Re: [dpdk-dev] app/testpmd: fix PPPOES flow API

["Zhao1, Wei" <wei.zhao1@intel.com>]

Hi, Ori

> -----Original Message-----
> From: Ori Kam <orika@mellanox.com>
> Sent: Monday, March 30, 2020 3:43 PM
> To: Zhang, Xiao <xiao.zhang@intel.com>; dev@dpdk.org
> Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> stable@dpdk.org
> Subject: RE: app/testpmd: fix PPPOES flow API
> 
> Hi Xiao
> 
> > -----Original Message-----
> > From: Zhang, Xiao <xiao.zhang@intel.com>
> > Sent: Monday, March 30, 2020 5:09 AM
> > To: Ori Kam <orika@mellanox.com>; dev@dpdk.org
> > Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> > <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> > stable@dpdk.org
> > Subject: RE: app/testpmd: fix PPPOES flow API
> >
> > Hi Ori,
> >
> > > -----Original Message-----
> > > From: Ori Kam <orika@mellanox.com>
> > > Sent: Sunday, March 29, 2020 8:46 PM
> > > To: Zhang, Xiao <xiao.zhang@intel.com>; dev@dpdk.org
> > > Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> > > <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> > stable@dpdk.org
> > > Subject: RE: app/testpmd: fix PPPOES flow API
> > >
> > > Hi Xiao,
> > >
> > > > -----Original Message-----
> > > > From: Zhang, Xiao <xiao.zhang@intel.com>
> > > > Sent: Sunday, March 29, 2020 3:00 PM
> > > > To: Ori Kam <orika@mellanox.com>; dev@dpdk.org
> > > > Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> > > > <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> > > > stable@dpdk.org
> > > > Subject: RE: app/testpmd: fix PPPOES flow API
> > > >
> > > > Hi Ori,
> > > >
> > > > > -----Original Message-----
> > > > > From: Ori Kam <orika@mellanox.com>
> > > > > Sent: Sunday, March 29, 2020 6:19 PM
> > > > > To: Zhang, Xiao <xiao.zhang@intel.com>; dev@dpdk.org
> > > > > Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> > > > > <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> > > > stable@dpdk.org
> > > > > Subject: RE: app/testpmd: fix PPPOES flow API
> > > > >
> > > > > Hi Xiao,
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Zhang, Xiao <xiao.zhang@intel.com>
> > > > > > Sent: Sunday, March 29, 2020 12:06 PM
> > > > > > To: Ori Kam <orika@mellanox.com>; dev@dpdk.org
> > > > > > Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> > > > > > <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> > > > > > stable@dpdk.org
> > > > > > Subject: RE: app/testpmd: fix PPPOES flow API
> > > > > >
> > > > > > Hi Ori,
> > > > > >
> > > > > > > -----Original Message-----
> > > > > > > From: Ori Kam <orika@mellanox.com>
> > > > > > > Sent: Sunday, March 29, 2020 2:28 PM
> > > > > > > To: Zhang, Xiao <xiao.zhang@intel.com>; dev@dpdk.org
> > > > > > > Cc: Wang, Ying A <ying.a.wang@intel.com>; Zhang, Qi Z
> > > > > > > <qi.z.zhang@intel.com>; Zhao1, Wei <wei.zhao1@intel.com>;
> > > > > > stable@dpdk.org
> > > > > > > Subject: RE: app/testpmd: fix PPPOES flow API
> > > > > > >
> > > > > > > Hi Xiao,
> > > > > > >
> > > > > > > Is the proto_id part of the basic header or not?
> > > > > >
> > > > > > Proto_id is part of PPPOE session header,
> > > > > >
> > > > >
> > > > > Where is the porto_id located? Inside the payload?
> > > >
> > > > Yes, my previous explanation was not clear. The protocol ID is in
> > > > the beginning of the payload in PPP Session Stage according to RFC2516.
> > > >
> > > >                      	         1
> 2                                3
> > > >    0 1 2  3  4  5 6 7 8 9 0 1  2 3 4  5 6 7 8 9  0 1 2 3  4  5 6 7
> > > > 8 9
> > > > 0 1  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > > >  |  VER   | TYPE   |      CODE      |
> SESSION_ID                    |
> > > >  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > > >  |                 LENGTH                    |
> payload                      ~
> > > >  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> > > >
> > >
> > > Yes this is what I thought, does proto_id must be the first part of the
> payload?
> >
> > It must be the first part of the payload for PPP Session Stage, not
> > all PPPOE packets.
> >
> > >
> > > > >
> > > > > > >
> > > > > > > From the spec it looks like a different header.
> > > > > > >
> > > > > > > If it is part of the original header then all documentations
> > > > > > > and rte_structs
> > > > > > should
> > > > > > > be changed, to reflect this.
> > > > > > >
> > > > > > > It will be very helpful if the patch message would explain
> > > > > > > the bug and why it
> > > > > > was
> > > > > > > changed.
> > > > > >
> > > > > > Okay, will add more message. The next value of the
> > > > ITEM_PPPOE_PROTO_ID
> > > > > > should be unsigned value but not item list.
> > > > > >
> > > > > > >
> > > > > > > Also please see inline other comment.
> > > > > > >
> > > > > > > Best,
> > > > > > > Ori
> > > > > > >
> > > > > > > > -----Original Message-----
> > > > > > > > From: Xiao Zhang <xiao.zhang@intel.com>
> > > > > > > > Sent: Friday, March 27, 2020 11:19 AM
> > > > > > > > To: dev@dpdk.org
> > > > > > > > Cc: Ori Kam <orika@mellanox.com>; ying.a.wang@intel.com;
> > > > > > > > qi.z.zhang@intel.com; wei.zhao1@intel.com; Xiao Zhang
> > > > > > > > <xiao.zhang@intel.com>; stable@dpdk.org
> > > > > > > > Subject: app/testpmd: fix PPPOES flow API
> > > > > > > >
> > > > > > > > The command line to create RTE flow for specific proto_id
> > > > > > > > of PPPOES is not correct. This patch is to fix this issue.
> > > > > > > >
> > > > > > > > Fixes: 226c6e60c35b ("ethdev: add PPPoE to flow API")
> > > > > > > > Cc: stable@dpdk.org
> > > > > > > >
> > > > > > > > Signed-off-by: Xiao Zhang <xiao.zhang@intel.com>
> > > > > > > > ---
> > > > > > > >  app/test-pmd/cmdline_flow.c | 13 +++----------
> > > > > > > >  1 file changed, 3 insertions(+), 10 deletions(-)
> > > > > > > >
> > > > > > > > diff --git a/app/test-pmd/cmdline_flow.c
> > > > > > > > b/app/test-pmd/cmdline_flow.c index a78154502..c25a2598d
> > > > > > > > 100644
> > > > > > > > --- a/app/test-pmd/cmdline_flow.c
> > > > > > > > +++ b/app/test-pmd/cmdline_flow.c
> > > > > > > > @@ -768,7 +768,6 @@ static const enum index next_item[] = {
> > > > > > > >  	ITEM_GTP_PSC,
> > > > > > > >  	ITEM_PPPOES,
> > > > > > > >  	ITEM_PPPOED,
> > > > > > > > -	ITEM_PPPOE_PROTO_ID,
> > > > > > > >  	ITEM_HIGIG2,
> > > > > > > >  	ITEM_TAG,
> > > > > > > >  	ITEM_L2TPV3OIP,
> > > > > > > > @@ -1030,11 +1029,6 @@ static const enum index
> > > > > > > > item_pppoed[] = {
> > > > > > > >
> > > > > > > >  static const enum index item_pppoes[] = {
> > > > > > > >  	ITEM_PPPOE_SEID,
> > > > > > > > -	ITEM_NEXT,
> > > > > > > > -	ZERO,
> > > > > > > > -};
> > > > > > > > -
> > > > > > > > -static const enum index item_pppoe_proto_id[] = {
> > > > > > > >  	ITEM_PPPOE_PROTO_ID,
> > > > > > > >  	ITEM_NEXT,
> > > > > > > >  	ZERO,
> > > > > > > > @@ -2643,10 +2637,9 @@ static const struct token token_list[] =
> {
> > > > > > > >  	[ITEM_PPPOE_PROTO_ID] = {
> > > > > > > >  		.name = "proto_id",
> > > > > > > >  		.help = "match PPPoE session protocol identifier",
> > > > > > > > -		.priv = PRIV_ITEM(PPPOE_PROTO_ID,
> > > > > > > > -				sizeof(struct
> > > > rte_flow_item_pppoe_proto_id)),
> > > > > > > > -		.next = NEXT(item_pppoe_proto_id),
> > > > > > > > -		.call = parse_vc,
> > > > > > > > +		.next = NEXT(item_pppoes, NEXT_ENTRY(UNSIGNED),
> > > > > > > > item_param),
> > > > > > > > +		.args = ARGS(ARGS_ENTRY_HTON
> > > > > > > > +			     (struct rte_flow_item_pppoe_proto_id,
> > > > proto_id)),
> > > > > > >
> > > > > > > Where is the memory for this proto_id is defined?
> > > > > >
> > > > > > Do you mean this?
> > > > > > lib/librte_ethdev/rte_flow.h
> > > > > > 1360 struct rte_flow_item_pppoe_proto_id {
> > > > > > 1361         rte_be16_t proto_id; /**< PPP protocol identifier. */
> > > > > > 1362 };
> > > > > >
> > > > >
> > > > > Yes. Why don't you use this one?
> > > >
> > > > I think I was using this, am I using it incorrectly?
> > > >
> > > > +		.args = ARGS(ARGS_ENTRY_HTON
> > > > +			     (struct rte_flow_item_pppoe_proto_id, proto_id)),
> > > >
> > >
> > > Yes but there is no space to save this data since you deleted the priv.
> > > I think you are trying to implement something like
> > > RTE_FLOW_ITEM_TYPE_IPV6_EXT.
> > >
> > > And I don't understand what was the problem with the previous
> > implementation.
> >
> > I deleted the priv because it changed to a subcommand in pppoes, the
> > command line will be like this:
> > testpmd> flow create 0 ingress pattern eth dst is 00:11:22:33:44:55 /
> > testpmd> pppoes
> > proto_id is 21
> >
> 
> The issue is that the pppoe struct doesn't have definition to the proto_id.
> If you wish a possible solution will be to add it to the pppoe struct, I'm not sure
> if this is the correct approach since this field is not a must.
> 
> Like I said there are examples on how to work with extended headers, which I
> think are more correct, buy may be the problem is that the pppoe struct is not
> aligned and this result in an issue when adding the last bytes.


There is a defination of RTE_FLOW_ITEM_TYPE_PPPOE_PROTO_ID, do you mean make use of that?
That is the reason for use extended header for this?
But that seems as you say, has some bug.


> 
> > The previous implementation would be infinite loop for proto_id
> > command and can not specific the value for proto_id.
> > testpmd> flow create 0 ingress pattern eth dst is 00:11:22:33:44:55 /
> > testpmd> proto_id
> >  proto_id [TOKEN]: match PPPoE session protocol identifier  / [TOKEN]:
> > specify next pattern item
> > testpmd> flow create 0 ingress pattern eth dst is 00:11:22:33:44:55 /
> > testpmd> proto_id
> > proto_id
> >  proto_id [TOKEN]: match PPPoE session protocol identifier  / [TOKEN]:
> > specify next pattern item
> > testpmd> flow create 0 ingress pattern eth dst is 00:11:22:33:44:55 /
> > testpmd> proto_id
> > proto_id proto_id
> >  proto_id [TOKEN]: match PPPoE session protocol identifier  / [TOKEN]:
> > specify next pattern item
> > testpmd> flow create 0 ingress pattern eth dst is 00:11:22:33:44:55 /
> > testpmd> proto_id
> > proto_id proto_id
> >
> > >
> > > > >
> > > > > > >
> > > > > > > >  	},
> > > > > > > >  	[ITEM_HIGIG2] = {
> > > > > > > >  		.name = "higig2",
> > > > > > > > --
> > > > > > > > 2.17.1