[ovs-dev] How to define enum raw constants for two new vendor actions

Ajmer Singh ajmersingh.tu06 at gmail.com
Wed Mar 16 12:36:30 UTC 2016


Hi Ben,

I have started adding OFPAT_ENCAP_GTP action in ofp-actions.c file as
following:
taking reference of already added action "OFPAT_ENQUEUE". I have prepared
the structure of Action, also written encode and decode for this new
action. but not able to understand how its *parse_ENCAP_GTP* and
*format_ENCAP_GTP* will be written.

Could you guide what need to take care while writing parse and format for
the action?


/* encap gtp action. */
struct ofp_encap_gtp {
    uint16_t type;                /* OFPAT_ENCAP_GTP */
    uint16_t len;                 /* Length is 16. */
    uint32_t teid;                /* tunnel id for GTP */
    uint32_t src_ip;              /* Source IP for external IP header */
    uint32_t dst_ip;              /* Destination IP for external IP header
*/
    unsigned char src_ipv6[16];   /* Source IP for external IPV6 header */
    unsigned char dst_ipv6[16];   /* Destination IP for external IPV6
header */
};
OFP_ASSERT(sizeof(struct ofp_strip_gtp) == 48);

static enum ofperr
*decode_OFPAT_RAW_ENCAP_GTP*(const struct ofp_encap_gtp *oegtp,
                           struct ofpbuf *out)
{
    struct ofpact_encap_gtp *encapgtp;
    uint32_t dest_ip = ntohl(oegtp->dst_ip);

    encapgtp = ofpact_put_ENCAP_GTP(out);
    encapgtp->teid = ntohl(oegtp->teid);
    if(dest_ip != 0){
       encapgtp->src_ip = ntohl(oegtp->src_ip);
       encapgtp->dst_ip = ntohl(oegtp->dst_ip);
    }else{
       memcpy(&(encapgtp->src_ipv6), &(oegtp->src_ipv6), 16);
       memcpy(&(encapgtp->dst_ipv6), &(oegtp->dst_ipv6), 16);
    }
    return 0;
}

static void
*encode_ENCAP_GTP*(const struct ofpact_encap_gtp *encapgtp,
               enum ofp_version ofp_version, struct ofpbuf *out)
{
    struct ofp_encap_gtp *oae;
    uint32_t dest_ip = htonl(encapgtp->dst_ip);

    oae = put_OFPACT10_ENCAP_GTP(out);
    oae->teid = htonl(encapgtp->teid);
    if(dest_ip != 0){
       oae->src_ip = htonl(encapgtp->src_ip);
       oae->dst_ip = htonl(encapgtp->dst_ip);
    }else{
       memcpy(&(oae->src_ipv6), &(encapgtp->src_ipv6), 16);
       memcpy(&(oae->dst_ipv6), &(encapgtp->dst_ipv6), 16);
    }
}

static char * OVS_WARN_UNUSED_RESULT
*parse_ENCAP_GTP(*char *arg, struct ofpbuf *ofpacts,
              enum ofputil_protocol *usable_protocols OVS_UNUSED)
{
    char *sp = NULL;
    char *port = strtok_r(arg, ":q,", &sp);
    char *queue = strtok_r(NULL, "", &sp);
    struct ofpact_enqueue *enqueue;

    if (port == NULL || queue == NULL) {
        return xstrdup("\"enqueue\" syntax is \"enqueue:PORT:QUEUE\" or "
                       "\"enqueue(PORT,QUEUE)\"");
    }

    enqueue = ofpact_put_ENQUEUE(ofpacts);
    if (!ofputil_port_from_string(port, &enqueue->port)) {
        return xasprintf("%s: enqueue to unknown port", port);
    }
    return str_to_u32(queue, &enqueue->queue);
}

static void
*format_**ENCAP_GTP*(const struct ofpact_enqueue *a, struct ds *s)
{
    ds_put_format(s, "enqueue:");
    ofputil_format_port(a->port, s);
    ds_put_format(s, ":%"PRIu32, a->queue);
}

Regards,
Ajmer

On Tue, Mar 15, 2016 at 8:55 PM, Ben Pfaff <blp at ovn.org> wrote:

> Please don't drop the mailing list.
>
> OK, so you'll have to add code to support the new vendor actions, some
> in build-aux/extract-ofp-msgs, some in ofp-actions.c.
>
> If it's your vendor code then you can use any types you want.
>
> On Tue, Mar 15, 2016 at 11:36:20AM +0530, Ajmer Singh wrote:
> > Hi Ben,
> >
> > With all due respect! I have read the entire comments for this data type
> > but have doubt for below two comments.
> >
> >  #The vendor is OF for standard OpenFlow actions, NX for Nicira
> >  *           extension actions.  (Support for other vendors can be added,
> > but
> >  *           it can't be done just based on a vendor ID definition alone
> >  *           because OpenFlow doesn't define a standard way to specify a
> >  *           subtype for vendor actions, so other vendors might do it
> > different
> >  *           from Nicira.)
> >
> > #The type, in parentheses, is the action type number (for standard
> >  *           OpenFlow actions) or subtype (for vendor extension actions).
> >
> > We are defining two new actions for GTP (STRIP & ENCAP). As per my
> > understanding these are not standard OpenFlow actions that's why OF will
> > not be the vendor, then question is who will? what should be the value of
> > Vendor here. NX will definitely not.
> >
> > Regarding type, I think these values don't match with standard OpenFlow
> > actions structure defined in Openflow specification. then what should be
> > the value in our case.
> >
> > Regards,
> > Ajmer
> >
> >
> > On Mon, Mar 14, 2016 at 8:55 PM, Ben Pfaff <blp at ovn.org> wrote:
> >
> > > On Mon, Mar 14, 2016 at 06:47:34PM +0530, Ajmer Singh wrote:
> > > > Hi,
> > > > I have a requirement of adding GTP headers in open Vswitch source
> code.
> > > We
> > > > need to define 2 new actions: STRIP_GTP and ENCAP_GTP. these are the
> > > > extensions.
> > > >
> > > > As per openflow specifications 1.0 and 1.1. If we add our 2 new
> actions
> > > > then following structure will look like as follows.
> > > >
> > > > enum ofp_action_type {
> > > >     OFPAT_OUTPUT = 0,        /* Output to switch port. */
> > > >     OFPAT_COPY_TTL_OUT = 11, /* Copy TTL "outwards" -- from
> > > > next-to-outermost
> > > >                                 to outermost */
> > > >     OFPAT_COPY_TTL_IN = 12,  /* Copy TTL "inwards" -- from outermost
> to
> > > >                                 next-to-outermost */
> > > >     OFPAT_SET_MPLS_TTL = 15, /* MPLS TTL */
> > > >     OFPAT_DEC_MPLS_TTL = 16, /* Decrement MPLS TTL */
> > > >     OFPAT_PUSH_VLAN = 17,    /* Push a new VLAN tag */
> > > >     OFPAT_POP_VLAN = 18,     /* Pop the outer VLAN tag */
> > > >     OFPAT_PUSH_MPLS = 19,    /* Push a new MPLS tag */
> > > >     OFPAT_POP_MPLS = 20,     /* Pop the outer MPLS tag */
> > > >     OFPAT_SET_QUEUE = 21,    /* Set queue id when outputting to a
> port */
> > > >     OFPAT_GROUP = 22,        /* Apply group. */
> > > >     OFPAT_SET_NW_TTL = 23,   /* IP TTL. */
> > > >     OFPAT_DEC_NW_TTL = 24,   /* Decrement IP TTL. */
> > > >     OFPAT_SET_FIELD = 25,    /* Set a header field using OXM TLV
> format.
> > > */
> > > >     OFPAT_PUSH_PBB = 26,     /*Push a new PBB service tag (I-TAG) */
> > > >     OFPAT_POP_PBB = 27,      /* Pop the outer PBB service tag
> (I-TAG) */
> > > >     OFPAT_STRIP_GTP = 0xfffd,  /* Strip GTP to get inner payload.  */
> > > >     OFPAT_ENCAP_GTP = 0xfffe,  /* Add external IP header, UDP header
> and
> > > > GTP header */
> > > >     OFPAT_EXPERIMENTER = 0xffff
> > > > };
> > > > but I try to support this in open Vswitch code. and I find there is
> one
> > > raw
> > > > action structure named as ofp_raw_action_type (/lib/ofp-actions.c),
> > > defined
> > > > as.
> > > >
> > > > enum ofp_raw_action_type {
> > > > /* ## ----------------- ## */
> > > > /* ## Standard actions. ## */
> > > > /* ## ----------------- ## */
> > > >
> > > >     /* OF1.0(0): struct ofp10_action_output. */
> > > >     OFPAT_RAW10_OUTPUT,
> > > >     /* OF1.1+(0): struct ofp11_action_output. */
> > > >     OFPAT_RAW11_OUTPUT,
> > > > }
> > > >
> > > > now I need to add Raw enum constants for these 2 new actions
> (STRIP_GTP
> > > and
> > > > ENCAP_GTP) in this structure. we have same structures for these 2 new
> > > > actions across all openflow versions. but facing issue in defining
> > > comment
> > > > line before these constants.
> > > > /* OF1.0(0): struct ofp10_action_output. */
> > > > here, OF->for standard open flow actions
> > > > 1.0->openflow protocol version
> > > > (0)->action type number in specification
> > > > struct ofp10_action_output -> corresponding structure for this
> action.
> > > >
> > > > Here, I m facing issue in defining these comments since these two new
> > > > actions are not standard openflow actions. these are vendor actions.
> then
> > > > how to write the comments for these new actions.
> > >
> > > Did you read the large comment just above enum ofp_raw_action_type?  It
> > > has lots of information.
> > >
>



More information about the dev mailing list