/* \summary: Pragmatic General Multicast (PGM) printer */
-#ifdef HAVE_CONFIG_H
+/* specification: RFC 3208
+
+ Plus https://dl.acm.org/doi/pdf/10.1145/347057.347390 for PGMCC,
+ whence the ACK packet type comes; there are some I-Ds for PGMCC,
+ draft-ietf-rmt-bb-pgmcc-00 through draft-ietf-rmt-bb-pgmcc-03,
+ but none of them give any description of the packet-level
+ changes to PGM, unlike the paper in question, which merely gives
+ an *insufficient* description of said changes. In particular,
+ it doesn't indicate what the packet type code for ACK is.
+
+ Luigi Rizzo's PGMCC code for FreeBSD, at
+
+ https://web.archive.org/web/20020302084503/http://info.iet.unipi.it/~luigi/pgm-code/
+
+ uses 0x0b (11) for ACK.
+
+ A capture file attached to
+
+ https://gitlab.com/wireshark/wireshark/-/issues/4798
+
+ has packets that use 0x0d for ACK, as does the Wireshark dissector
+ for PGM, and as does OpenPGM at https://github.com/steve-o/openpgm.
+ It may be that some proprietary PGMCC implementations, such as
+ SmartPGM, do so as well.
+
+ We use *both*, treating *either one* as a PGMCC ACK, pending
+ more information, such as an answer to
+
+ https://github.com/steve-o/openpgm/issues/75.
+
+ */
+
#include <config.h>
-#endif
#include "netdissect-stdinc.h"
};
typedef enum _pgm_type {
- PGM_SPM = 0, /* source path message */
- PGM_POLL = 1, /* POLL Request */
- PGM_POLR = 2, /* POLL Response */
- PGM_ODATA = 4, /* original data */
- PGM_RDATA = 5, /* repair data */
- PGM_NAK = 8, /* NAK */
- PGM_NULLNAK = 9, /* Null NAK */
- PGM_NCF = 10, /* NAK Confirmation */
- PGM_ACK = 11, /* ACK for congestion control */
- PGM_SPMR = 12, /* SPM request */
- PGM_MAX = 255
+ PGM_SPM = 0x00, /* source path message */
+ PGM_POLL = 0x01, /* POLL Request */
+ PGM_POLR = 0x02, /* POLL Response */
+ PGM_ODATA = 0x04, /* original data */
+ PGM_RDATA = 0x05, /* repair data */
+ PGM_NAK = 0x08, /* NAK */
+ PGM_NULLNAK = 0x09, /* Null NAK */
+ PGM_NCF = 0x0a, /* NAK Confirmation */
+ PGM_ACK = 0x0b, /* ACK for congestion control? */
+ PGM_SPMR = 0x0c, /* SPM request */
+ PGM_ACK2 = 0x0d, /* Also ACK for congestion control? */
} pgm_type;
#define PGM_OPT_BIT_PRESENT 0x01
pgm->pgm_gsid[3],
pgm->pgm_gsid[4],
pgm->pgm_gsid[5]);
+ bp += sizeof(struct pgm_header);
switch (pgm_type_val) {
case PGM_SPM: {
const struct pgm_spm *spm;
- spm = (const struct pgm_spm *)(pgm + 1);
+ spm = (const struct pgm_spm *)bp;
ND_TCHECK_SIZE(spm);
- bp = (const u_char *) (spm + 1);
+ bp += sizeof(struct pgm_spm);
switch (GET_BE_U_2(spm->pgms_nla_afi)) {
case AFNUM_IP:
break;
default:
goto invalid;
- break;
}
ND_PRINT("SPM seq %u trail %u lead %u nla %s",
const struct pgm_poll *pgm_poll;
uint32_t ivl, rnd, mask;
- pgm_poll = (const struct pgm_poll *)(pgm + 1);
+ pgm_poll = (const struct pgm_poll *)bp;
ND_TCHECK_SIZE(pgm_poll);
- bp = (const u_char *) (pgm_poll + 1);
+ bp += sizeof(struct pgm_poll);
switch (GET_BE_U_2(pgm_poll->pgmp_nla_afi)) {
case AFNUM_IP:
break;
default:
goto invalid;
- break;
}
ivl = GET_BE_U_4(bp);
case PGM_POLR: {
const struct pgm_polr *polr_msg;
- polr_msg = (const struct pgm_polr *)(pgm + 1);
+ polr_msg = (const struct pgm_polr *)bp;
ND_TCHECK_SIZE(polr_msg);
ND_PRINT("POLR seq %u round %u",
GET_BE_U_4(polr_msg->pgmp_seq),
GET_BE_U_2(polr_msg->pgmp_round));
- bp = (const u_char *) (polr_msg + 1);
+ bp += sizeof(struct pgm_polr);
break;
}
case PGM_ODATA: {
const struct pgm_data *odata;
- odata = (const struct pgm_data *)(pgm + 1);
+ odata = (const struct pgm_data *)bp;
ND_PRINT("ODATA trail %u seq %u",
GET_BE_U_4(odata->pgmd_trailseq),
GET_BE_U_4(odata->pgmd_seq));
- bp = (const u_char *) (odata + 1);
+ bp += sizeof(struct pgm_data);
break;
}
case PGM_RDATA: {
const struct pgm_data *rdata;
- rdata = (const struct pgm_data *)(pgm + 1);
+ rdata = (const struct pgm_data *)bp;
ND_PRINT("RDATA trail %u seq %u",
GET_BE_U_4(rdata->pgmd_trailseq),
GET_BE_U_4(rdata->pgmd_seq));
- bp = (const u_char *) (rdata + 1);
+ bp += sizeof(struct pgm_data);
break;
}
const struct pgm_nak *nak;
char source_buf[INET6_ADDRSTRLEN], group_buf[INET6_ADDRSTRLEN];
- nak = (const struct pgm_nak *)(pgm + 1);
+ nak = (const struct pgm_nak *)bp;
ND_TCHECK_SIZE(nak);
- bp = (const u_char *) (nak + 1);
+ bp += sizeof(struct pgm_nak);
/*
* Skip past the source, saving info along the way
break;
default:
goto invalid;
- break;
}
/*
break;
default:
goto invalid;
- break;
}
/*
break;
}
- case PGM_ACK: {
+ case PGM_ACK:
+ case PGM_ACK2: {
const struct pgm_ack *ack;
- ack = (const struct pgm_ack *)(pgm + 1);
+ ack = (const struct pgm_ack *)bp;
ND_TCHECK_SIZE(ack);
ND_PRINT("ACK seq %u",
GET_BE_U_4(ack->pgma_rx_max_seq));
- bp = (const u_char *) (ack + 1);
+ bp += sizeof(struct pgm_ack);
break;
}
break;
default:
goto invalid;
- break;
}
ND_PRINT(" REDIRECT %s", nla_buf);
case PGM_OPT_PATH_NLA:
ND_PRINT(" PATH_NLA [%u]", opt_len);
- bp += opt_len;
+ bp += opt_len - 2;
opts_len -= opt_len;
break;
case PGM_OPT_CR:
ND_PRINT(" CR");
- bp += opt_len;
+ bp += opt_len - 2;
opts_len -= opt_len;
break;
break;
default:
goto invalid;
- break;
}
ND_PRINT(" PGMCC DATA %u %s", offset, nla_buf);
break;
default:
goto invalid;
- break;
}
ND_PRINT(" PGMCC FEEDBACK %u %s", offset, nla_buf);
default:
ND_PRINT(" OPT_%02X [%u] ", opt_type, opt_len);
- bp += opt_len;
+ bp += opt_len - 2;
opts_len -= opt_len;
break;
}
projects / tcpdump / blobdiff