From: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
To: Stephen Hemminger <stephen@networkplumber.org>
Cc: Usama Nadeem <usama.nadeem@emumba.com>,
"thomas@monjalon.net" <thomas@monjalon.net>,
"dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH v2] Warns if IPv4, UDP or TCP checksum offload not available
Date: Wed, 15 Sep 2021 08:43:03 +0000 [thread overview]
Message-ID: <DM6PR11MB4491B8CF1C2308C91F7A87359ADB9@DM6PR11MB4491.namprd11.prod.outlook.com> (raw)
In-Reply-To: <20210914164415.020dd967@hermes.local>
> -----Original Message-----
> From: Stephen Hemminger <stephen@networkplumber.org>
> Sent: Wednesday, September 15, 2021 12:44 AM
> To: Ananyev, Konstantin <konstantin.ananyev@intel.com>
> Cc: Usama Nadeem <usama.nadeem@emumba.com>; thomas@monjalon.net; dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] Warns if IPv4, UDP or TCP checksum offload not available
>
> On Tue, 14 Sep 2021 22:22:04 +0000
> "Ananyev, Konstantin" <konstantin.ananyev@intel.com> wrote:
>
> > >
> > > From: usamanadeem321 <usama.nadeem@emumba.com>
> > >
> > > Checks if IPV4, UDP and TCP Checksum offloads are available.
> > > If not available, prints a warning message.
> > >
> > > Bugzilla ID: 545
> > > Signed-off-by: usamanadeem321 <usama.nadeem@emumba.com>
> > > ---
> > > examples/l3fwd/main.c | 22 +++++++++++++++++++++-
> > > 1 file changed, 21 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/examples/l3fwd/main.c b/examples/l3fwd/main.c
> > > index 00ac267af1..ae62bc570d 100644
> > > --- a/examples/l3fwd/main.c
> > > +++ b/examples/l3fwd/main.c
> > > @@ -123,7 +123,6 @@ static struct rte_eth_conf port_conf = {
> > > .mq_mode = ETH_MQ_RX_RSS,
> > > .max_rx_pkt_len = RTE_ETHER_MAX_LEN,
> > > .split_hdr_size = 0,
> > > - .offloads = DEV_RX_OFFLOAD_CHECKSUM,
> > > },
> > > .rx_adv_conf = {
> > > .rss_conf = {
> > > @@ -1039,6 +1038,27 @@ l3fwd_poll_resource_setup(void)
> > > local_port_conf.txmode.offloads |=
> > > DEV_TX_OFFLOAD_MBUF_FAST_FREE;
> > >
> > > + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_IPV4_CKSUM)
> > > + local_port_conf.rxmode.offloads |=
> > > + DEV_RX_OFFLOAD_IPV4_CKSUM;
> > > + else {
> > > + printf("WARNING: IPV4 Checksum offload not available.\n");
> > > + }
> > > +
> > > + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_UDP_CKSUM)
> > > + local_port_conf.rxmode.offloads |=
> > > + DEV_RX_OFFLOAD_UDP_CKSUM;
> > > +
> > > + else
> > > + printf("WARNING: UDP Checksum offload not available.\n");
> > > +
> > > + if (dev_info.rx_offload_capa & DEV_RX_OFFLOAD_TCP_CKSUM)
> > > + local_port_conf.rxmode.offloads |=
> > > + DEV_RX_OFFLOAD_TCP_CKSUM;
> > > +
> > > + else
> > > + printf("WARNING: TCP Checksum offload not available.\n");
> > > +
> >
> > Sorry, but I didn't get the logic:
> > Application expects some offloads to be supported by HW.
>
> The application is expecting more offloads than is necessary for basic
> IP level forwarding which is all the example is documented to do.
>
> "The application performs L3 forwarding."
>
> > You add the code that checks for offloads, but if they are not supported just prints warning
> > and continues, as if everything is ok. Doesn't look like correct behaviour to me.
> > I think, it should either terminate with error message or be prepared to work properly
> > on HW without these offloads (check cksums in SW if necessary).
> > In fact I don't see what was wrong with original behaviour, one thing that probably
> > was missing - more descriptive error message.
>
> It is not a problem with your patch, it is fine.
>
> It is a problem in how l3fwd has grown and changed and no longer really what
> was intended in the original version. There is no reason that the application
> should be looking at L4 data. In fact, it shouldn't care if it gets TCP, UDP, SCP or DCCP;
> but the application now depends on ptype.
>
> It should be possible to do L3 forwarding independent of packet type.
> The application only needs to look at Ether type and do IPv4 or IPv6 based on that.
>
As I remember l3fwd cares about L4 headers (chan cksums) because it can do FWD decisions
based on 5-tuple (exact-macth mode).
I presume that's the reason L4 cksum offloads was enabled at first place.
For LPM/FIB I believe ipv4 cksum check should be sufficient.
If we believe that some offloads are excessive,
then I think right way is to simply remove them
(with updating docs and source in a proper way etc.).
Just printing warnings and continuing seems wrong to me.
next prev parent reply other threads:[~2021-09-15 8:43 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-13 12:09 [dpdk-dev] [PATCH] examples: ipv4, udp and tcp checksum offload warning usamanadeem321
2021-09-13 15:11 ` Stephen Hemminger
2021-09-14 18:08 ` [dpdk-dev] [PATCH v2] Warns if IPv4, UDP or TCP checksum offload not available Usama Nadeem
2021-09-14 18:28 ` Stephen Hemminger
2021-09-14 22:22 ` Ananyev, Konstantin
2021-09-14 23:44 ` Stephen Hemminger
2021-09-15 8:43 ` Ananyev, Konstantin [this message]
2021-10-08 15:51 ` [dpdk-dev] [PATCH v3] ipv4 and udp/tcp cksum verification through software Usama Nadeem
2021-10-14 18:43 ` [dpdk-dev] [PATCH v4] examples/l3fwd: " Usama Nadeem
2021-11-01 8:33 ` Usama Nadeem
2021-11-04 11:11 ` Walsh, Conor
2021-11-04 16:19 ` Medvedkin, Vladimir
2021-11-16 5:20 ` Usama Nadeem
2021-11-16 5:21 ` Usama Nadeem
2023-06-30 21:50 ` Stephen Hemminger
2021-11-04 13:19 ` Ananyev, Konstantin
2021-11-16 5:18 ` Usama Nadeem
2022-01-14 9:30 ` Usama Nadeem
2022-01-14 12:05 ` Ananyev, Konstantin
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=DM6PR11MB4491B8CF1C2308C91F7A87359ADB9@DM6PR11MB4491.namprd11.prod.outlook.com \
--to=konstantin.ananyev@intel.com \
--cc=dev@dpdk.org \
--cc=stephen@networkplumber.org \
--cc=thomas@monjalon.net \
--cc=usama.nadeem@emumba.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).