[dpdk-stable] [PATCH v2] example/ipv4_multicast: fix app hanging when using clone

[Zhao1, Wei]

Hi, Konstantin

> -----Original Message-----
> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Ananyev,
> Konstantin
> Sent: Wednesday, November 14, 2018 7:19 PM
> To: Wang, Dong1 <dong1.wang at intel.com>; Lipiec, Herakliusz
> <herakliusz.lipiec at intel.com>; dev at dpdk.org
> Cc: thomas at monjalon.net; Burakov, Anatoly <anatoly.burakov at intel.com>;
> stable at dpdk.org
> Subject: Re: [dpdk-dev] [PATCH v2] example/ipv4_multicast: fix app hanging
> when using clone
> 
> 
> 
> > -----Original Message-----
> > From: Wang, Dong1
> > Sent: Wednesday, November 14, 2018 11:06 AM
> > To: Ananyev, Konstantin <konstantin.ananyev at intel.com>; Lipiec,
> > Herakliusz <herakliusz.lipiec at intel.com>; dev at dpdk.org
> > Cc: thomas at monjalon.net; Burakov, Anatoly
> <anatoly.burakov at intel.com>;
> > stable at dpdk.org
> > Subject: RE: [PATCH v2] example/ipv4_multicast: fix app hanging when
> > using clone
> >
> > Hi Konstantin & Herakliusz,
> >
> > In mcast_out_pkt(), new header added and to be the first segment. Driver
> will check first segment's ol_flags when sending mbuf.
> 
> It will check for *TX* flags.
> TX flags are not set in cloned mbuf (only RX ones).
> 
> > So why new first segment didn't inherit old one's ol_flags?
> 
> Because they are RX flags and they don't' really belong to new packet we are
> going to send.
> Also, as that bug shows it is error prone.
> 
> > This function already copied other mbuf fields from old one.
> 
> And from my perspective it is not needed and better be removed.
> Though it is a subject of different patch probably.

In the code , function mcast_out_pkt() has the following code,

	/* copy metadata from source packet*/
	hdr->port = pkt->port;
	hdr->vlan_tci = pkt->vlan_tci;
	hdr->vlan_tci_outer = pkt->vlan_tci_outer;
	hdr->tx_offload = pkt->tx_offload;
	hdr->hash = pkt->hash;

all these are useless , is that right?
I will commit a patch to delete them later.



> 
> >
> > Since this is an example and reference code, hope developer can reuse
> > it smoothly. If they didn't notice ol_flags not be copied because ol_flags is
> not changed in this example, they may take some time to debug their
> application.
> 
> For that scenario it is not needed.
> Surely when user linking two packets manually (as we are doing here) it is his
> responsibility to update the mbuf fileds properly.
> I think this is obvious thing and probably already mentioned in the docs.
> Konstantin
> 
> >
> > This is a tiny change to the patch, I hope it cloud be done.
> >
> > Thanks,
> > Dong
> >
> > -----Original Message-----
> > From: Ananyev, Konstantin
> > Sent: Wednesday, November 14, 2018 18:18
> > To: Wang, Dong1 <dong1.wang at intel.com>; Lipiec, Herakliusz
> > <herakliusz.lipiec at intel.com>; dev at dpdk.org
> > Cc: thomas at monjalon.net; Burakov, Anatoly
> <anatoly.burakov at intel.com>;
> > stable at dpdk.org
> > Subject: RE: [PATCH v2] example/ipv4_multicast: fix app hanging when
> > using clone
> >
> >
> > >
> > > Hi Konstantin,
> > >
> > > In this example, no flags be set by user, so no need to copy flags from
> original  mbuf.
> >
> > Then why to bother?
> >
> > > I'm thinking about some DPDK developers may create their own
> > > application based on this example,  they add some flags before packet be
> cloned.
> >
> > Hmm you receive a packet change l2 header and transmit it.
> > What flags you might need to copy from original header considering that
> only RX flags would be set at present moment?
> > If you let say want to insert vlan tag (or so), then you'll have to set new TX
> flags.
> > Again, for timestamp you'll need to copy the timestamp value too, which
> we don't support in that example.
> > Konstantin
> >
> > >
> > > Regards,
> > > Dong
> > >
> > > -----Original Message-----
> > > From: Ananyev, Konstantin
> > > Sent: Wednesday, November 14, 2018 17:03
> > > To: Wang, Dong1 <dong1.wang at intel.com>; Lipiec, Herakliusz
> > > <herakliusz.lipiec at intel.com>; dev at dpdk.org
> > > Cc: thomas at monjalon.net; Burakov, Anatoly
> > > <anatoly.burakov at intel.com>; stable at dpdk.org
> > > Subject: RE: [PATCH v2] example/ipv4_multicast: fix app hanging when
> > > using clone
> > >
> > > Hi Dong,
> > >
> > > > Hi Herakliusz,
> > > >
> > > > Since pkt->ol_flags might set some other bits except
> > > > IND_ATTACHED_MBUF, how about set its value to original
> > > > pkt->ol_flag
> > >
> > > It could be done, but not sure what is the point?
> > > Which flags you think we need to copy from cloned mbuf (segment #2) to
> new header?
> > > Konstantin
> > >
> > > >
> > > > Regards,
> > > > Dong
> > > >
> > > > -----Original Message-----
> > > > From: Lipiec, Herakliusz
> > > > Sent: Tuesday, November 13, 2018 19:49
> > > > To: dev at dpdk.org
> > > > Cc: Ananyev, Konstantin <konstantin.ananyev at intel.com>; Wang,
> > > > Dong1 <dong1.wang at intel.com>; thomas at monjalon.net; Burakov,
> > > > Anatoly <anatoly.burakov at intel.com>; Lipiec, Herakliusz
> > > > <herakliusz.lipiec at intel.com>; stable at dpdk.org
> > > > Subject: [PATCH v2] example/ipv4_multicast: fix app hanging when
> > > > using clone
> > > >
> > > > The ipv4_multicast sample application was dropping packets when
> > > > using mbuf clone. When creating an L2 header and copying metadata
> > > > from the source packet, the ol_flags were also copied along with
> > > > all the other metadata. Because the cloned packet had
> > > > IND_ATTACHED_MBUF
> > > flag set in its ol_flags, this caused the packets to never be freed when
> using rte_pktmbuf_free.
> > > > Since copying ol_flags from the cloned packet is not necessary in the
> first place, just don't do it.
> > > >
> > > > Fixes: af75078fece3 ("first public release")
> > > > CC: stable at dpdk.org
> > > >
> > > > Reported-by: Wang Dong <dong1.wang at intel.com>
> > > > Signed-off-by: Herakliusz Lipiec <herakliusz.lipiec at intel.com>
> > > > ---
> > > >  doc/guides/sample_app_ug/ipv4_multicast.rst | 1 -
> > > >  examples/ipv4_multicast/main.c              | 2 --
> > > >  2 files changed, 3 deletions(-)
> > > >
> > > > diff --git a/doc/guides/sample_app_ug/ipv4_multicast.rst
> > > > b/doc/guides/sample_app_ug/ipv4_multicast.rst
> > > > index ce1474ec7..f6efa7f6f 100644
> > > > --- a/doc/guides/sample_app_ug/ipv4_multicast.rst
> > > > +++ b/doc/guides/sample_app_ug/ipv4_multicast.rst
> > > > @@ -319,7 +319,6 @@ It is the mcast_out_pkt() function that performs
> the packet duplication (either
> > > >          hdr->pkt.in_port = pkt->pkt.in_port;
> > > >          hdr->pkt.vlan_macip = pkt->pkt.vlan_macip;
> > > >          hdr->pkt.hash = pkt->pkt.hash;
> > > > -        hdr->ol_flags = pkt->ol_flags;
> > > >          rte_mbuf_sanity_check(hdr, RTE_MBUF_PKT, 1);
> > > >
> > > >          return hdr;
> > > > diff --git a/examples/ipv4_multicast/main.c
> > > > b/examples/ipv4_multicast/main.c index 4073a4907..428ca4694 100644
> > > > --- a/examples/ipv4_multicast/main.c
> > > > +++ b/examples/ipv4_multicast/main.c
> > > > @@ -266,8 +266,6 @@ mcast_out_pkt(struct rte_mbuf *pkt, int
> use_clone)
> > > >  	hdr->tx_offload = pkt->tx_offload;
> > > >  	hdr->hash = pkt->hash;
> > > >
> > > > -	hdr->ol_flags = pkt->ol_flags;
> > > > -
> > > >  	__rte_mbuf_sanity_check(hdr, 1);
> > > >  	return hdr;
> > > >  }
> > > > --
> > > > 2.17.1