patches for DPDK stable branches
 help / color / mirror / Atom feed
From: "Li, Xiaoyun" <xiaoyun.li@intel.com>
To: "Olivier Matz" <olivier.matz@6wind.com>,
	"Morten Brørup" <mb@smartsharesystems.com>
Cc: "Yigit, Ferruh" <ferruh.yigit@intel.com>,
	"Ananyev, Konstantin" <konstantin.ananyev@intel.com>,
	"stephen@networkplumber.org" <stephen@networkplumber.org>,
	"dev@dpdk.org" <dev@dpdk.org>,
	"stable@dpdk.org" <stable@dpdk.org>,
	"Medvedkin, Vladimir" <vladimir.medvedkin@intel.com>
Subject: RE: [dpdk-dev] [PATCH v3] app/testpmd: fix l4 sw csum over multi segments
Date: Fri, 3 Dec 2021 11:31:14 +0000	[thread overview]
Message-ID: <DM4PR11MB55340CF7594BE3F3E43E8D9D996A9@DM4PR11MB5534.namprd11.prod.outlook.com> (raw)
In-Reply-To: <YXuw0I/02IcYEAPX@platinum>

Hi

> -----Original Message-----
> From: Olivier Matz <olivier.matz@6wind.com>
> Sent: Friday, October 29, 2021 09:29
> To: Morten Brørup <mb@smartsharesystems.com>
> Cc: Yigit, Ferruh <ferruh.yigit@intel.com>; Li, Xiaoyun <xiaoyun.li@intel.com>;
> Ananyev, Konstantin <konstantin.ananyev@intel.com>;
> stephen@networkplumber.org; dev@dpdk.org; stable@dpdk.org;
> Medvedkin, Vladimir <vladimir.medvedkin@intel.com>
> Subject: Re: [dpdk-dev] [PATCH v3] app/testpmd: fix l4 sw csum over multi
> segments
> 
> On Wed, Oct 27, 2021 at 01:29:52PM +0200, Morten Brørup wrote:
> > > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Ferruh Yigit
> > > Sent: Wednesday, 27 October 2021 12.49
> > >
> > > On 10/20/2021 11:12 AM, Xiaoyun Li wrote:
> > > > In csum forwarding mode, software UDP/TCP csum calculation only
> > > > takes the first segment into account while using the whole packet
> > > > length so the calculation will read invalid memory region with
> > > > multi-segments packets and will get wrong value.
> > > > This patch fixes this issue.
> > > >
> > > > Fixes: af75078fece3 ("first public release")
> > > > Cc: stable@dpdk.org
> > > >
> > > > Signed-off-by: Xiaoyun Li <xiaoyun.li@intel.com>
> > > > ---
> > > > v3:
> > > >   * Use rte_raw_cksum() for multi-segs case instead of copying the
> > > whole
> > > >   * packet.
> > > > v2:
> > > >   * Use static stack memory instead of dynamic allocating in
> > > > datapath
> > > > ---
> > > >   app/test-pmd/csumonly.c | 68
> > > > ++++++++++++++++++++++++++++++++------
> > > ---
> > > >   1 file changed, 53 insertions(+), 15 deletions(-)
> > > >
> > > > diff --git a/app/test-pmd/csumonly.c b/app/test-pmd/csumonly.c
> > > > index 090797318a..f3e60eb3c3 100644
> > > > --- a/app/test-pmd/csumonly.c
> > > > +++ b/app/test-pmd/csumonly.c
> > > > @@ -91,12 +91,41 @@ struct simple_gre_hdr {
> > > >   } __rte_packed;
> > > >
> > > >   static uint16_t
> > > > -get_udptcp_checksum(void *l3_hdr, void *l4_hdr, uint16_t
> > > > ethertype)
> > > > +get_udptcp_checksum(void *l3_hdr, struct rte_mbuf *m, uint16_t
> > > l4_off,
> > > > +		    uint16_t ethertype)
> > > >   {
> > > > +	uint16_t off = l4_off;
> > > > +	uint32_t cksum = 0;
> > > > +	char *buf;
> > > > +
> > > > +	while (m != NULL) {
> > > > +		buf = rte_pktmbuf_mtod_offset(m, char *, off);
> > > > +		cksum += rte_raw_cksum(buf, m->data_len - off);
> > > > +		off = 0;
> > > > +		m = m->next;
> > > > +	}
> > > >   	if (ethertype == _htons(RTE_ETHER_TYPE_IPV4))
> > > > -		return rte_ipv4_udptcp_cksum(l3_hdr, l4_hdr);
> > > > +		cksum += rte_ipv4_phdr_cksum(l3_hdr, 0);
> > > >   	else /* assume ethertype == RTE_ETHER_TYPE_IPV6 */
> > > > -		return rte_ipv6_udptcp_cksum(l3_hdr, l4_hdr);
> > > > +		cksum += rte_ipv6_phdr_cksum(l3_hdr, 0);
> > > > +
> > >
> > > Hi Xiaoyun,
> > >
> > > I can see 'rte_ipv[46]_udptcp_cksum()' is not taking multi segment
> > > mbuf into account, so this fix is required, but instead of
> > > implementing this logic into testpmd, what do you think to have APIs
> > > to support multi segment mbufs?
> > > This way other applications also benefit from it and we don't need
> > > to maintain ip4/6 checksum related code in testpmd.
> >
> > +1
> >
> > Also, there is no need to implement the multi-segment raw checksum loop
> in test-pmd.
> >
> > You can use the multi-segment raw checksum function in the net library
> instead:
> > http://code.dpdk.org/dpdk/latest/source/lib/net/rte_ip.h#L224
> 
> +1
> 
> We can have mbuf variants of udptcp checksum functions:
> 
> rte_ipv4_udptcp_cksum()
> rte_ipv4_udptcp_cksum_verify()
> rte_ipv6_udptcp_cksum()
> rte_ipv6_udptcp_cksum_verify()
> 
> Adding a "_mbuf" suffix would be consistent with rte_raw_cksum_mbuf().

Thanks for the suggestion. Since it's API change, I'll send these in 22.03 release with release note.
V4 will come soon.

> 
> 
> Olivier
--------------------------------------------------------------
Intel Research and Development Ireland Limited
Registered in Ireland
Registered Office: Collinstown Industrial Park, Leixlip, County Kildare
Registered Number: 308263


This e-mail and any attachments may contain confidential material for the sole
use of the intended recipient(s). Any review or distribution by others is
strictly prohibited. If you are not the intended recipient, please contact the
sender and delete all copies.


      reply	other threads:[~2021-12-03 11:31 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-10-15  5:13 [dpdk-stable] [PATCH] " Xiaoyun Li
2021-10-15  8:09 ` [dpdk-stable] [dpdk-dev] " David Marchand
2021-10-18  2:02   ` Li, Xiaoyun
2021-10-18  2:16 ` [dpdk-stable] [PATCH v2] " Xiaoyun Li
2021-10-18  3:00 ` [dpdk-stable] [dpdk-dev] [PATCH] " Stephen Hemminger
2021-10-18  3:16   ` Li, Xiaoyun
2021-10-18  4:40     ` Li, Xiaoyun
2021-10-18 10:15     ` Ananyev, Konstantin
2021-10-19  1:54       ` Li, Xiaoyun
2021-10-20 10:12 ` [dpdk-stable] [PATCH v3] " Xiaoyun Li
2021-10-27 10:48   ` Ferruh Yigit
2021-10-27 11:29     ` [dpdk-stable] [dpdk-dev] " Morten Brørup
2021-10-29  8:29       ` Olivier Matz
2021-12-03 11:31         ` Li, Xiaoyun [this message]

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=DM4PR11MB55340CF7594BE3F3E43E8D9D996A9@DM4PR11MB5534.namprd11.prod.outlook.com \
    --to=xiaoyun.li@intel.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=konstantin.ananyev@intel.com \
    --cc=mb@smartsharesystems.com \
    --cc=olivier.matz@6wind.com \
    --cc=stable@dpdk.org \
    --cc=stephen@networkplumber.org \
    --cc=vladimir.medvedkin@intel.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).