From: Michael Chan <michael.chan@broadcom.com>
To: Song Liu <songliubraving@fb.com>
Cc: Eric Dumazet <edumazet@google.com>,
David Miller <davem@davemloft.net>,
Netdev <netdev@vger.kernel.org>,
Ariel Elior <ariel.elior@cavium.com>,
Eric Dumazet <eric.dumazet@gmail.com>,
tariqt@mellanox.com, Saeed Mahameed <saeedm@mellanox.com>,
jeffrey.t.kirsher@intel.com,
Jakub Kicinski <jakub.kicinski@netronome.com>,
j.vosburgh@gmail.com, vfalico@gmail.com,
Andy Gospodarek <andy@greyhouse.net>
Subject: Re: [PATCH net 00/15] netpoll: avoid capture effects for NAPI drivers
Date: Tue, 25 Sep 2018 07:02:28 -0700 [thread overview]
Message-ID: <CACKFLinRzz8h+Q12uTU6cORf9f8-hRTTAT2vi+GQskPM9MT55A@mail.gmail.com> (raw)
In-Reply-To: <ACD08072-B5A0-4A5D-8E02-30E5F763124D@fb.com>
On Mon, Sep 24, 2018 at 2:18 PM Song Liu <songliubraving@fb.com> wrote:
>
>
>
> > On Sep 24, 2018, at 2:05 PM, Eric Dumazet <edumazet@google.com> wrote:
> >
> >>
> >> Interesting, maybe a bnxt specific issue.
> >>
> >> It seems their model is to process TX/RX notification in the same queue,
> >> they throw away RX events if budget == 0
> >>
> >> It means commit e7b9569102995ebc26821789628eef45bd9840d8 is wrong and
> >> must be reverted.
> >>
> >> Otherwise, we have a possibility of blocking a queue under netpoll pressure.
> >
> > Hmm, actually a revert might not be enough, since code at lines 2030-2031
> > would fire and we might not call napi_complete_done() anyway.
> >
> > Unfortunately this driver logic is quite complex.
> >
> > Could you test on other NIC eventually ?
> >
>
> It actually runs OK on ixgbe.
>
> @Michael, could you please help us with this?
>
I've taken a quick look using today's net tree plus Eric's
poll_one_napi() patch. The problem I'm seeing is that netpoll calls
bnxt_poll() with budget 0. And since work_done >= budget of 0, we
return without calling napi_complete_done() and without arming the
interrupt. netpoll doesn't always call us back until we call
napi_complete_done(), right? So I think if there are in-flight TX
completions, we'll miss those.
next prev parent reply other threads:[~2018-09-25 20:10 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-21 22:27 [PATCH net 00/15] netpoll: avoid capture effects for NAPI drivers Eric Dumazet
2018-09-21 22:27 ` [PATCH net 01/15] netpoll: make ndo_poll_controller() optional Eric Dumazet
2018-09-21 22:27 ` [PATCH net 02/15] bonding: use netpoll_poll_dev() helper Eric Dumazet
2018-09-22 10:23 ` Jay Vosburgh
2018-09-21 22:27 ` [PATCH net 03/15] ixgbe: remove ndo_poll_controller Eric Dumazet
2018-09-21 22:27 ` [PATCH net 04/15] ixgbevf: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 05/15] fm10k: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 06/15] ixgb: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 07/15] igb: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 08/15] ice: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 09/15] i40evf: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 10/15] mlx4: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 11/15] mlx5: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 12/15] bnx2x: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 13/15] bnxt: " Eric Dumazet
2018-09-21 22:27 ` [PATCH net 14/15] nfp: " Eric Dumazet
2018-09-21 23:41 ` Jakub Kicinski
2018-09-21 22:27 ` [PATCH net 15/15] tun: " Eric Dumazet
2018-09-23 19:29 ` [PATCH net 00/15] netpoll: avoid capture effects for NAPI drivers David Miller
2018-09-23 19:47 ` Eric Dumazet
2018-09-24 5:04 ` David Miller
2018-09-24 15:30 ` Eric Dumazet
2018-09-24 19:30 ` Song Liu
2018-09-24 19:41 ` Eric Dumazet
2018-09-24 20:00 ` Song Liu
2018-09-24 20:56 ` Eric Dumazet
2018-09-24 21:05 ` Eric Dumazet
2018-09-24 21:18 ` Song Liu
2018-09-25 14:02 ` Michael Chan [this message]
2018-09-25 14:20 ` Eric Dumazet
2018-09-25 14:43 ` Michael Chan
2018-09-25 18:25 ` Song Liu
2018-09-25 21:09 ` Michael Chan
2018-09-25 23:36 ` Song Liu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CACKFLinRzz8h+Q12uTU6cORf9f8-hRTTAT2vi+GQskPM9MT55A@mail.gmail.com \
--to=michael.chan@broadcom.com \
--cc=andy@greyhouse.net \
--cc=ariel.elior@cavium.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=eric.dumazet@gmail.com \
--cc=j.vosburgh@gmail.com \
--cc=jakub.kicinski@netronome.com \
--cc=jeffrey.t.kirsher@intel.com \
--cc=netdev@vger.kernel.org \
--cc=saeedm@mellanox.com \
--cc=songliubraving@fb.com \
--cc=tariqt@mellanox.com \
--cc=vfalico@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).