From: Alexander Duyck <alexander.duyck@gmail.com>
To: Eric Dumazet <edumazet@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>,
"David S . Miller" <davem@davemloft.net>,
Jakub Kicinski <kuba@kernel.org>, netdev <netdev@vger.kernel.org>,
Coco Li <lixiaoyan@google.com>
Subject: Re: [PATCH net-next 09/15] net: increase MAX_SKB_FRAGS
Date: Thu, 3 Feb 2022 09:56:42 -0800 [thread overview]
Message-ID: <CAKgT0UeTvj_6DWUskxxaRiQQxcwg6j0u+UHDaougJSMdkogKWA@mail.gmail.com> (raw)
In-Reply-To: <CANn89iKxGvbXQqoRZZ5j22-5YkpiCLS13EGoQ1OYe3EHjEss6A@mail.gmail.com>
On Thu, Feb 3, 2022 at 9:34 AM Eric Dumazet <edumazet@google.com> wrote:
>
> On Thu, Feb 3, 2022 at 9:26 AM Alexander H Duyck
> <alexander.duyck@gmail.com> wrote:
> >
> > On Wed, 2022-02-02 at 17:51 -0800, Eric Dumazet wrote:
> > > From: Eric Dumazet <edumazet@google.com>
> > >
> > > Currently, MAX_SKB_FRAGS value is 17.
> > >
> > > For standard tcp sendmsg() traffic, no big deal because tcp_sendmsg()
> > > attempts order-3 allocations, stuffing 32768 bytes per frag.
> > >
> > > But with zero copy, we use order-0 pages.
> > >
> > > For BIG TCP to show its full potential, we increase MAX_SKB_FRAGS
> > > to be able to fit 45 segments per skb.
> > >
> > > This is also needed for BIG TCP rx zerocopy, as zerocopy currently
> > > does not support skbs with frag list.
> > >
> > > We have used this MAX_SKB_FRAGS value for years at Google before
> > > we deployed 4K MTU, with no adverse effect.
> > > Back then, goal was to be able to receive full size (64KB) GRO
> > > packets without the frag_list overhead.
> > >
> > > Signed-off-by: Eric Dumazet <edumazet@google.com>
> >
> > So a big issue I see with this patch is the potential queueing issues
> > it may introduce on Tx queues. I suspect it will cause a number of
> > performance regressions and deadlocks as it will change the Tx queueing
> > behavior for many NICs.
> >
> > As I recall many of the Intel drivers are using MAX_SKB_FRAGS as one of
> > the ingredients for DESC_NEEDED in order to determine if the Tx queue
> > needs to stop. With this change the value for igb for instance is
> > jumping from 21 to 49, and the wake threshold is twice that, 98. As
> > such the minimum Tx descriptor threshold for the driver would need to
> > be updated beyond 80 otherwise it is likely to deadlock the first time
> > it has to pause.
>
> Are these limits hard coded in Intel drivers and firmware, or do you
> think this can be changed ?
This is all code in the drivers. Most drivers have them as the logic
is used to avoid having to return NETIDEV_TX_BUSY. Basically the
assumption is there is a 1:1 correlation between descriptors and
individual frags. So most drivers would need to increase the size of
their Tx descriptor rings if they were optimized for a lower value.
The other thing is that most of the tuning for things like interrupt
moderation assume a certain fill level on the queues and those would
likely need to be updated to account for this change.
> I could make MAX_SKB_FRAGS a config option, and default to 17, until
> all drivers have been fixed.
>
> Alternative is that I remove this patch from the series and we apply
> it to Google production kernels,
> as we did before.
A config option would probably be preferred. The big issue as I see it
is that changing MAX_SKB_FRAGS is going to have ripples throughout the
ecosystem as the shared info size will be increasing and the queueing
behavior for most drivers will be modified as a result.
next prev parent reply other threads:[~2022-02-03 17:57 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-03 1:51 [PATCH net-next 00/15] tcp: BIG TCP implementation Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 01/15] net: add netdev->tso_ipv6_max_size attribute Eric Dumazet
2022-02-03 16:34 ` Jakub Kicinski
2022-02-03 16:56 ` Eric Dumazet
2022-02-03 18:58 ` Jakub Kicinski
2022-02-03 19:12 ` Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 02/15] ipv6: add dev->gso_ipv6_max_size Eric Dumazet
2022-02-03 8:57 ` Paolo Abeni
2022-02-03 15:34 ` Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 03/15] tcp_cubic: make hystart_ack_delay() aware of BIG TCP Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 04/15] ipv6: add struct hop_jumbo_hdr definition Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 05/15] ipv6/gso: remove temporary HBH/jumbo header Eric Dumazet
2022-02-03 18:53 ` Alexander H Duyck
2022-02-03 19:17 ` Eric Dumazet
2022-02-03 19:45 ` Alexander Duyck
2022-02-03 19:59 ` Eric Dumazet
2022-02-03 21:08 ` Alexander H Duyck
2022-02-03 21:41 ` Eric Dumazet
2022-02-04 0:05 ` Alexander Duyck
2022-02-04 0:27 ` Eric Dumazet
2022-02-04 1:14 ` Eric Dumazet
2022-02-04 1:48 ` Eric Dumazet
2022-02-04 2:15 ` Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 06/15] ipv6/gro: insert " Eric Dumazet
2022-02-03 9:19 ` Paolo Abeni
2022-02-03 15:48 ` Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 07/15] ipv6: add GRO_IPV6_MAX_SIZE Eric Dumazet
2022-02-03 2:18 ` Eric Dumazet
2022-02-03 10:44 ` Paolo Abeni
2022-02-03 1:51 ` [PATCH net-next 08/15] ipv6: Add hop-by-hop header to jumbograms in ip6_output Eric Dumazet
2022-02-03 9:07 ` Paolo Abeni
2022-02-03 16:31 ` Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 09/15] net: increase MAX_SKB_FRAGS Eric Dumazet
2022-02-03 5:02 ` kernel test robot
2022-02-03 5:20 ` Eric Dumazet
2022-02-03 5:31 ` Jakub Kicinski
2022-02-03 6:35 ` Eric Dumazet
2022-02-03 5:23 ` kernel test robot
2022-02-03 5:43 ` kernel test robot
2022-02-03 16:01 ` Paolo Abeni
2022-02-03 17:26 ` Alexander H Duyck
2022-02-03 17:34 ` Eric Dumazet
2022-02-03 17:56 ` Alexander Duyck [this message]
2022-02-03 19:18 ` Jakub Kicinski
2022-02-03 19:20 ` Eric Dumazet
2022-02-03 19:54 ` Eric Dumazet
2022-02-04 10:18 ` David Laight
2022-02-04 15:46 ` Alexander Duyck
2022-02-03 1:51 ` [PATCH net-next 10/15] net: loopback: enable BIG TCP packets Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 11/15] bonding: update dev->tso_ipv6_max_size Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 12/15] macvlan: enable BIG TCP Packets Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 13/15] ipvlan: " Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 14/15] mlx4: support BIG TCP packets Eric Dumazet
2022-02-03 13:04 ` Tariq Toukan
2022-02-03 15:54 ` Eric Dumazet
2022-02-03 1:51 ` [PATCH net-next 15/15] mlx5: " Eric Dumazet
2022-02-03 7:27 ` Tariq Toukan
2022-02-04 4:03 ` kernel test robot
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=CAKgT0UeTvj_6DWUskxxaRiQQxcwg6j0u+UHDaougJSMdkogKWA@mail.gmail.com \
--to=alexander.duyck@gmail.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=eric.dumazet@gmail.com \
--cc=kuba@kernel.org \
--cc=lixiaoyan@google.com \
--cc=netdev@vger.kernel.org \
/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).