[dpdk-stable] [PATCH v4 2/2] ethdev: update qfi definition
Ori Kam
orika at nvidia.com
Tue Apr 13 11:24:32 CEST 2021
Hi Raslan,
> -----Original Message-----
> From: Raslan Darawsheh <rasland at nvidia.com>
> Subject: RE: [PATCH v4 2/2] ethdev: update qfi definition
>
> Hi,
>
> > -----Original Message-----
> > From: Ferruh Yigit <ferruh.yigit at intel.com>
> > Sent: Tuesday, April 6, 2021 7:10 PM
> > To: Raslan Darawsheh <rasland at nvidia.com>; dev at dpdk.org; Ori Kam
> > <orika at nvidia.com>; andrew.rybchenko at oktetlabs.ru
> > Cc: ivan.malov at oktetlabs.ru; ying.a.wang at intel.com;
> > olivier.matz at 6wind.com; Slava Ovsiienko <viacheslavo at nvidia.com>; Shiri
> > Kuzin <shirik at nvidia.com>; stable at dpdk.org; David Marchand
> > <david.marchand at redhat.com>; NBU-Contact-Thomas Monjalon
> > <thomas at monjalon.net>
> > Subject: Re: [PATCH v4 2/2] ethdev: update qfi definition
> >
> > On 4/4/2021 8:45 AM, Raslan Darawsheh wrote:
> > > qfi field is 8 bits which represent single bit for
> > > PPP (paging Policy Presence) single bit for RQI
> > > (Reflective QoS Indicator) and 6 bits for qfi
> > > (QoS Flow Identifier) based on RFC 38415-g30
> > >
> > > This update the doxygen format and the mask for qfi
> > > to properly identify the full 8 bits of the field.
> > >
> > > note: changing the default mask would cause different
> > > patterns generated by testpmd.
> > >
> > > Fixes: 346553db5bd1 ("ethdev: add GTP extension header to flow API")
> > > Cc: ying.a.wang at intel.com
> > > Cc: stable at dpdk.org
> > >
> > > Signed-off-by: Raslan Darawsheh <rasland at nvidia.com>
> > > ---
> > > doc/guides/testpmd_app_ug/testpmd_funcs.rst | 3 ++-
> > > lib/librte_ethdev/rte_flow.h | 20 +++++++++++++++++---
> > > 2 files changed, 19 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/doc/guides/testpmd_app_ug/testpmd_funcs.rst
> > b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
> > > index f59eb8a27d..dd39c4c3c2 100644
> > > --- a/doc/guides/testpmd_app_ug/testpmd_funcs.rst
> > > +++ b/doc/guides/testpmd_app_ug/testpmd_funcs.rst
> > > @@ -3742,7 +3742,8 @@ This section lists supported pattern items and
> > their attributes, if any.
> > > - ``gtp_psc``: match GTP PDU extension header with type 0x85.
> > >
> > > - ``pdu_type {unsigned}``: PDU type.
> > > - - ``qfi {unsigned}``: QoS flow identifier.
> > > +
> > > + - ``qfi {unsigned}``: PPP, RQI and QoS flow identifier.
> > >
> > > - ``pppoes``, ``pppoed``: match PPPoE header.
> > >
> > > diff --git a/lib/librte_ethdev/rte_flow.h b/lib/librte_ethdev/rte_flow.h
> > > index 6cc57136ac..20b4389429 100644
> > > --- a/lib/librte_ethdev/rte_flow.h
> > > +++ b/lib/librte_ethdev/rte_flow.h
> > > @@ -20,6 +20,7 @@
> > > #include <rte_arp.h>
> > > #include <rte_common.h>
> > > #include <rte_ether.h>
> > > +#include <rte_gtp.h>
> > > #include <rte_icmp.h>
> > > #include <rte_ip.h>
> > > #include <rte_sctp.h>
> > > @@ -1421,16 +1422,29 @@ static const struct rte_flow_item_meta
> > rte_flow_item_meta_mask = {
> > > *
> > > * Matches a GTP PDU extension header with type 0x85.
> > > */
> > > +RTE_STD_C11
> > > struct rte_flow_item_gtp_psc {
> > > - uint8_t pdu_type; /**< PDU type. */
> > > - uint8_t qfi; /**< QoS flow identifier. */
> > > + union {
> > > + struct {
> > > + /*
> > > + * These fields are retained for compatibility.
> > > + * Please switch to the new header field below.
> > > + */
> > > + uint8_t pdu_type; /**< PDU type. */
> > > + uint8_t qfi; /**< PPP, RQI, QoS flow identifier. */
> > > +
> > > + };
> > > + struct rte_gtp_psc_hdr hdr;
> > > + };
> > > };
> >
> > This will change the struct size even with union, since old version is missing
> > all fields from protocol header. Struct size will go from 2 --> 5 bytes [1].
> >
> > Since this is public struct, we can't change its size.
> >
> > @Ori, Andrew,
> >
> > Do you have a suggestion for next step?
> >
I think Ferruh is right, and I think that we should at this point just update the documentation.
Sorry for the detour
Just small comment about the original patch.
I don’t think you should change the default mask since it may break existing apps.
> > - We can still add the "struct rte_gtp_psc_hdr", and add a deprecation notice
> > for "struct rte_flow_item_gtp_psc" to say it will use "struct
> > rte_gtp_psc_hdr"
> > on 21.11.
> >
> > - And for this release use the Raslan's first version:
> > - uint8_t qfi; /**< QoS flow identifier. */
> > + uint8_t qfi; /**< PPP, RQI, QoS flow identifier. */
> >
> >
> > Does it make sense? Any comments?
> >
> @Ori Kam, @andrew.rybchenko at oktetlabs.ru,
> This is a kind remainder of this patch,
>
> >
> >
> > [1]
> > struct rte_gtp_psc_hdr {
> > uint8_t ext_hdr_len; /* 0 1 */
> > uint8_t type:4; /* 1: 0 1 */
> > uint8_t qmp:1; /* 1: 4 1 */
> >
> > /* XXX 3 bits hole, try to pack */
> >
> > union {
> > struct {
> > uint8_t snp:1; /* 2: 0 1 */
> > uint8_t spare_dl1:2; /* 2: 1 1 */
> > }; /* 2 1 */
> > struct {
> > uint8_t dl_delay_ind:1; /* 2: 0 1 */
> > uint8_t ul_delay_ind:1; /* 2: 1 1 */
> > uint8_t snp_ul1:1; /* 2: 2 1 */
> > }; /* 2 1 */
> > }; /* 2 1 */
> > union {
> > struct {
> > uint8_t ppp:1; /* 3: 0 1 */
> > uint8_t rqi:1; /* 3: 1 1 */
> > }; /* 3 1 */
> > struct {
> > uint8_t n_delay_ind:1; /* 3: 0 1 */
> > uint8_t spare_ul2:1; /* 3: 1 1 */
> > }; /* 3 1 */
> > }; /* 3 1 */
> > uint8_t qfi:6; /* 4: 0 1 */
> >
> > /* XXX 2 bits hole, try to pack */
> >
> > uint8_t data[]; /* 5 0 */
> >
> > /* size: 5, cachelines: 1, members: 7 */
> > /* sum members: 3 */
> > /* sum bitfield members: 11 bits, bit holes: 2, sum bit holes: 5 bits */
> > /* last cacheline: 5 bytes */
> > };
> >
> > >
> > > /** Default mask for RTE_FLOW_ITEM_TYPE_GTP_PSC. */
> > > #ifndef __cplusplus
> > > static const struct rte_flow_item_gtp_psc
> > > rte_flow_item_gtp_psc_mask = {
> > > - .qfi = 0x3f,
> > > + .hdr.ppp = 1,
> > > + .hdr.rqi = 1,
> > > + .hdr.qfi = 0x3f,
> > > };
> > > #endif
> > >
> > >
More information about the stable
mailing list