diff options
author | Anthony Coddington <anthony.coddington@endace.com> | 2017-06-01 20:34:25 +1200 |
---|---|---|
committer | Guy Harris <guy@alum.mit.edu> | 2017-07-17 02:08:52 +0000 |
commit | f3181f706b39955a4f4bc26f1d6d75166a67c235 (patch) | |
tree | 4d1d9d5ab2a17a05219211537ba15693d46cd9b4 /wiretap/erf.c | |
parent | 32ec45dc92a699e34e23197d2de48a043dfad426 (diff) |
ERF_TYPE_META write and comment support
Support per-packet comments in ERF_TYPE_META through a new Anchor ID
extension header with per-Host unique 48-bit Anchor ID which links an
ERF_TYPE_META record with a packet record. There may be more than one
Anchor ID associated with a packet, where they are grouped by Host ID
extension header in the extension header list. Like other ERF_TYPE_META
existing comments should not be overwritten and instead a new record
generated. See erf_write_anchor_meta_update_phdr() for detailed comments
on the extension header stack required.
As Wireshark only supports one comment currently, use the one one with
the latest metadata generation time (gen_time). Do this for capture
comment too.
Write various wtap metadata in periodic per-second ERF_TYPE_META records
if non-WTAP_ENCAP_ERF or we have an updated capture comment.
Refactor erf_dump to create fake ERF header first then follow common
pseudoheadr and payload write code rather than two separate code paths.
Support an ERF_HOST_ID environment variable to define Wireshark's Host
ID when writing. Defaults to 0 for now.
ERF dissector updates to support Anchor ID extension header with basic
frame linking.
Update ERF_TYPE_META naming and descriptions to official name
(Provenance)
Core changes:
Add has_comment_changed to wtap_pkthdr, TRUE when a packet
opt_comment has unsaved changes by the user.
Add needs_reload to wtap_dumper which forces a full reload of the file
on save, otherwise wireshark gets confused by additional packets being
written.
Change-Id: I0bb04411548c7bcd2d6ed82af689fbeed104546c
Ping-Bug: 12303
Reviewed-on: https://code.wireshark.org/review/21873
Petri-Dish: Anders Broman <a.broman58@gmail.com>
Tested-by: Petri Dish Buildbot <buildbot-no-reply@wireshark.org>
Reviewed-by: Stephen Donnelly <stephen.donnelly@endace.com>
Reviewed-by: Guy Harris <guy@alum.mit.edu>
Diffstat (limited to 'wiretap/erf.c')
-rw-r--r-- | wiretap/erf.c | 1726 |
1 files changed, 1590 insertions, 136 deletions
diff --git a/wiretap/erf.c b/wiretap/erf.c index 702420b55d..d7e846065d 100644 --- a/wiretap/erf.c +++ b/wiretap/erf.c @@ -57,20 +57,55 @@ #include "pcapng.h" #include "erf.h" +struct erf_anchor_mapping { + guint64 host_id; + guint64 anchor_id; + guint64 gen_time; + gchar *comment; +}; + static gboolean erf_read_header(wtap *wth, FILE_T fh, struct wtap_pkthdr *phdr, erf_header_t *erf_header, int *err, gchar **err_info, guint32 *bytes_read, - guint32 *packet_size); + guint32 *packet_size, + GPtrArray *anchor_mappings_to_update); static gboolean erf_read(wtap *wth, int *err, gchar **err_info, gint64 *data_offset); static gboolean erf_seek_read(wtap *wth, gint64 seek_off, struct wtap_pkthdr *phdr, Buffer *buf, int *err, gchar **err_info); static void erf_close(wtap *wth); -static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_header *pseudo_header, guint32 packet_size); + +static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_header *pseudo_header, guint32 packet_size, GPtrArray *anchor_mappings_to_update); +static int erf_update_anchors_from_header(erf_t *erf_priv, struct wtap_pkthdr *phdr, union wtap_pseudo_header *pseudo_header, guint64 host_id, GPtrArray *anchor_mappings_to_update); + +typedef struct { + gboolean write_next_extra_meta; + gboolean last_meta_periodic; + guint64 host_id; + guint64 implicit_host_id; + guint64 prev_frame_ts; + guint8 prev_erf_type; + guint64 gen_time; + guint32 first_frame_time_sec; + guint32 prev_inserted_time_sec; + gchar* user_comment_ptr; + GPtrArray* periodic_sections; + GArray *periodic_extra_ehdrs; + GRand *rand; +} erf_dump_t; + +erf_dump_t* erf_dump_priv_create(void); +static void erf_dump_priv_free(erf_dump_t *dump_priv); +static gboolean erf_dump_priv_compare_capture_comment(wtap_dumper *wdh, erf_dump_t *dump_priv,const union wtap_pseudo_header *pseudo_header, const guint8 *pd); +static gboolean erf_comment_to_sections(wtap_dumper *wdh, guint16 section_type, guint16 section_id, gchar *comment, GPtrArray *sections); +static gboolean erf_wtap_info_to_sections(wtap_dumper *wdh, GPtrArray *sections); +static gboolean get_user_comment_string(wtap_dumper *wdh, gchar** user_comment_ptr); + +static gboolean erf_write_meta_record(wtap_dumper *wdh, erf_dump_t *dump_priv, guint64 timestamp, GPtrArray *sections, GArray *extra_ehdrs, int *err); static const struct { int erf_encap_value; @@ -89,7 +124,9 @@ static const struct { #define NUM_ERF_ENCAPS (sizeof erf_to_wtap_map / sizeof erf_to_wtap_map[0]) #define ERF_META_TAG_HEADERLEN 4 -#define ERF_META_TAG_ALIGNED_LENGTH(taglength) ((((guint32)taglength + 0x3U) & ~0x3U) + ERF_META_TAG_HEADERLEN) +#define ERF_META_TAG_TOTAL_ALIGNED_LENGTH(taglength) ((((guint32)taglength + 0x3U) & ~0x3U) + ERF_META_TAG_HEADERLEN) +#define ERF_META_TAG_ALIGNED_LENGTH(taglength) ((((guint32)taglength + 0x3U) & ~0x3U)) +#define ERF_PADDING_TO_8(len) ((8 - len % 8) % 8) struct erf_if_info { int if_index; @@ -109,10 +146,20 @@ struct erf_if_mapping { struct erf_if_info interfaces[4]; gchar *module_filter_str; + /*here because we could have captures from multiple hosts in the file*/ + gchar *capture_filter_str; gint8 module_fcs_len; guint32 module_snaplen; int interface_metadata; - gboolean module_metadata; + guint64 interface_gentime; + guint64 module_gentime; +}; + +struct erf_meta_section { + guint16 type; + guint16 section_id; + guint16 section_length; + GPtrArray *tags; }; struct erf_meta_tag { @@ -132,9 +179,41 @@ struct erf_meta_read_state { guint16 parentsectiontype; guint16 parentsectionid; + guint64 gen_time; + int interface_metadata; }; +static gboolean erf_wtap_blocks_to_erf_sections(wtap_block_t block, GPtrArray *sections, guint16 section_type, guint16 section_id, wtap_block_foreach_func func); + +static guint32 erf_meta_read_tag(struct erf_meta_tag*, guint8*, guint32); + +static guint erf_anchor_mapping_hash(gconstpointer key) { + const struct erf_anchor_mapping *anchor_map = (const struct erf_anchor_mapping*) key; + + return ((guint32)anchor_map->host_id ^ (guint32)anchor_map->anchor_id); + +} + +static gboolean erf_anchor_mapping_equal(gconstpointer a, gconstpointer b) { + const struct erf_anchor_mapping *anchor_map_a = (const struct erf_anchor_mapping*) a ; + const struct erf_anchor_mapping *anchor_map_b = (const struct erf_anchor_mapping*) b ; + + return (anchor_map_a->host_id) == (anchor_map_b->host_id) && + (anchor_map_a->anchor_id & ERF_EXT_HDR_TYPE_ANCHOR_ID) == (anchor_map_b->anchor_id & ERF_EXT_HDR_TYPE_ANCHOR_ID); +} + +static void erf_anchor_mapping_destroy(gpointer key) { + struct erf_anchor_mapping *anchor_map = (struct erf_anchor_mapping*) key; + + if(anchor_map->comment != NULL) { + g_free(anchor_map->comment); + anchor_map->comment = NULL; + } + g_free(anchor_map); + anchor_map = NULL; +} + static gboolean erf_if_mapping_equal(gconstpointer a, gconstpointer b) { const struct erf_if_mapping *if_map_a = (const struct erf_if_mapping*) a; @@ -169,8 +248,7 @@ static struct erf_if_mapping* erf_if_mapping_create(guint64 host_id, guint8 sour int i = 0; struct erf_if_mapping *if_map = NULL; - if_map = (struct erf_if_mapping*) g_malloc(sizeof(struct erf_if_mapping)); - memset(if_map, 0, sizeof(struct erf_if_mapping)); + if_map = (struct erf_if_mapping*) g_malloc0(sizeof(struct erf_if_mapping)); if_map->host_id = host_id; if_map->source_id = source_id; @@ -182,20 +260,22 @@ static struct erf_if_mapping* erf_if_mapping_create(guint64 host_id, guint8 sour if_map->module_fcs_len = -1; if_map->module_snaplen = (guint32) -1; - /* everything else 0 by memset */ + /* everything else 0 by g_malloc0*/ return if_map; } + erf_t *erf_priv_create(void) { erf_t *erf_priv; erf_priv = (erf_t*) g_malloc(sizeof(erf_t)); + erf_priv->anchor_map = g_hash_table_new_full(erf_anchor_mapping_hash, erf_anchor_mapping_equal, erf_anchor_mapping_destroy, NULL); erf_priv->if_map = g_hash_table_new_full(erf_if_mapping_hash, erf_if_mapping_equal, erf_if_mapping_destroy, NULL); erf_priv->implicit_host_id = ERF_META_HOST_ID_IMPLICIT; - erf_priv->capture_metadata = FALSE; - erf_priv->host_metadata = FALSE; + erf_priv->capture_gentime = 0; + erf_priv->host_gentime = 0; return erf_priv; } @@ -204,6 +284,7 @@ erf_t* erf_priv_free(erf_t* erf_priv) { if (erf_priv) { + g_hash_table_destroy(erf_priv->anchor_map); g_hash_table_destroy(erf_priv->if_map); g_free(erf_priv); } @@ -211,6 +292,77 @@ erf_t* erf_priv_free(erf_t* erf_priv) return NULL; } +static void erf_dump_priv_free(erf_dump_t *dump_priv) { + if(dump_priv) { + if(dump_priv->periodic_sections) { + g_ptr_array_free(dump_priv->periodic_sections, TRUE); + } + if(dump_priv->periodic_extra_ehdrs) { + g_array_free(dump_priv->periodic_extra_ehdrs, TRUE); + } + if(dump_priv->user_comment_ptr) { + g_free(dump_priv->user_comment_ptr); + } + + g_free(dump_priv->rand); + + g_free(dump_priv); + } + +} + +static void erf_meta_section_free(gpointer data) { + struct erf_meta_section *section_ptr = (struct erf_meta_section*) data; + if (section_ptr) { + g_ptr_array_free(section_ptr->tags, TRUE); + section_ptr->tags = NULL; + } + g_free(section_ptr); +} + +static void erf_meta_tag_free(gpointer data) { + struct erf_meta_tag *tag_ptr = (struct erf_meta_tag*) data; + if (tag_ptr) { + g_free(tag_ptr->value); + tag_ptr->value = NULL; + } + g_free(tag_ptr); +} + + +static gboolean erf_dump_finish(struct wtap_dumper *wdh, int *err) { + erf_dump_t *dump_priv = (erf_dump_t*)wdh->priv; + gboolean ret = TRUE; + + /* Write final metadata record. There are some corner cases where we should + * do this (file <1 second, last record was ERF_TYPE_META with an out of date + * comment) and there is no harm doing this always if we have already written + * some metadata. */ + if(dump_priv->write_next_extra_meta) { + if (!dump_priv->periodic_sections) { + dump_priv->periodic_sections = g_ptr_array_new_with_free_func(erf_meta_section_free); + if (dump_priv->prev_erf_type == ERF_TYPE_META && dump_priv->last_meta_periodic) { + erf_comment_to_sections(wdh, ERF_META_SECTION_CAPTURE, 0, dump_priv->user_comment_ptr, dump_priv->periodic_sections); + } else { + /* If we get here, metadata record was not found in the first ~1 sec + * but we have either a capture comment or a non-ERF file (see + * erf_dump_open) */ + erf_wtap_info_to_sections(wdh, dump_priv->periodic_sections); + } + } + + if (!erf_write_meta_record(wdh, dump_priv, dump_priv->prev_frame_ts, dump_priv->periodic_sections, dump_priv->periodic_extra_ehdrs, err)) ret = FALSE; + } + + /* Clean up */ + erf_dump_priv_free(dump_priv); + /* Avoid double freeing by setting it to NULL*/ + wdh->priv = NULL; + + return ret; + +} + static void erf_free_data(gpointer data, gpointer user_data _U_) { @@ -425,31 +577,41 @@ static gboolean erf_read(wtap *wth, int *err, gchar **err_info, { erf_header_t erf_header; guint32 packet_size, bytes_read; + GPtrArray *anchor_mappings_to_update; *data_offset = file_tell(wth->fh); + anchor_mappings_to_update = g_ptr_array_new_with_free_func(erf_anchor_mapping_destroy); + do { if (!erf_read_header(wth, wth->fh, &wth->phdr, &erf_header, - err, err_info, &bytes_read, &packet_size)) { + err, err_info, &bytes_read, &packet_size, + anchor_mappings_to_update)) { + g_ptr_array_free(anchor_mappings_to_update, TRUE); return FALSE; } if (!wtap_read_packet_bytes(wth->fh, wth->frame_buffer, packet_size, - err, err_info)) + err, err_info)) { + g_ptr_array_free(anchor_mappings_to_update, TRUE); return FALSE; + } /* - * If MetaERF, frame buffer could hold the meta erf tags. Only look until - * we have seen a description of every interface. + * If Provenance metadata record, frame buffer could hold the meta erf tags. + * It can also contain per packet comments which can be associated to another + * frame. */ if ((erf_header.type & 0x7F) == ERF_TYPE_META && packet_size > 0) { - populate_summary_info((erf_t*) wth->priv, wth, &wth->phdr.pseudo_header, packet_size); + populate_summary_info((erf_t*) wth->priv, wth, &wth->phdr.pseudo_header, packet_size, anchor_mappings_to_update); } } while ( erf_header.type == ERF_TYPE_PAD ); + g_ptr_array_free(anchor_mappings_to_update, TRUE); + return TRUE; } @@ -459,27 +621,54 @@ static gboolean erf_seek_read(wtap *wth, gint64 seek_off, { erf_header_t erf_header; guint32 packet_size; + GPtrArray *anchor_mappings_to_update; if (file_seek(wth->random_fh, seek_off, SEEK_SET, err) == -1) return FALSE; + anchor_mappings_to_update = g_ptr_array_new_with_free_func(erf_anchor_mapping_destroy); + do { if (!erf_read_header(wth, wth->random_fh, phdr, &erf_header, - err, err_info, NULL, &packet_size)) + err, err_info, NULL, &packet_size, anchor_mappings_to_update)) { + g_ptr_array_free(anchor_mappings_to_update, TRUE); return FALSE; + } } while ( erf_header.type == ERF_TYPE_PAD ); + g_ptr_array_free(anchor_mappings_to_update, TRUE); + return wtap_read_packet_bytes(wth->random_fh, buf, packet_size, err, err_info); } +static struct erf_anchor_mapping* erf_find_anchor_mapping(erf_t *priv, + guint64 host_id, + guint64 anchor_id) +{ + struct erf_anchor_mapping mapping = { + host_id, + anchor_id, + 0, + NULL + }; + + if (!priv) { + return NULL; + } + + return (struct erf_anchor_mapping*)g_hash_table_lookup(priv->anchor_map, &mapping); + +} + static gboolean erf_read_header(wtap *wth, FILE_T fh, struct wtap_pkthdr *phdr, erf_header_t *erf_header, int *err, gchar **err_info, guint32 *bytes_read, - guint32 *packet_size) + guint32 *packet_size, + GPtrArray *anchor_mappings_to_update) { union wtap_pseudo_header *pseudo_header = &phdr->pseudo_header; guint8 erf_exhdr[8]; @@ -491,6 +680,7 @@ static gboolean erf_read_header(wtap *wth, FILE_T fh, guint32 skiplen = 0; int i = 0; int max = sizeof(pseudo_header->erf.ehdr_list)/sizeof(struct erf_ehdr); + erf_t *priv = (erf_t*)wth->priv; guint64 host_id = ERF_META_HOST_ID_IMPLICIT; guint8 source_id = 0; @@ -537,7 +727,7 @@ static gboolean erf_read_header(wtap *wth, FILE_T fh, * purposes, but currently ft_specific_record_phdr clashes with erf_mc_phdr * and the PCAP-NG dumper assumes it is a PCAP-NG block type. Ideally we * would register a block handler with PCAP-NG and write out the closest - * PCAP-NG block, or a custom block/MetaERF record. + * PCAP-NG block, or a custom block/Provenance record. * */ #if 0 @@ -547,7 +737,7 @@ static gboolean erf_read_header(wtap *wth, FILE_T fh, * What to do about ENCAP_ERF in PCAP/PCAP-NG? Filetype dissector is * chosen by wth->file_type_subtype? */ - /* For now just treat all MetaERF records as reports */ + /* For now just treat all Provenance records as reports */ phdr->rec_type = REC_TYPE_FT_SPECIFIC_REPORT; /* XXX: phdr ft_specific_record_phdr? */ } @@ -593,7 +783,7 @@ static gboolean erf_read_header(wtap *wth, FILE_T fh, * XXX: Only want first Source ID and Host ID, and want to preserve HID n SID 0 (see * erf_populate_interface) */ - switch (type & 0x7f) { + switch (type & 0x7FU) { case ERF_EXT_HDR_TYPE_HOST_ID: if (!host_id_found) host_id = erf_exhdr_sw & ERF_EHDR_HOST_ID_MASK; @@ -601,17 +791,23 @@ static gboolean erf_read_header(wtap *wth, FILE_T fh, host_id_found = TRUE; /* Fall through */ case ERF_EXT_HDR_TYPE_FLOW_ID: + /* Source ID is present in both Flow ID and Host ID extension headers */ if (!source_id) source_id = (erf_exhdr_sw >> 48) & 0xff; break; + case ERF_EXT_HDR_TYPE_ANCHOR_ID: + /* handled below*/ + break; } - i++; } /* XXX: erf_priv pointer needs to change if used as common function for other dissectors! */ phdr->interface_id = (guint) erf_populate_interface((erf_t*) wth->priv, wth, pseudo_header, host_id, source_id, if_num); + /* Try to find comment links using Anchor ID. Done here after we found the first Host ID and have updated the implicit Host ID. */ + erf_update_anchors_from_header(priv, phdr, pseudo_header, host_id, anchor_mappings_to_update); + switch (erf_header->type & 0x7F) { case ERF_TYPE_IPV4: case ERF_TYPE_IPV6: @@ -791,6 +987,777 @@ static gboolean erf_write_phdr(wtap_dumper *wdh, int encap, const union wtap_pse return TRUE; } + +static void erf_dump_priv_init_gen_time(erf_dump_t *dump_priv) { + GTimeVal real_time; + + g_get_current_time(&real_time); + /* Convert TimeVal to ERF timestamp */ + dump_priv->gen_time = ((guint64) real_time.tv_sec << 32) + ((guint64) real_time.tv_usec << 32) / 1000 / 1000; +} + + +static void erf_write_wtap_option_to_capture_tag(wtap_block_t block _U_, + guint option_id, + wtap_opttype_e option_type _U_, + wtap_optval_t *optval, + void* user_data) { + + struct erf_meta_section *section_ptr = (struct erf_meta_section*) user_data; + struct erf_meta_tag *tag_ptr = NULL; + + tag_ptr = (struct erf_meta_tag*) g_malloc0(sizeof(struct erf_meta_tag)); + + switch(option_id) { + case OPT_SHB_USERAPPL: + tag_ptr->type = ERF_META_TAG_app_name; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + case OPT_COMMENT: + tag_ptr->type = ERF_META_TAG_comment; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + default: + erf_meta_tag_free(tag_ptr); + tag_ptr = NULL; + return; + } + + if (tag_ptr) + g_ptr_array_add(section_ptr->tags, tag_ptr); +} + +static void erf_write_wtap_option_to_host_tag(wtap_block_t block _U_, + guint option_id, + wtap_opttype_e option_type _U_, + wtap_optval_t *optval, + void* user_data) { + + struct erf_meta_section *section_ptr = (struct erf_meta_section*) user_data; + struct erf_meta_tag *tag_ptr = NULL; + + tag_ptr = (struct erf_meta_tag*) g_malloc0(sizeof(struct erf_meta_tag)); + + switch(option_id) { + case OPT_SHB_HARDWARE: + tag_ptr->type = ERF_META_TAG_cpu; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + case OPT_SHB_OS: + tag_ptr->type = ERF_META_TAG_os; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + default: + erf_meta_tag_free(tag_ptr); + tag_ptr = NULL; + return; + } + + if (tag_ptr) + g_ptr_array_add(section_ptr->tags, tag_ptr); + +} + +static void erf_write_wtap_option_to_interface_tag(wtap_block_t block _U_, + guint option_id, + wtap_opttype_e option_type _U_, + wtap_optval_t *optval, + void* user_data) { + + struct erf_meta_section *section_ptr = (struct erf_meta_section*) user_data; + struct erf_meta_tag *tag_ptr = NULL; + + tag_ptr = (struct erf_meta_tag*) g_malloc0(sizeof(struct erf_meta_tag)); + + switch(option_id) { + case OPT_COMMENT: + tag_ptr->type = ERF_META_TAG_comment; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + case OPT_IDB_NAME: + tag_ptr->type = ERF_META_TAG_name; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + case OPT_IDB_DESCR: + tag_ptr->type = ERF_META_TAG_descr; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + case OPT_IDB_OS: + tag_ptr->type = ERF_META_TAG_os; + tag_ptr->value = (guint8*)g_strdup(optval->stringval); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + break; + case OPT_IDB_TSOFFSET: + tag_ptr->type = ERF_META_TAG_ts_offset; + tag_ptr->length = 8; + tag_ptr->value = (guint8*)g_malloc(sizeof(optval->uint64val)); + /* convert to relative ERF timestamp */ + phtolell(tag_ptr->value, optval->uint64val << 32); + break; + case OPT_IDB_SPEED: + tag_ptr->type = ERF_META_TAG_if_speed; + tag_ptr->length = 8; + tag_ptr->value = (guint8*)g_malloc(sizeof(optval->uint64val)); + phtonll(tag_ptr->value, optval->uint64val); + break; + case OPT_IDB_IP4ADDR: + tag_ptr->type = ERF_META_TAG_if_ipv4; + tag_ptr->length = 4; + tag_ptr->value = (guint8*)g_malloc(sizeof(optval->ipv4val)); + memcpy(tag_ptr->value, &optval->ipv4val, sizeof(optval->ipv4val)); + break; + case OPT_IDB_IP6ADDR: + tag_ptr->type = ERF_META_TAG_if_ipv6; + tag_ptr->length = 16; + tag_ptr->value = (guint8*)g_malloc(sizeof(optval->ipv6val)); + memcpy(tag_ptr->value, &optval->ipv6val, sizeof(optval->ipv6val)); + break; + case OPT_IDB_FILTER: + { + wtapng_if_descr_filter_t *filter; + tag_ptr->type = 0xF800; + filter = (wtapng_if_descr_filter_t*)&optval->customval; + if(filter->if_filter_str) { + tag_ptr->type = ERF_META_TAG_filter; + tag_ptr->value = (guint8*)g_strdup(filter->if_filter_str); + tag_ptr->length = (guint16)strlen((char*)tag_ptr->value); + } + } + break; + case OPT_IDB_FCSLEN: + tag_ptr->type = ERF_META_TAG_fcs_len; + tag_ptr->length = 4; + tag_ptr->value = (guint8*)g_malloc(tag_ptr->length); + phtonl(tag_ptr->value, (guint32)optval->uint8val); + break; + /* TODO: Don't know what to do with these yet */ + case OPT_IDB_EUIADDR: +#if 0 + tag_ptr->type = ERF_META_TAG_if_eui; + tag_ptr->length = 8; + tag_ptr->value = (guint8*)g_malloc(sizeof(optval->eui64val)); + memcpy(tag_ptr->value, &optval->euival, sizeof(optval->eui64val)); + break; +#endif + case OPT_IDB_MACADDR: +#if 0 + tag_ptr->type = ERF_META_TAG_if_mac; + tag_ptr->length = 6; + /*value same format as PCAP-NG (6-byte canonical, padded by write + * function automatically to 32-bit boundary)*/ + tag_ptr->value = (guint8*)g_malloc(sizeof(optval->macval)); + memcpy(tag_ptr->value, &optval->macval, sizeof(optval->macval)); + break; +#endif + case OPT_IDB_TSRESOL: + case OPT_IDB_TZONE: + /* Fall through */ + default: + erf_meta_tag_free(tag_ptr); + tag_ptr = NULL; + break; + } + + if (tag_ptr) + g_ptr_array_add(section_ptr->tags, tag_ptr); + +} + +static void erf_populate_section_length_by_tags(struct erf_meta_section *section_ptr) { + guint i = 0; + struct erf_meta_tag *tag_ptr; + + section_ptr->section_length = 8; + + for(;i < section_ptr->tags->len; i++) { + tag_ptr = (struct erf_meta_tag*)g_ptr_array_index(section_ptr->tags, i); + section_ptr->section_length += ERF_META_TAG_TOTAL_ALIGNED_LENGTH(tag_ptr->length); + } +} + +/** + * @brief Converts a wtap_block_t block to ERF metadata sections + * @param block a wtap_block_t block + * @param sections pointer to a GPtrArray containing pointers to sections + * @param section_type the pre-specified section_type + * @param section_id Section ID to assign + * @param func a wtap_block_foreach_func call back function to specify + * what needs to be done on the block + * @return TRUE if success, FALSE if failed + */ +static gboolean erf_wtap_blocks_to_erf_sections(wtap_block_t block, GPtrArray *sections, guint16 section_type, guint16 section_id, wtap_block_foreach_func func) { + + if(!block || !sections || !func) { + return FALSE; + } + + struct erf_meta_section *section_ptr; + + section_ptr = (struct erf_meta_section*) g_malloc(sizeof(struct erf_meta_section)); + section_ptr->tags = g_ptr_array_new_with_free_func(erf_meta_tag_free); + section_ptr->type = section_type; + section_ptr->section_id = section_id; + + wtap_block_foreach_option(block, func, (void*)section_ptr); + erf_populate_section_length_by_tags(section_ptr); + g_ptr_array_add(sections, section_ptr); + + return TRUE; +} + + +static gboolean erf_meta_write_tag(wtap_dumper *wdh, struct erf_meta_tag *tag_ptr, int *err) { + + guint16 data[2]; + guint pad = 0; + /* we only need to pad up to 32 bits*/ + guint32 padbuf = 0; + + pad = ERF_META_TAG_ALIGNED_LENGTH(tag_ptr->length) - tag_ptr->length; + data[0] = g_htons(tag_ptr->type); + data[1] = g_htons(tag_ptr->length); + + if(!wtap_dump_file_write(wdh, data, sizeof(data), err)) return FALSE; + wdh->bytes_dumped += sizeof(data); + + if(!wtap_dump_file_write(wdh, tag_ptr->value, tag_ptr->length, err)) return FALSE; + wdh->bytes_dumped += tag_ptr->length; + + if(pad) { + if(!wtap_dump_file_write(wdh, &padbuf, pad, err)) return FALSE; + wdh->bytes_dumped += pad; + } + + return TRUE; + +} + +static gboolean erf_meta_write_section(wtap_dumper *wdh, struct erf_meta_section *section_ptr, int *err) { + + struct erf_meta_tag *tag_ptr; + guint i; + guint16 data[4]; + + data[0] = g_htons(section_ptr->type); + data[1] = g_htons(4); /*section header length*/ + data[2] = g_htons(section_ptr->section_id); + data[3] = g_htons(section_ptr->section_length); + + if(!wtap_dump_file_write(wdh, data, sizeof(data), err)) return FALSE; + wdh->bytes_dumped += sizeof(data); + + for(i = 0; i < section_ptr->tags->len; i++) { + tag_ptr = (struct erf_meta_tag*)g_ptr_array_index(section_ptr->tags, i); + if(!erf_meta_write_tag(wdh, tag_ptr, err)) return FALSE; + } + + return TRUE; + +} + +static gboolean erf_wtap_info_to_sections(wtap_dumper *wdh, GPtrArray *sections) { + wtap_block_t block; + guint i = 0; + + block = g_array_index(wdh->shb_hdrs, wtap_block_t, 0); + erf_wtap_blocks_to_erf_sections(block, sections, ERF_META_SECTION_CAPTURE, 0, erf_write_wtap_option_to_capture_tag); + + block = g_array_index(wdh->shb_hdrs, wtap_block_t, 0); + erf_wtap_blocks_to_erf_sections(block, sections, ERF_META_SECTION_HOST, 0, erf_write_wtap_option_to_host_tag); + + /*TODO: support >4 interfaces by using more Source IDs. Affects more than this + * function as need more metadata records. Just dump them all out for now. */ + for(i = 0; i < wdh->interface_data->len; i++) { + block = g_array_index(wdh->interface_data, wtap_block_t, i); + erf_wtap_blocks_to_erf_sections(block, sections, ERF_META_SECTION_INTERFACE, (gint16)i+1, erf_write_wtap_option_to_interface_tag); + } + + return TRUE; +} + +static gboolean erf_comment_to_sections(wtap_dumper *wdh _U_, guint16 section_type, guint16 section_id, gchar *comment, GPtrArray *sections){ + struct erf_meta_section *section_ptr; + struct erf_meta_tag *comment_tag_ptr = NULL; + struct erf_meta_tag *user_tag_ptr = NULL; + const gchar *user = NULL; + + /* Generate the section */ + section_ptr = (struct erf_meta_section*) g_malloc(sizeof(struct erf_meta_section)); + section_ptr->type = section_type; + section_ptr->section_id = section_id; + section_ptr->tags = g_ptr_array_new_with_free_func(erf_meta_tag_free); + + /* Generate the comment tag */ + comment_tag_ptr = (struct erf_meta_tag*) g_malloc(sizeof(struct erf_meta_tag)); + comment_tag_ptr->type = ERF_META_TAG_comment; + /* XXX: if the comment has been cleared write the empty string (which + * conveniently is all a zero length tag which means the value is + * invalidated) */ + comment_tag_ptr->value = (guint8*)g_strdup(comment ? comment : ""); + comment_tag_ptr->length = (guint16)strlen((char*)comment_tag_ptr->value); + g_ptr_array_add(section_ptr->tags, comment_tag_ptr); + + user = g_get_user_name(); + if (user) { + /* Generate username tag */ + user_tag_ptr = (struct erf_meta_tag*) g_malloc(sizeof(struct erf_meta_tag)); + user_tag_ptr->type = ERF_META_TAG_user; + user_tag_ptr->value = (guint8*)g_strdup(user); + user_tag_ptr->length = (guint16)strlen((char*)user_tag_ptr->value); + g_ptr_array_add(section_ptr->tags, user_tag_ptr); + } + + erf_populate_section_length_by_tags(section_ptr); + + g_ptr_array_add(sections, section_ptr); + + return TRUE; +} + +static guint64 erf_get_random_anchor_id(erf_dump_t *dump_priv) { + return (((guint64)g_rand_int(dump_priv->rand) << 32) | (guint64)g_rand_int(dump_priv->rand)) >> 16; +} + +static guint64 erf_metaid_ext_hdr(guint8 exthdr_type, guint64 id, guint8 srcid_flags) { + guint64 ext_hdr; + + ext_hdr = id & ERF_EHDR_HOST_ID_MASK; + ext_hdr |= ((guint64)srcid_flags) << 48; + ext_hdr |= ((guint64)exthdr_type) << 56; + + return ext_hdr; +} +#define erf_host_id_ext_hdr(host_id, source_id) erf_metaid_ext_hdr(ERF_EXT_HDR_TYPE_HOST_ID, host_id, source_id) +#define erf_anchor_id_ext_hdr(anchor_id, flags) erf_metaid_ext_hdr(ERF_EXT_HDR_TYPE_ANCHOR_ID, anchor_id, flags) + +static inline gboolean erf_add_ext_hdr_to_list(guint64 ext_hdr, guint64 comparison_mask, GArray *extra_ehdrs) { + /* check for existing Host ID in set and add */ + guint i = 0; + struct erf_ehdr ehdr_tmp; + struct erf_ehdr *ehdr_ptr = NULL; + + if (!extra_ehdrs) + return FALSE; + + ext_hdr = ext_hdr & ~ERF_EHDR_MORE_EXTHDR_MASK; + if (comparison_mask == 0) + comparison_mask = G_MAXUINT64; + + comparison_mask &= ~ERF_EHDR_MORE_EXTHDR_MASK; + + for (i = 0; i < extra_ehdrs->len; i++) { + ehdr_ptr = &g_array_index(extra_ehdrs, struct erf_ehdr, i); + /* Check if we already have this Host ID extension header */ + if (ext_hdr == (ehdr_ptr->ehdr & comparison_mask)) { + return TRUE; + } + } + + /* set more flag on last extension header */ + if (ehdr_ptr) { + ehdr_ptr->ehdr |= ERF_EHDR_MORE_EXTHDR_MASK; + } + + ehdr_tmp.ehdr = ext_hdr; /*more flag already cleared above*/ + g_array_append_val(extra_ehdrs, ehdr_tmp); + + return TRUE; +} + +static inline gboolean erf_append_ext_hdr_to_list(guint64 ext_hdr, GArray *extra_ehdrs) { + struct erf_ehdr ehdr_tmp; + + if (!extra_ehdrs) + return FALSE; + + ehdr_tmp.ehdr = ext_hdr & ~ERF_EHDR_MORE_EXTHDR_MASK; + + /* set more flag on last extension header */ + if (extra_ehdrs->len) { + g_array_index(extra_ehdrs, struct erf_ehdr, extra_ehdrs->len - 1).ehdr |= ERF_EHDR_MORE_EXTHDR_MASK; + } + + g_array_append_val(extra_ehdrs, ehdr_tmp); + + return TRUE; +} + +static gboolean erf_update_host_id_ext_hdrs_list(erf_dump_t *dump_priv, const union wtap_pseudo_header *pseudo_header, GArray *extra_ehdrs) { + guint8 type; + guint8 erf_type; + int has_more; + guint64 hdr; + int i = 0; + guint8 source_id = 0; + guint64 host_id = 0; + gboolean host_id_found = FALSE; + + if (!extra_ehdrs) + return FALSE; + + erf_type = pseudo_header->erf.phdr.type & 0x7f; + has_more = pseudo_header->erf.phdr.type & 0x80; + + while (has_more && i < MAX_ERF_EHDR) { + hdr = pseudo_header->erf.ehdr_list[i].ehdr; + type = (guint8) (hdr >> 56); + + switch (type & 0x7f) { + case ERF_EXT_HDR_TYPE_HOST_ID: + host_id = hdr & ERF_EHDR_HOST_ID_MASK; + source_id = (hdr >> 48) & 0xff; + + /* Don't add the wireshark Host ID Source ID 0 twice since we already add it to metadata records */ + if (host_id != dump_priv->host_id || source_id != 0) + if (!erf_add_ext_hdr_to_list(hdr, 0, extra_ehdrs)) return FALSE; + + if (!host_id_found) { + /* XXX: Take the opportunity to update the implicit Host ID if we + * don't know it yet. Ideally we should pass this through from the + * reader as a custom option or similar. */ + if (erf_type == ERF_TYPE_META && ((hdr >> 48) & 0xff) > 0) { + if (dump_priv->implicit_host_id == ERF_META_HOST_ID_IMPLICIT) { + dump_priv->implicit_host_id = host_id; + } + } + } + + host_id_found = TRUE; + break; + case ERF_EXT_HDR_TYPE_FLOW_ID: + if (source_id == 0) /* If no Host ID extension header use the first Source ID only */ + source_id = (hdr >> 48) & 0xff; + break; + } + + has_more = type & 0x80; + i++; + } + + /* Add Source ID with implicit Host ID if not found */ + if (!host_id_found) { + guint64 implicit_host_id = dump_priv->implicit_host_id == ERF_META_HOST_ID_IMPLICIT ? 0 : dump_priv->implicit_host_id; + /* Don't add the wireshark Host ID Source ID 0 twice since we already add it to metadata records */ + if (implicit_host_id != dump_priv->host_id || source_id != 0) + if (!erf_add_ext_hdr_to_list(erf_host_id_ext_hdr(implicit_host_id, source_id), 0, extra_ehdrs)) return FALSE; + } + + return TRUE; +} + +/** + * Writes a metadata record with a randomly generated Anchor ID with the + * user comment attached to its comment section, also updates the + * modified frame header to include a Host ID extension header and + * a Anchor ID extension header to link the records together. + * @param wdh the wtap_dumper structure + * @param phdr packet header to get user comment from + * @param mutable_hdr pseudo_header to update with Anchor ID for comment record + * @param err the error value + * @return A gboolean value to indicate whether the dump was successful + */ +static gboolean erf_write_anchor_meta_update_phdr(wtap_dumper *wdh, erf_dump_t *dump_priv, const struct wtap_pkthdr *phdr, union wtap_pseudo_header *mutable_hdr, int *err) { + GArray *meta_ehdrs; + GPtrArray* sections = NULL; + guint8 has_more; + guint8 i = 0; + guint8 ext_hdr_count = 0; + guint8 j = 0; + guint64 host_id_src_hdr = ERF_META_HOST_ID_IMPLICIT; + guint64 host_id_own_hdr = erf_host_id_ext_hdr(dump_priv->host_id, 0); + guint64 flow_id_hdr = 0; + guint64 anchor_id_hdr = 0; + gboolean found_host_id = FALSE; + gboolean found_own_host_id = FALSE; + gboolean found_flow_id = FALSE; + gint new_ext_hdrs = 0; + guint8 insert_idx = 0; + guint8 source_id = 0; + gboolean ret = FALSE; + guint64 implicit_host_id = dump_priv->implicit_host_id == ERF_META_HOST_ID_IMPLICIT ? 0 : dump_priv->implicit_host_id; + + + /* + * There are 3 possible scenarios: + * a. The record has a source Host ID but not our Host ID. We need to add our + * Host ID extension header then our Anchor ID extension header. + * b. The record already has our Host ID extension header on it. We should + * insert the Anchor ID at the end of the list for that Host ID just + * before the next Host ID extension header. + * c. The record has no Host ID extension header at all. We need to add the Host ID + * extension header making the Implicit Host ID explicit before we add our + * one to avoid claiming the packet was captured by us. + */ + + /* + * Extract information from the packet extension header stack + * 1. original source Host ID extension header. + * 2. Anchor ID extension header insertion point (see b., above). + * 3. Flow ID extension header so we can add it for reference to the metadata + * record. + * 4. Enough information to generate an explicit Host ID extension header if + * there wasn't one (see erf_get_source_from_header). + */ + + has_more = mutable_hdr->erf.phdr.type & 0x80; + + while (has_more && (i < MAX_ERF_EHDR)) { + guint64 hdr = mutable_hdr->erf.ehdr_list[i].ehdr; + guint8 type = (guint8) (hdr >> 56); + + switch (type & 0x7f) { + case ERF_EXT_HDR_TYPE_HOST_ID: + /* Set insertion point of anchor ID to be at end of Host ID list (i.e. + * just before the next one). */ + if (found_own_host_id && !insert_idx) + insert_idx = i; + + if ((hdr & ERF_EHDR_HOST_ID_MASK) == dump_priv->host_id){ + found_own_host_id = TRUE; + } + + if (!found_host_id) + host_id_src_hdr = hdr; + + found_host_id = TRUE; + break; + + case ERF_EXT_HDR_TYPE_FLOW_ID: + /*XXX: we only use this when making the implicit host id explicit, + * otherwise we'd need to check the one in Host ID header too*/ + if (source_id == 0) + source_id = (guint8)(hdr >> 48); + + if (!found_flow_id) + flow_id_hdr = hdr; + + found_flow_id = TRUE; + break; + } + + has_more = type & 0x80; + i += 1; + } + + ext_hdr_count = i; + + if (!insert_idx) + insert_idx = i; + + /* Don't need to add our own Host ID twice if it is the same as the implicit*/ + if (!found_host_id && implicit_host_id == dump_priv->host_id) { + found_own_host_id = TRUE; + } + + /* + * Update the packet record pseudo_header with Anchor ID and extension header(s) + */ + new_ext_hdrs = 1 /*anchor id*/ + (found_own_host_id?0:1) + (found_host_id?0:1); + + if(ext_hdr_count + new_ext_hdrs > MAX_ERF_EHDR + || mutable_hdr->erf.phdr.rlen + new_ext_hdrs * 8 > 65535) { + /* Not enough extension header slots to add Anchor ID */ + *err = WTAP_ERR_PACKET_TOO_LARGE; + return FALSE; + } + + mutable_hdr->erf.phdr.rlen += new_ext_hdrs * 8; + + /* Set the more extension headers flag */ + mutable_hdr->erf.phdr.type |= 0x80; + if (insert_idx > 0) { + mutable_hdr->erf.ehdr_list[insert_idx-1].ehdr |= ERF_EHDR_MORE_EXTHDR_MASK; + } + + /* Generate the Anchor ID extension header */ + anchor_id_hdr = erf_anchor_id_ext_hdr(erf_get_random_anchor_id(dump_priv), 0); + + /* Either we can insert Anchor ID at the end of the list for our Host ID or we + * need to append the Host ID(s) and Anchor ID */ + if (insert_idx < ext_hdr_count) { + /* shuffle up any following extension headers FIRST - we know we have room now */ + for (j = ext_hdr_count; j > insert_idx; j--) { + mutable_hdr->erf.ehdr_list[j].ehdr = mutable_hdr->erf.ehdr_list[j-1].ehdr; + } + + /* copy more extension headers bit from previous extension header */ + anchor_id_hdr |= ERF_EHDR_MORE_EXTHDR_MASK; + } + + if(!found_host_id) { + /* No Host ID extension header found and we have an implicit Host ID which + * we want to make explicit */ + + /* XXX: it is important that we know the implicit Host ID here or we end + * up semi-permentantly associating the packet with Host 0 (unknown), we should + * pass it through from the reader. In theory we should be on the + * original capture machine if we have no Host ID extension headers. */ + host_id_src_hdr = erf_host_id_ext_hdr(implicit_host_id, source_id); + mutable_hdr->erf.ehdr_list[insert_idx++].ehdr = ERF_EHDR_SET_MORE_EXTHDR(host_id_src_hdr); + } + + if(!found_own_host_id) { + /* Add our Host ID extension header */ + mutable_hdr->erf.ehdr_list[insert_idx++].ehdr = ERF_EHDR_SET_MORE_EXTHDR(host_id_own_hdr); + } + + /*Add the Anchor ID extension header */ + mutable_hdr->erf.ehdr_list[insert_idx].ehdr = anchor_id_hdr; + + + /* + * Now construct the metadata Anchor record with the same Anchor ID + */ + + meta_ehdrs = g_array_new(FALSE, FALSE, sizeof(struct erf_ehdr)); + + /* We need up to 4 extension headers on the Provenance metadata record */ + /*Required*/ + /* 1. Added by erf_write_meta_record: HostID exthdr to indicate this Anchor + * record was generated by this host. Source ID 0 to avoid changing the + * implicit Host ID. */ + + /* 2. AnchorID exthdr with 'unique' per-host Anchor ID assigned by this host + * (in this case Wireshark). Anchor defintion flag set to 1 to indicate this + * record contains a defintion of the ID, in this case a comment on a single + * packet. Tied to above extension header by ordering like a list */ + erf_append_ext_hdr_to_list(anchor_id_hdr | ERF_EHDR_ANCHOR_ID_DEFINITION_MASK, meta_ehdrs); + + /*Helpful for indexing*/ + /* 3. HostID exthdr with the original Source (first Host ID extension header) of the packet record */ + erf_append_ext_hdr_to_list(host_id_src_hdr, meta_ehdrs); + + /* Flow ID extension header from the packet record if we have one */ + if (found_flow_id) { + /* 4. FlowID exthdr with Flow ID from the packet so a flow search will find the comment + * record too. Must come here so the (redundant here) Source ID is scoped to the + * correct Host ID. */ + /* Clear the stack type just in case something tries to assume we're an IP + * packet without looking at the ERF type. Clear Source ID too just in case + * we're trying to associate with the wrong Host ID. */ + erf_append_ext_hdr_to_list(flow_id_hdr & ~(ERF_EHDR_FLOW_ID_STACK_TYPE_MASK|ERF_EHDR_FLOW_ID_SOURCE_ID_MASK), meta_ehdrs); + } + + /* Generate the metadata payload with the packet comment */ + sections = g_ptr_array_new_with_free_func(erf_meta_section_free); + erf_comment_to_sections(wdh, ERF_META_SECTION_INFO, 0x8000 /*local to record*/, phdr->opt_comment, sections); + + /* Write the metadata record, but not the packet record as what we do depends + * on the WTAP_ENCAP */ + ret = erf_write_meta_record(wdh, dump_priv, mutable_hdr->erf.phdr.ts, sections, meta_ehdrs, err); + g_ptr_array_free(sections, TRUE); + g_array_free(meta_ehdrs, TRUE); + + return ret; +} + +static gboolean erf_write_meta_record(wtap_dumper *wdh, erf_dump_t *dump_priv, guint64 timestamp, GPtrArray *sections, GArray *extra_ehdrs, int *err) { + union wtap_pseudo_header other_header; + struct erf_meta_tag gen_time_tag; + struct erf_meta_section *section_ptr; + guint total_wlen = 0; + guint total_rlen = 0; + gint64 alignbytes = 0; + guint i; + guint num_extra_ehdrs = 0; + + if(!sections || sections->len <= 0) + return FALSE; + + for(i = 0; i < sections->len; i++) { + section_ptr = (struct erf_meta_section*)g_ptr_array_index(sections, i); + total_wlen += section_ptr->section_length; + } + + gen_time_tag.type = ERF_META_TAG_gen_time; + gen_time_tag.length = 8U; + gen_time_tag.value = (guint8*)&dump_priv->gen_time; + total_wlen += gen_time_tag.length + 4; + + total_rlen = total_wlen + 24; /* 24 is the header + extension header length */ + if (extra_ehdrs) { + num_extra_ehdrs = MIN(extra_ehdrs->len, MAX_ERF_EHDR); + total_rlen += num_extra_ehdrs * 8; + } + /*padding to 8 byte alignment*/ + total_rlen += ERF_PADDING_TO_8(total_rlen); + + if(total_rlen > 65535) { + *err = WTAP_ERR_PACKET_TOO_LARGE; + return FALSE; + } + + other_header.erf.phdr.ts = timestamp; + other_header.erf.phdr.type = ERF_TYPE_META | 0x80; + other_header.erf.phdr.flags = 0x04; /* Varying record length */ + other_header.erf.phdr.lctr = 0; + other_header.erf.phdr.wlen = (guint16)total_wlen; + other_header.erf.phdr.rlen = (guint16)total_rlen; + /*Add our Host ID in Host ID extension header indicating we generated this + * record. Source ID 0 to avoid affecting implicit Host ID. */ + other_header.erf.ehdr_list[0].ehdr = erf_host_id_ext_hdr(dump_priv->host_id, 0); + /*Additional extension headers*/ + /*XXX: If we end up cutting the list short, erf_write_phdr will correct the + * unterminated extension header list*/ + if (num_extra_ehdrs > 0) { + other_header.erf.ehdr_list[0].ehdr |= ERF_EHDR_MORE_EXTHDR_MASK; + memcpy(&other_header.erf.ehdr_list[1], extra_ehdrs->data, sizeof(struct erf_ehdr) * num_extra_ehdrs); + } + + /* Make sure we always write out rlen, regardless of what happens */ + alignbytes = wdh->bytes_dumped + other_header.erf.phdr.rlen; + + if(!erf_write_phdr(wdh, WTAP_ENCAP_ERF, &other_header, err)) return FALSE; + + /* Generation time */ + erf_meta_write_tag(wdh, &gen_time_tag, err); + + /* Section(s) */ + for(i = 0; i < sections->len; i++) { + section_ptr = (struct erf_meta_section*)g_ptr_array_index(sections, i); + erf_meta_write_section(wdh, section_ptr, err); + } + + while(wdh->bytes_dumped < alignbytes){ + if(!wtap_dump_file_write(wdh, "", 1, err)) return FALSE; + wdh->bytes_dumped++; + } + + /* We wrote new packets, reloading is required */ + wdh->needs_reload = TRUE; + + return TRUE; + +} + +erf_dump_t *erf_dump_priv_create(void) { + erf_dump_t *dump_priv; + + dump_priv = (erf_dump_t*)g_malloc(sizeof(erf_dump_t)); + dump_priv->write_next_extra_meta = FALSE; + dump_priv->last_meta_periodic = FALSE; + dump_priv->gen_time = 0; + dump_priv->host_id = ERF_WS_DEFAULT_HOST_ID; + dump_priv->implicit_host_id = ERF_META_HOST_ID_IMPLICIT; + dump_priv->first_frame_time_sec = 0; + dump_priv->prev_inserted_time_sec = 0; + dump_priv->prev_frame_ts = 0; + dump_priv->prev_erf_type = 0; + dump_priv->user_comment_ptr = NULL; + dump_priv->periodic_sections = NULL; + dump_priv->periodic_extra_ehdrs = g_array_new(FALSE, FALSE, sizeof(struct erf_ehdr)); + dump_priv->rand = g_rand_new(); + + return dump_priv; +} + static gboolean erf_dump( wtap_dumper *wdh, const struct wtap_pkthdr *phdr, @@ -801,11 +1768,16 @@ static gboolean erf_dump( const union wtap_pseudo_header *pseudo_header = &phdr->pseudo_header; union wtap_pseudo_header other_phdr; int encap; + int erf_type; gint64 alignbytes = 0; - int i; + guint padbytes = 0; int round_down = 0; gboolean must_add_crc = FALSE; guint32 crc32 = 0x00000000; + erf_dump_t *dump_priv = (erf_dump_t*)wdh->priv; + /* Host ID extension header with Host ID 0 (unknown). For now use Source ID 1. */ + /* TODO: How to know if record was captured by this Wireshark? */ + guint64 non_erf_host_id_ehdr = erf_host_id_ext_hdr(0, 1); /* Don't write anything bigger than we're willing to read. */ if(phdr->caplen > WTAP_MAX_PACKET_SIZE_STANDARD) { @@ -819,80 +1791,184 @@ static gboolean erf_dump( encap = wdh->encap; } - if(encap == WTAP_ENCAP_ERF){ - /* We've been handed an ERF record, so there's not much to do here. */ - alignbytes = wdh->bytes_dumped + pseudo_header->erf.phdr.rlen; + if(!dump_priv->gen_time) { + erf_dump_priv_init_gen_time(dump_priv); + dump_priv->first_frame_time_sec = (guint32)phdr->ts.secs; + } + + if (encap != WTAP_ENCAP_ERF) { + unsigned int total_rlen;; + unsigned int total_wlen; - if(!erf_write_phdr(wdh, encap, pseudo_header, err)) return FALSE; + /*Non-ERF*/ - if(!wtap_dump_file_write(wdh, pd, phdr->caplen, err)) return FALSE; - wdh->bytes_dumped += phdr->caplen; + total_rlen = phdr->caplen+16; + total_wlen = phdr->len; - /*XXX: this pads the record to its original length, which is fine in most - * cases. However with >MAX_ERF_EHDR unnecessary padding will be added, and - * if the record was truncated this will be incorrectly treated as payload. - * More than 8 extension headers is unusual though, only the first 8 are - * written out anyway and fixing properly would require major refactor.*/ - while(wdh->bytes_dumped < alignbytes){ - if(!wtap_dump_file_write(wdh, "", 1, err)) return FALSE; - wdh->bytes_dumped++; + /* We can only convert packet records. */ + if (phdr->rec_type != REC_TYPE_PACKET) { + *err = WTAP_ERR_UNWRITABLE_REC_TYPE; + return FALSE; } - return TRUE; - } - /* We can only convert packet records. */ - if (phdr->rec_type != REC_TYPE_PACKET) { - *err = WTAP_ERR_UNWRITABLE_REC_TYPE; - return FALSE; - } + if ((erf_type = wtap_wtap_encap_to_erf_encap(encap)) == -1) { + *err = WTAP_ERR_UNWRITABLE_ENCAP; + return FALSE; + } - /*generate a fake header in other_phdr using data that we know*/ - /*covert time erf timestamp format*/ - other_phdr.erf.phdr.ts = ((guint64) phdr->ts.secs << 32) + (((guint64) phdr->ts.nsecs <<32) / 1000 / 1000 / 1000); - other_phdr.erf.phdr.type = wtap_wtap_encap_to_erf_encap(encap); - other_phdr.erf.phdr.flags = 0x4; /*vlen flag set because we're creating variable length records*/ - other_phdr.erf.phdr.lctr = 0; - /*now we work out rlen, accounting for all the different headers and missing fcs(eth)*/ - other_phdr.erf.phdr.rlen = phdr->caplen+16; - other_phdr.erf.phdr.wlen = phdr->len; - switch(other_phdr.erf.phdr.type){ - case ERF_TYPE_ETH: - other_phdr.erf.phdr.rlen += 2; /*2 bytes for erf eth_type*/ - if (pseudo_header->eth.fcs_len != 4) { - /* Either this packet doesn't include the FCS - (pseudo_header->eth.fcs_len = 0), or we don't - know whether it has an FCS (= -1). We have to - synthesize an FCS.*/ - if(!(phdr->caplen < phdr->len)){ /*don't add FCS if packet has been snapped off*/ + /* Generate a fake header in other_phdr using data that we know*/ + memset(&other_phdr, 0, sizeof(union wtap_pseudo_header)); + /* Convert time erf timestamp format*/ + other_phdr.erf.phdr.ts = ((guint64) phdr->ts.secs << 32) + (((guint64) phdr->ts.nsecs <<32) / 1000 / 1000 / 1000); + other_phdr.erf.phdr.type = (guint8)erf_type; + /* Support up to 4 interfaces */ + /* TODO: use multiple Source IDs and metadata records to support >4 interfaces */ + other_phdr.erf.phdr.flags = phdr->interface_id % ERF_MAX_INTERFACES; + other_phdr.erf.phdr.flags |= 0x4; /*vlen flag set because we're creating variable length records*/ + + other_phdr.erf.phdr.lctr = 0; + + /*now we work out rlen, accounting for all the different headers and missing fcs(eth)*/ + switch(other_phdr.erf.phdr.type & 0x7F){ + case ERF_TYPE_ETH: + total_rlen += 2; /*2 bytes for erf eth_type*/ + if (pseudo_header->eth.fcs_len != 4) { + /* Either this packet doesn't include the FCS + (pseudo_header->eth.fcs_len = 0), or we don't + know whether it has an FCS (= -1). We have to + synthesize an FCS.*/ + if(!(phdr->caplen < phdr->len)){ /*don't add FCS if packet has been snapped off*/ + crc32 = crc32_ccitt_seed(pd, phdr->caplen, 0xFFFFFFFF); + total_rlen += 4; /*4 bytes for added checksum*/ + total_wlen += 4; + must_add_crc = TRUE; + } + } + break; + case ERF_TYPE_HDLC_POS: + /*we assume that it's missing a FCS checksum, make one up*/ + if(!(phdr->caplen < phdr->len)){ /*unless of course, the packet has been snapped off*/ crc32 = crc32_ccitt_seed(pd, phdr->caplen, 0xFFFFFFFF); - other_phdr.erf.phdr.rlen += 4; /*4 bytes for added checksum*/ - other_phdr.erf.phdr.wlen += 4; - must_add_crc = TRUE; + total_rlen += 4; /*4 bytes for added checksum*/ + total_wlen += 4; + must_add_crc = TRUE; /* XXX - these never have an FCS? */ + } + break; + default: + break; + } + + /* Add Host ID extension header with Host ID 0 (unknown). For now use Source ID 1. */ + other_phdr.erf.phdr.type |= 0x80; + other_phdr.erf.ehdr_list[0].ehdr = non_erf_host_id_ehdr; + total_rlen += 8; + + padbytes = ERF_PADDING_TO_8(total_rlen); /*calculate how much padding will be required */ + if(phdr->caplen < phdr->len){ /*if packet has been snapped, we need to round down what we output*/ + round_down = (8 - padbytes) % 8; + total_rlen -= round_down; + }else{ + total_rlen += padbytes; + } + + if (total_rlen > G_MAXUINT16 || total_wlen > G_MAXUINT16) { + *err = WTAP_ERR_PACKET_TOO_LARGE; + return FALSE; + } + + other_phdr.erf.phdr.rlen = (guint16)total_rlen; + other_phdr.erf.phdr.wlen = (guint16)total_wlen; + + pseudo_header = &other_phdr; + } + + /* We now have a (real or fake) ERF record */ + erf_type = pseudo_header->erf.phdr.type & 0x7FU; + + /* Accumulate Host ID/Source ID to put in updated periodic metadata */ + /* TODO: pass these through from read interface list instead? */ + /* Note: this includes the one we made for the fake ERF header */ + erf_update_host_id_ext_hdrs_list(dump_priv, pseudo_header, dump_priv->periodic_extra_ehdrs); + + /* Insert new metadata record depending on whether the capture comment has + * changed. Write metadata each second at boundaries. If there is metadata + * write at the end of each of metadata records so we update the metadata. */ + if (erf_type == ERF_TYPE_META) { + /* Check whether the capture comment string has changed */ + /* Updates write_next_extra_meta */ + dump_priv->last_meta_periodic = erf_dump_priv_compare_capture_comment(wdh, dump_priv, pseudo_header, pd); + } else { /* don't want to insert a new metadata record while looking at another */ + if (dump_priv->prev_erf_type == ERF_TYPE_META && dump_priv->last_meta_periodic) { + /* Last frame was a periodic (non-comment) metadata record (and this frame is not), check if we + * need to insert one to update metdata. */ + + if(dump_priv->write_next_extra_meta) { + if (!dump_priv->periodic_sections) { + /* If we've seen metadata just insert the capture comment and not the + * rest of the metadata */ + dump_priv->periodic_sections = g_ptr_array_new_with_free_func(erf_meta_section_free); + erf_comment_to_sections(wdh, ERF_META_SECTION_CAPTURE, 0, dump_priv->user_comment_ptr, dump_priv->periodic_sections); } + + if (!erf_write_meta_record(wdh, dump_priv, dump_priv->prev_frame_ts, dump_priv->periodic_sections, dump_priv->periodic_extra_ehdrs, err)) return FALSE; + dump_priv->prev_inserted_time_sec = (guint32) phdr->ts.secs; + /*TODO: clear accumulated existing extension headers here?*/ } - break; - case ERF_TYPE_HDLC_POS: - /*we assume that it's missing a FCS checksum, make one up*/ - if(!(phdr->caplen < phdr->len)){ /*unless of course, the packet has been snapped off*/ - crc32 = crc32_ccitt_seed(pd, phdr->caplen, 0xFFFFFFFF); - other_phdr.erf.phdr.rlen += 4; /*4 bytes for added checksum*/ - other_phdr.erf.phdr.wlen += 4; - must_add_crc = TRUE; /* XXX - these never have an FCS? */ + + /* If we have seen a metadata record in the first ~1 second it + * means that we are dealing with an ERF file with metadata already in them. + * We dont want to write extra metadata if nothing has changed. We can't + * trust the Wireshark representation since we massage the fields on + * read. */ + /* restart searching for next meta record to update capture comment at */ + dump_priv->write_next_extra_meta = FALSE; + } else if (phdr->ts.secs > dump_priv->first_frame_time_sec + 1U + && dump_priv->prev_inserted_time_sec != phdr->ts.secs) { + /* For compatibility, don't insert metadata for older ERF files with no changed metadata */ + if (dump_priv->write_next_extra_meta) { + if (!dump_priv->periodic_sections) { + /* If we get here, metadata record was not found in the first ~1 sec + * but we have either a capture comment or a non-ERF file (see + * erf_dump_open) */ + /* Start inserting metadata records from wtap data at second boundaries */ + dump_priv->periodic_sections = g_ptr_array_new_with_free_func(erf_meta_section_free); + erf_wtap_info_to_sections(wdh, dump_priv->periodic_sections); + } } - break; - default: - break; + + /* At second boundaries insert either the updated comment (if we've seen some metadata records + * already) or the full metadata */ + if (dump_priv->periodic_sections) { + if (!erf_write_meta_record(wdh, dump_priv, (guint64)(phdr->ts.secs) << 32, dump_priv->periodic_sections, dump_priv->periodic_extra_ehdrs, err)) return FALSE; + dump_priv->prev_inserted_time_sec = (guint32) phdr->ts.secs; + } + } } - alignbytes = (8 - (other_phdr.erf.phdr.rlen % 8)) % 8; /*calculate how much padding will be required */ - if(phdr->caplen < phdr->len){ /*if packet has been snapped, we need to round down what we output*/ - round_down = (8 - (guint)alignbytes) % 8; - other_phdr.erf.phdr.rlen -= round_down; - }else{ - other_phdr.erf.phdr.rlen += (gint16)alignbytes; + /* If the packet user comment has changed, we need to + * construct a new header with additional Host ID and Anchor ID + * and insert a metadata record before that frame */ + /*XXX: The user may have changed the comment to cleared! */ + if(phdr->opt_comment || phdr->has_comment_changed) { + if (encap == WTAP_ENCAP_ERF) { + /* XXX: What about ERF-in-PCAPNG with existing comment (that wasn't + * modified)? */ + if(phdr->has_comment_changed) { + memcpy(&other_phdr, pseudo_header, sizeof(union wtap_pseudo_header)); + if(!erf_write_anchor_meta_update_phdr(wdh, dump_priv, phdr, &other_phdr, err)) return FALSE; + pseudo_header = &other_phdr; + } + } else { + /* Always write the comment if non-ERF */ + if(!erf_write_anchor_meta_update_phdr(wdh, dump_priv, phdr, &other_phdr, err)) return FALSE; + } } - if(!erf_write_phdr(wdh, WTAP_ENCAP_ERF, &other_phdr, err)) return FALSE; + /* Make sure we always write out rlen, regardless of what happens */ + alignbytes = wdh->bytes_dumped + pseudo_header->erf.phdr.rlen; + + if(!erf_write_phdr(wdh, WTAP_ENCAP_ERF, pseudo_header, err)) return FALSE; + if(!wtap_dump_file_write(wdh, pd, phdr->caplen - round_down, err)) return FALSE; wdh->bytes_dumped += phdr->caplen - round_down; @@ -901,14 +1977,21 @@ static gboolean erf_dump( if(!wtap_dump_file_write(wdh, &crc32, 4, err)) return FALSE; wdh->bytes_dumped += 4; } - /*records should be 8byte aligned, so we add padding*/ - if(round_down == 0){ - for(i = (gint16)alignbytes; i > 0; i--){ - if(!wtap_dump_file_write(wdh, "", 1, err)) return FALSE; - wdh->bytes_dumped++; - } + + /*XXX: In the case of ENCAP_ERF, this pads the record to its original length, which is fine in most + * cases. However with >MAX_ERF_EHDR unnecessary padding will be added, and + * if the record was truncated this will be incorrectly treated as payload. + * More than 8 extension headers is unusual though, only the first 8 are + * written out anyway and fixing properly would require major refactor.*/ + /*records should be 8byte aligned, so we add padding to our calculated rlen */ + while(wdh->bytes_dumped < alignbytes){ + if(!wtap_dump_file_write(wdh, "", 1, err)) return FALSE; + wdh->bytes_dumped++; } + dump_priv->prev_erf_type = pseudo_header->erf.phdr.type & 0x7FU; + dump_priv->prev_frame_ts = pseudo_header->erf.phdr.ts; + return TRUE; } @@ -926,7 +2009,31 @@ int erf_dump_can_write_encap(int encap) int erf_dump_open(wtap_dumper *wdh, int *err _U_) { + erf_dump_t *dump_priv; + gchar *s; + guint64 host_id; + + dump_priv = erf_dump_priv_create(); + wdh->subtype_write = erf_dump; + wdh->priv = dump_priv; + wdh->subtype_finish = erf_dump_finish; + + /* Get the capture comment string */ + get_user_comment_string(wdh, &dump_priv->user_comment_ptr); + /* XXX: If we have a capture comment or a non-ERF file assume we need to + * write metadata unless we see existing metadata in the first second. */ + if (dump_priv->user_comment_ptr || wdh->encap != WTAP_ENCAP_ERF) + dump_priv->write_next_extra_meta = TRUE; + + /* Read Host ID from environment variable */ + /* TODO: generate one from MAC address? */ + if ((s = getenv("ERF_HOST_ID")) != NULL) { + /* TODO: support both decimal and hex strings (base 0)? */ + if (ws_hexstrtou64(s, NULL, &host_id)) { + dump_priv->host_id = host_id & ERF_EHDR_HOST_ID_MASK; + } + } return TRUE; } @@ -1046,6 +2153,7 @@ static struct erf_if_mapping* erf_find_interface_mapping(erf_t *erf_priv, guint6 { struct erf_if_mapping if_map_lookup; + /* XXX: erf_priv should never be NULL here */ if (!erf_priv) return NULL; @@ -1085,6 +2193,100 @@ static void erf_set_interface_descr(wtap_block_t block, guint option_id, guint64 } } +static int erf_update_anchors_from_header(erf_t *erf_priv, struct wtap_pkthdr *phdr, union wtap_pseudo_header *pseudo_header, guint64 host_id, GPtrArray *anchor_mappings_to_update) +{ + guint8 type; + guint8 has_more; + guint64 hdr; + guint64 comment_gen_time = 0; + guint64 host_id_current; + guint64 anchor_id_current = 0; + int i = 0; + gchar *comment = NULL; + + if (!phdr || !pseudo_header) + return -1; + + /* Start with the first Host ID that was found on the record + * as the Anchor ID isn't required to be the first extension header' */ + host_id_current = host_id == ERF_META_HOST_ID_IMPLICIT ? erf_priv->implicit_host_id : host_id; + + has_more = pseudo_header->erf.phdr.type & 0x80; + + while (has_more && (i < MAX_ERF_EHDR)) { + hdr = pseudo_header->erf.ehdr_list[i].ehdr; + type = (guint8) (hdr >> 56); + + switch (type & 0x7f) { + case ERF_EXT_HDR_TYPE_HOST_ID: + host_id_current = hdr & ERF_EHDR_HOST_ID_MASK; + break; + + case ERF_EXT_HDR_TYPE_ANCHOR_ID: + { + anchor_id_current = hdr & ERF_EHDR_ANCHOR_ID_MASK; + if (!(ERF_ANCHOR_ID_IS_DEFINITION(hdr))) { + /* + * Anchor definiton flag is 0, attempt to associate a comment with this record + * XXX: currently the comment count may be wrong on the first pass! + */ + /* We may not have found the implicit Host ID yet, if so we are unlikely to find anything */ + struct erf_anchor_mapping* lookup_result; + lookup_result = erf_find_anchor_mapping(erf_priv, host_id_current, anchor_id_current); + if (lookup_result) { + if (lookup_result->gen_time > comment_gen_time) { + /* XXX: we might have a comment that clears the comment (i.e. + * empty string)! */ + if (lookup_result->comment && lookup_result->comment[0] != '\0') { + comment = lookup_result->comment; + } + comment_gen_time = lookup_result->gen_time; + } + } + } + else { + if (anchor_mappings_to_update && (pseudo_header->erf.phdr.type & 0x7f) == ERF_TYPE_META) { + /* + * Anchor definition flag is 1, put the mapping in an array + * which we will later update when we walk through + * the metadata tags + */ + /* Only Provenance record can contain the information we need */ + struct erf_anchor_mapping *mapping_ptr = + (struct erf_anchor_mapping*)g_malloc0(sizeof(struct erf_anchor_mapping)); + /* May be ERF_META_HOST_ID_IMPLICIT */ + mapping_ptr->host_id = host_id_current; + mapping_ptr->anchor_id = anchor_id_current; + g_ptr_array_add(anchor_mappings_to_update, mapping_ptr); + } + } + break; + } + } + + has_more = type & 0x80; + i += 1; + } + + if (comment) { + phdr->opt_comment = g_strdup(comment); + phdr->presence_flags |= WTAP_HAS_COMMENTS; + } else { + /* WTAP_HAS_COMMENT has no visible effect? + * Need to set opt_comment to NULL to prevent other packets + * from displaying the same comment + */ + /* XXX: We cannot free the old comment because it can be for a different + * frame and still in use, wiretap should be handling this better! */ + phdr->opt_comment = NULL; + } + + return 0; +} + +/** + * @brief Update the implicit Host ID and Anchor Mapping information + */ static int erf_update_implicit_host_id(erf_t *erf_priv, wtap *wth, guint64 implicit_host_id) { GHashTableIter iter; @@ -1095,6 +2297,8 @@ static int erf_update_implicit_host_id(erf_t *erf_priv, wtap *wth, guint64 impli struct erf_if_mapping* if_map = NULL; struct erf_if_mapping* if_map_other = NULL; struct erf_if_info* if_info = NULL; + struct erf_anchor_mapping* anchor_mapping = NULL; + struct erf_anchor_mapping* anchor_mapping_other = NULL; gchar *oldstr = NULL; char portstr_buf[16]; int i; @@ -1122,7 +2326,7 @@ static int erf_update_implicit_host_id(erf_t *erf_priv, wtap *wth, guint64 impli /* Pull mapping for update */ /* XXX: Can't add while iterating hash table so use list instead */ g_hash_table_iter_steal(&iter); - implicit_list = g_list_append(implicit_list, if_map); + implicit_list = g_list_prepend(implicit_list, if_map); } else { /* * XXX: We have duplicate interfaces in this case, but not much else we @@ -1192,6 +2396,51 @@ static int erf_update_implicit_host_id(erf_t *erf_priv, wtap *wth, guint64 impli } while ((item = g_list_next(item))); g_list_free(implicit_list); + implicit_list = NULL; + } + + /* + * We also need to update the anchor comment mappings + * to the correct Host ID. + */ + g_hash_table_iter_init(&iter, erf_priv->anchor_map); + + /* Remove the implicit mappings from the mapping table */ + while (g_hash_table_iter_next(&iter, &iter_value, NULL)) { + anchor_mapping = (struct erf_anchor_mapping*) iter_value; + + if (anchor_mapping->host_id == ERF_META_HOST_ID_IMPLICIT) { + /* Check we don't have an existing anchor that matches */ + anchor_mapping_other = erf_find_anchor_mapping(erf_priv, implicit_host_id, + anchor_mapping->anchor_id); + + if (anchor_mapping_other && anchor_mapping_other->gen_time >= anchor_mapping->gen_time) { + /* + * XXX: Duplicate entry of anchor mapping, keep the one with newer + * gen_time. + */ + g_hash_table_iter_remove(&iter); + } else { + /* Pull mapping for update */ + /* XXX: Can't add while iterating hash table so use list instead */ + g_hash_table_iter_steal(&iter); + implicit_list = g_list_prepend(implicit_list, anchor_mapping); + /* existing entry (if any) will be removed by g_hash_table_replace */ + } + } + } + + /* Re-add the non-clashing items under the real implicit Host ID */ + if (implicit_list) { + item = implicit_list; + do { + anchor_mapping = (struct erf_anchor_mapping*) item->data; + anchor_mapping->host_id = implicit_host_id; + g_hash_table_replace(erf_priv->anchor_map, anchor_mapping, anchor_mapping); + } while ((item = g_list_next(item))); + + g_list_free(implicit_list); + implicit_list = NULL; } return 0; @@ -1280,7 +2529,7 @@ static guint32 erf_meta_read_tag(struct erf_meta_tag* tag, guint8 *tag_ptr, guin /* length (2 bytes) */ taglength = pntoh16(&tag_ptr[2]); - tagtotallength = ERF_META_TAG_ALIGNED_LENGTH(taglength); + tagtotallength = ERF_META_TAG_TOTAL_ALIGNED_LENGTH(taglength); if (remaining_len < tagtotallength) { return 0; @@ -1318,20 +2567,30 @@ static int populate_capture_host_info(erf_t *erf_priv, wtap *wth, union wtap_pse switch (state->sectiontype) { case ERF_META_SECTION_CAPTURE: { - if (erf_priv->capture_metadata == TRUE) { + if (erf_priv->capture_gentime > state->gen_time) { return 0; } switch (tag.type) { case ERF_META_TAG_comment: - wtap_block_add_string_option(shb_hdr, OPT_COMMENT, tag.value, tag.length); + { + gchar *existing_comment = NULL; + /*XXX: hack to make changing capture comment work since Wireshark only + * displays one. For now just overwrite the comment as we won't + * pick up all of them yet due to the gen_time check above */ + if (wtap_block_get_nth_string_option_value(shb_hdr, OPT_COMMENT, 0, &existing_comment) == WTAP_OPTTYPE_SUCCESS) { + wtap_block_set_nth_string_option_value(shb_hdr, OPT_COMMENT, 0, tag.value, tag.length); + } else { + wtap_block_add_string_option(shb_hdr, OPT_COMMENT, tag.value, tag.length); + } break; + } } /* Fall through */ } case ERF_META_SECTION_HOST: { - if (erf_priv->host_metadata == TRUE) { + if (erf_priv->host_gentime > state->gen_time) { return 0; } @@ -1361,6 +2620,10 @@ static int populate_capture_host_info(erf_t *erf_priv, wtap *wth, union wtap_pse break; /* TODO: dag_version? */ /* TODO: could concatenate comment(s)? */ + case ERF_META_TAG_filter: + g_free(state->if_map->capture_filter_str); + state->if_map->capture_filter_str = g_strndup((gchar*) tag.value, tag.length); + break; default: break; } @@ -1424,9 +2687,9 @@ static int populate_capture_host_info(erf_t *erf_priv, wtap *wth, union wtap_pse g_free(cpu); if (state->sectiontype == ERF_META_SECTION_CAPTURE) { - erf_priv->capture_metadata = TRUE; + erf_priv->capture_gentime = state->gen_time; } else { - erf_priv->host_metadata = TRUE; + erf_priv->host_gentime = state->gen_time; } return 1; @@ -1441,34 +2704,34 @@ static int populate_module_info(erf_t *erf_priv _U_, wtap *wth, union wtap_pseud if (!wth || !state) return -1; - if (state->if_map->module_metadata == TRUE) { + if (state->if_map->module_gentime > state->gen_time) { return 0; } while ((tagtotallength = erf_meta_read_tag(&tag, state->tag_ptr, state->remaining_len)) && !ERF_META_IS_SECTION(tag.type)) { - switch (tag.type) { - case ERF_META_TAG_fcs_len: - if (tag.length >= 4) { - state->if_map->module_fcs_len = (gint8) pntoh32(tag.value); - } - break; - case ERF_META_TAG_snaplen: - /* XXX: this is generally per stream */ - if (tag.length >= 4) { - state->if_map->module_snaplen = pntoh32(tag.value); - } - break; - case ERF_META_TAG_filter: - g_free(state->if_map->module_filter_str); - state->if_map->module_filter_str = g_strndup((gchar*) tag.value, tag.length); - break; - } + switch (tag.type) { + case ERF_META_TAG_fcs_len: + if (tag.length >= 4) { + state->if_map->module_fcs_len = (gint8) pntoh32(tag.value); + } + break; + case ERF_META_TAG_snaplen: + /* XXX: this is generally per stream */ + if (tag.length >= 4) { + state->if_map->module_snaplen = pntoh32(tag.value); + } + break; + case ERF_META_TAG_filter: + g_free(state->if_map->module_filter_str); + state->if_map->module_filter_str = g_strndup((gchar*) tag.value, tag.length); + break; + } state->tag_ptr += tagtotallength; state->remaining_len -= tagtotallength; } - state->if_map->module_metadata = TRUE; + state->if_map->module_gentime = state->gen_time; return 1; } @@ -1493,7 +2756,7 @@ static int populate_interface_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo if_num = state->sectionid - 1; /* * Get or create the interface (there can be multiple interfaces in - * a MetaERF record). + * a Provenance record). */ if (if_num < 4) { /* Note: -1u > 4*/ if_info = &state->if_map->interfaces[if_num]; @@ -1552,7 +2815,10 @@ static int populate_interface_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo * We also don't support metadata for >4 interfaces per Host + Source * as we only use interface ID. */ - if (!int_data || state->if_map->interface_metadata & (1 << if_num)) + if (!int_data) + return 0; + + if (state->if_map->interface_gentime > state->gen_time && state->if_map->interface_metadata & (1 << if_num)) return 0; while ((tagtotallength = erf_meta_read_tag(&tag, state->tag_ptr, state->remaining_len)) && !ERF_META_IS_SECTION(tag.type)) { @@ -1632,14 +2898,20 @@ static int populate_interface_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo * XXX: Missing exposed existence/type-check. No way currently to check if * been set in the optionblock. */ - if (state->if_map->module_filter_str && !if_info->set_flags.filter) { - /* Duplicate because might use with multiple interfaces */ - if_filter.if_filter_str = state->if_map->module_filter_str; - wtap_block_add_custom_option(int_data, OPT_IDB_FILTER, &if_filter, sizeof if_filter); - /* - * Don't set flag because stream is more specific than module. Interface - * metadata bit is set so we don't look at the filter again regardless. - */ + if (!if_info->set_flags.filter) { + if (state->if_map->module_filter_str) { + /* Duplicate because might use with multiple interfaces */ + if_filter.if_filter_str = state->if_map->module_filter_str; + wtap_block_add_custom_option(int_data, OPT_IDB_FILTER, &if_filter, sizeof if_filter); + /* + * Don't set flag because stream is more specific than module. + */ + } else if (state->if_map->capture_filter_str) { + /* TODO: display separately? Note that we could have multiple captures + * from multiple hosts in the file */ + if_filter.if_filter_str = state->if_map->capture_filter_str; + wtap_block_add_custom_option(int_data, OPT_IDB_FILTER, &if_filter, sizeof if_filter); + } } if (state->if_map->module_fcs_len != -1 && !if_info->set_flags.fcs_len) { @@ -1796,8 +3068,71 @@ static int populate_stream_info(erf_t *erf_priv _U_, wtap *wth, union wtap_pseud return 1; } +static int populate_anchor_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_header *pseudo_header, struct erf_meta_read_state *state, GPtrArray *anchor_mappings_to_update) { + struct erf_meta_tag tag = {0, 0, NULL}; + guint32 tagtotallength; + gchar *comment_ptr = NULL; + guint i = 0; + + if (!wth || !state || !pseudo_header) + return -1; + + if (!anchor_mappings_to_update || anchor_mappings_to_update->len == 0) + return 0; + + while ((tagtotallength = erf_meta_read_tag(&tag, state->tag_ptr, state->remaining_len)) && !ERF_META_IS_SECTION(tag.type)) { + /* XXX:Always gets the first comment tag in the section */ + switch(tag.type) { + case ERF_META_TAG_comment: + if(!comment_ptr) { + comment_ptr = g_strndup((gchar*)tag.value, tag.length); + } + break; + default: + break; + } + + state->tag_ptr += tagtotallength; + state->remaining_len -= tagtotallength; + } + + if(comment_ptr) { + for(i = 0; i < anchor_mappings_to_update->len; i++) { + struct erf_anchor_mapping *mapping; + struct erf_anchor_mapping *lookup_result; + + mapping = (struct erf_anchor_mapping*)g_ptr_array_index(anchor_mappings_to_update, i); + lookup_result = (struct erf_anchor_mapping*)g_hash_table_lookup(erf_priv->anchor_map, mapping); + + /* Use the most recent comment, across all anchors associated with the + * record. */ + if(lookup_result) { + if(lookup_result->gen_time < state->gen_time) { + lookup_result->gen_time = state->gen_time; + g_free(lookup_result->comment); + lookup_result->comment = g_strdup(comment_ptr); + } + } + else { + /* !lookup_result */ + struct erf_anchor_mapping *new_mapping; + new_mapping = (struct erf_anchor_mapping *)g_malloc0(sizeof(struct erf_anchor_mapping)); + new_mapping->anchor_id = mapping->anchor_id; + new_mapping->host_id = mapping->host_id; + new_mapping->gen_time = state->gen_time; + new_mapping->comment = g_strdup(comment_ptr); + g_hash_table_replace(erf_priv->anchor_map, new_mapping, new_mapping); + } + } + } + + g_free(comment_ptr); + + return 1; +} + /* Populates the capture and interface information for display on the Capture File Properties */ -static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_header *pseudo_header, guint32 packet_size) +static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_header *pseudo_header, guint32 packet_size, GPtrArray *anchor_mappings_to_update) { struct erf_meta_read_state state; struct erf_meta_read_state *state_post = NULL; @@ -1829,14 +3164,6 @@ static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_h } - /* - * Skip the record if we already have enough metadata (seen one section for - * each type for the source). - */ - if ((state.if_map->interface_metadata & 0x03) - && erf_priv->host_metadata && erf_priv->capture_metadata) { - return 0; - } state.tag_ptr = wth->frame_buffer->data; state.remaining_len = packet_size; @@ -1844,10 +3171,35 @@ static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_h /* Read until see next section tag */ while ((tagtotallength = erf_meta_read_tag(&tag, state.tag_ptr, state.remaining_len))) { /* - * Skip until we get to the next section tag (which could be the current tag - * after an empty section or successful parsing). + * Obtain the gen_time from the non-section at the beginning of the record */ if (!ERF_META_IS_SECTION(tag.type)) { + if(state.gen_time == 0U + && tag.type == ERF_META_TAG_gen_time + ) { + memcpy(&state.gen_time, tag.value, sizeof(state.gen_time)); + + /* + * Since wireshark doesn't have a concept of different summary metadata + * over time, skip the record if metadata is older than what we already have. + */ + /* TODO: This doesn't work very well for some tags that wireshark only + * supports one copy of, we'll only end up with the first one. + * wtap_block_set_*_value() currently fails on WTAP_OPTTYPE_NOT_FOUND + * for everything except strings. + * Currently we only particularly care about updating the capture comment + * and a few counters anyway. + */ + if ((state.if_map->interface_metadata & 0x03) + && state.gen_time < erf_priv->host_gentime && state.gen_time < erf_priv->capture_gentime + && (!anchor_mappings_to_update || !anchor_mappings_to_update->len)) { + return 0; + } + } + /* + * Skip until we get to the next section tag (which could be the current tag + * after an empty section or successful parsing). + */ /* adjust offset */ state.tag_ptr += tagtotallength; state.remaining_len -= tagtotallength; @@ -1862,7 +3214,7 @@ static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_h /* Update parent section. Implicit grouping is by a change in section except Interface and Stream. */ if (tag.type != state.sectiontype) { if ((tag.type == ERF_META_SECTION_STREAM && state.sectiontype == ERF_META_SECTION_INTERFACE) || - (tag.type == ERF_META_SECTION_INTERFACE && state.sectiontype == ERF_META_SECTION_STREAM)) { + (tag.type == ERF_META_SECTION_INTERFACE && state.sectiontype == ERF_META_SECTION_STREAM)) { /* do nothing */ } else { state.parentsectiontype = state.sectiontype; @@ -1901,8 +3253,13 @@ static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_h /* * Skip sections that don't apply to the general set of records * (extension point for per-packet/event metadata). + * Unless we need to update the anchor info + * in which case, read into it */ if (state.sectionid & 0x8000) { + if(state.sectiontype & (ERF_META_SECTION_INFO)) { + populate_anchor_info(erf_priv, wth, pseudo_header, &state, anchor_mappings_to_update); + } continue; } @@ -1959,11 +3316,108 @@ static int populate_summary_info(erf_t *erf_priv, wtap *wth, union wtap_pseudo_h * Update known metadata so we only examine the first set of metadata. Need to * do this here so can have interface and stream in same record. */ - state.if_map->interface_metadata |= state.interface_metadata; + if (state.interface_metadata) { + state.if_map->interface_metadata |= state.interface_metadata; + state.if_map->interface_gentime = state.gen_time; + } return 0; } +static gboolean get_user_comment_string(wtap_dumper *wdh, gchar** user_comment_ptr) { + wtap_block_t wtap_block; + gboolean ret; + + wtap_block = NULL; + + if(wdh->shb_hdrs && (wdh->shb_hdrs->len > 0)) { + wtap_block = g_array_index(wdh->shb_hdrs, wtap_block_t, 0); + } + + if(wtap_block != NULL) { + ret = wtap_block_get_nth_string_option_value(wtap_block, OPT_COMMENT, 0, user_comment_ptr); + if(ret) { + return FALSE; + } + } + + return TRUE; +} + +static gboolean erf_dump_priv_compare_capture_comment(wtap_dumper *wdh _U_, erf_dump_t *dump_priv, const union wtap_pseudo_header *pseudo_header, const guint8 *pd){ + struct erf_meta_read_state state; + struct erf_meta_tag tag = {0, 0, NULL}; + guint32 tagtotallength; + gboolean found_capture_section = FALSE; + gboolean found_normal_section = FALSE; + gchar* comment_ptr = NULL; + + memset(&state, 0, sizeof(struct erf_meta_read_state)); + state.remaining_len = pseudo_header->erf.phdr.wlen; + memcpy(&(state.tag_ptr), &pd, sizeof(pd)); + + while((tagtotallength = erf_meta_read_tag(&tag, state.tag_ptr, state.remaining_len))) { + if (ERF_META_IS_SECTION(tag.type)) { + state.sectiontype = tag.type; + if (tag.length >= 4) { + state.sectionid = pntoh16(tag.value); + } else { + state.sectionid = 0; + } + + /* Skip sections that don't apply to the general set of records */ + if (!(state.sectionid & 0x8000)) { + found_normal_section = TRUE; + + if(tag.type == ERF_META_SECTION_CAPTURE) { + /* Found the Capture Section */ + found_capture_section = TRUE; + } + } + } else { + if (state.sectiontype == ERF_META_SECTION_CAPTURE && !(state.sectionid & 0x8000)) { + if (tag.type == ERF_META_TAG_comment) { + /* XXX: Only compare the first comment tag */ + if(!comment_ptr) { + comment_ptr = g_strndup((char*)tag.value, tag.length); + } + break; + } + } + } + + /* Read until we have the Capture section */ + state.tag_ptr += tagtotallength; + state.remaining_len -= tagtotallength; + } + + if(found_capture_section && (comment_ptr || dump_priv->user_comment_ptr)) { + if(g_strcmp0(comment_ptr, dump_priv->user_comment_ptr) + && !(dump_priv->user_comment_ptr == NULL && comment_ptr && comment_ptr[0] == '\0')) { + /* Also treat "" in ERF as equivalent to NULL as that is how we clear the comment on write. */ + + /* Comments are different, we should write extra metadata record at the end of the list */ + dump_priv->write_next_extra_meta = TRUE; + g_free(comment_ptr); + return TRUE; + } else { + /* We have a capture comment but there is no change, we don't + * need to insert the 'changed' comment. This most likely happened + * because we were looking at list of periodic records and got up to the + * one where the comment was last set. */ + dump_priv->write_next_extra_meta = FALSE; + } + /* Otherwise no effect on whether we need to write extra metadata record */ + } + /* We didn't find a capture section (e.g. looking at a comment Anchor + * record), or the comment hadn't changed. */ + + g_free(comment_ptr); + /* Return whether we found any non-local metadata (i.e. whether the record has + * metadata that is more than just packet 'comments') */ + return found_normal_section; +} + static void erf_close(wtap *wth) { erf_t* erf_priv = (erf_t*)wth->priv; |