'm expecting a respin of this.  Patch #1 appears to be 'net' material, and
Eric asked for some other minor tweaks as well.

Thanks.

[toc] | [prev] | [standalone]


Back to top | Article view | linux.kernel


csiph-web